FACTORS that influence prepaid churn subscribers when choosing a

mobile- service provider

E. Mankum

Student number: 10665392

A research project submitted to the Gordon Institute of Business Science,

University of Pretoria, in partial fulfilment of the requirements for the degree of
Master of Business Administration

9 November 2011

© University of Pretoria i
Abstract

High prepaid subscriber churn impacts on the service provider‟s profitability,

financial evaluation and resource utilisation. Simultaneously it deprives

subscribers of improved service because service providers are committing

resources to subscriber acquisition and the management of subscriber retention

instead of committing these resources to improving the value propositions

offered to subscribers.

A prepaid churn rate of just less than 50% in South Africa (2009) highlights the

need for research into the factors that influence churned prepaid subscribers‟

choice of mobile-service provider. This in turn will increase the knowledge of

retention and acquisition of prepaid subscribers.

Using a telephonic survey on a sample of 861 prepaid subscribers, quantitative

research was conducted to determine those factors that churned prepaid

subscribers consider important when choosing a mobile-service provider.

By making use of Analysis of Variance 4 out of 11 factors that, prepaid

subscribers considered very important were identified when choosing a mobile-

service provider. These included, in order of importance, Customer service

quality, Mobile tariffs, Prepaid airtime and Starter-pack availability and

Promotions.

Discriminant Analysis highlighted 9 factors to be used to predict and classify

groups of subscribers based on average monthly spend. The discriminate

equation coefficients highlighted how these two groups of subscribers view the

9 factors when choosing a mobile- service provider.

ii
Keywords

Churn: Churn is the discontinuation of the services of a mobile-service provider, and

calculated by dividing the annualised number of disconnections during the period by

the average monthly reported mobile subscriber base during the period. It is reported

as an annualised percentage figure.

Subscribers: Subscribers are the ultimate users i.e. customers, that subscribe to a

mobile-telecommunication service.

Prepaid: A product offering that requires the holder to purchase call credit/ service in

advance before the service/ call credit is used.

Mobile-service provider: A company that provides mobile-telecommunication

services typically using its own mobile network.

iii
Declaration

I declare that this research project is my own work. It is submitted in partial

fulfilment of the requirements for the degree of Master of Business

Administration at the Gordon Institute of Business Science, University of

Pretoria.

It has not been submitted before for any degree or examination in any other

University. I further declare that I have obtained the necessary authorisation

and consent to carry out this research.

E. Mankum

________________

9 November 2011

iv
Acknowledgements

I would like to thank my wife Yugeshree and my children Kirah and Senthan

who sacrificed a lot from their time and life for me to finish my studies. Without

their support, completion of my studies would have not been possible.

In addition, I would like to direct special thanks to my supervisor Professor

Louis-Francois Pau for his support and encouragement during my completion of

the research project.

v
Table Contents

1. Research formulation ..................................................................................1

1.1 Research problem .......................................................................................1
1.2 Research objective......................................................................................6
2. Literature review .........................................................................................8
2.1 Network quality ...........................................................................................8
2.2 Mobile tariffs................................................................................................9
2.3 Mobile tariff plan types ..............................................................................10
2.4 Distribution and availability ........................................................................11
2.5 Promotions ................................................................................................12
2.6 Loyalty programmes..................................................................................14
2.7 Customer service quality ...........................................................................15
2.8 Product bundles ........................................................................................17
2.9 Content and location-based services.........................................................19
2.10 Brand perception .......................................................................................19
3. Research propositions ..............................................................................22
4. Methodology .............................................................................................25
4.1 Statement of method .................................................................................25
4.1.1 Type of research study..............................................................................25
4.1.2 Degree of research question crystallisation ...............................................26
4.1.3 Method of data collection ..........................................................................26
4.1.4 Control of variables ...................................................................................27
4.1.5 The purpose of the study...........................................................................27
4.1.6 The time dimension ...................................................................................28
4.1.7 The topical scope ......................................................................................28
4.1.8 The research environment ........................................................................29
4.2 Unit of analysis..........................................................................................29
4.3 Population .................................................................................................29
4.4 Sampling ...................................................................................................31
4.4.1 Relevant population ..................................................................................32
4.4.2 Parameters of interest ...............................................................................32
4.4.3 Sample frame............................................................................................32
4.4.4 Sample type ..............................................................................................33
4.4.5 Sample size ..............................................................................................34

vi
 4.4.5.1 Stratified sampling.....................................................................................34
4.4.6 Communication method ............................................................................35
4.4.7 Data-collection method .............................................................................35
4.4.8 Measurement Instruments.........................................................................36
4.4.8.1 Data types.................................................................................................36
4.4.8.2 Error sources ............................................................................................36
4.4.8.3 Characteristics of sound measurement .....................................................37
4.4.8.4 Rating scales ............................................................................................37
4.4.9 Data analysis ............................................................................................37
5. Results ......................................................................................................39
5.1. Reliability analysis .....................................................................................39
5.1.1 Case-processing summary........................................................................39
5.1.2 Reliability statistics ....................................................................................39
5.1.3 Hotelling‟s T-squared test..........................................................................40
5.1.4 Intraclass correlation coefficient ................................................................41
5.2. Analysis of Variance between factors that are considered important by
prepaid mobile subscribers when choosing a mobile-service provider.......41
5.2.1 Test of homogeneity of variances..............................................................41
5.2.2 Analysis of Variances ................................................................................41
5.2.3 Robust tests of equality of means .............................................................42
5.2.4 Post-hoc test of means .............................................................................42
5.3. Descriptive statistics of sub factors relating to the 11 factors that are
considered by churned subscribers when choosing a mobile-service
provider .....................................................................................................47
5.3.1 Descriptive statistics of 3 factors that influence network quality .................47
5.3.1.1. Descriptive statistics of the importance of call quality ................................47
5.3.1.2. Descriptive statistics of the importance of drop calls .................................47
5.3.1.3. Descriptive statistics of the importance of geographical coverage .............48
5.3.2 Descriptive statistics of 3 factors that influence mobile tariffs ....................48
5.3.2.1 Descriptive statistics of the importance of price of calls .............................48
5.3.2.2 Descriptive statistics of the importance of the price of data .......................48
5.3.2.3 Descriptive statistics of the importance of price of an SMS .......................49
5.3.3 Descriptive statistics of 4 factors that influence mobile-tariff plan types .....49
5.3.3.1 Descriptive statistics of the importance of per-second tariffs .....................49
5.3.3.2 Descriptive statistics of the importance of a per-minute tariff plan .............49
5.3.3.3 Descriptive statistics of the importance of a peak and off peak tariff plan ..50

vii
5.3.3.4 Descriptive statistics of the importance of a flat tariff plan .........................50
5.3.4 Descriptive statistics of 6 factors of different prepaid product bundles .......50
5.3.4.1 Descriptive statistics of the importance of prepaid service with SIM only
offering ......................................................................................................50
5.3.4.2 Descriptive statistics of the importance of prepaid service with SIM and
phone offering ...........................................................................................51
5.3.4.3 Descriptive statistics of the importance of prepaid service with SIM and free
air-time offering .........................................................................................51
5.3.4.4 Descriptive statistics of the importance of prepaid service with SIM bundle
offering ......................................................................................................51
5.3.4.5 Descriptive statistics of the importance of prepaid service with data bundle
offering ......................................................................................................52
5.3.4.6 Descriptive statistics of the importance of prepaid service with music
download ..................................................................................................52
5.3.5 Descriptive statistics of 5 factors that influence access to mobile content
and location services.................................................................................53
5.3.5.1 Descriptive statistics of the importance of prepaid service with access to
mobile video ..............................................................................................53
5.3.5.2 Descriptive statistics of the importance of prepaid service with access to
mobile banking ..........................................................................................53
5.3.5.3 Descriptive statistics of the importance of prepaid service with access to
music download ........................................................................................53
5.3.5.4 Descriptive statistics of the importance of prepaid service with access to
ring tones ..................................................................................................54
5.3.5.5 Descriptive statistics of the importance of prepaid service with access to
instant messaging .....................................................................................54
5.3.6 Descriptive statistics of 4 factors for prepaid airtime and starter- pack
availability .................................................................................................55
5.3.6.1 Descriptive statistics of the importance of prepaid airtime and starter- pack
availability through food retail stores .........................................................55
5.3.6.2 Descriptive statistics of the importance of prepaid airtime and starter- pack
availability through clothing retail stores ....................................................55
5.3.6.3 Descriptive statistics of the importance of prepaid airtime and starter- pack
availability through local neighbourhood stores .........................................56
5.3.6.4 Descriptive statistics of the importance of prepaid airtime and starter-pack
availability through mobile-service provider stores ....................................56
5.4. Discriminant Analysis ................................................................................56
5.4.1 Discriminant Analysis descriptive statistics ................................................56
5.4.1.1. Discriminant Analysis descriptive case processing summary ....................57
5.4.1.2. Discriminant Analysis group statistics........................................................57

viii
 5.4.1.3. Test of equality of group means ................................................................59
5.4.1.4. Pooled within group matrices ....................................................................59
5.4.2 Discriminant Analysis ................................................................................60
5.4.2.1 Box‟s test of equality of covariance matrices .............................................60
5.4.2.2 Summary of canonical discriminant functions ............................................61
5.4.3 Discriminant Analysis classification statistics.............................................63
5.4.3.1 Classification processing summary ...........................................................63
5.4.3.2 Prior probabilities for groups .....................................................................64
5.4.3.3 Classification function coefficients .............................................................64
5.4.3.4 Classification results .................................................................................65
6. Data quality ...............................................................................................67
6.1 Reliability of scale .....................................................................................67
6.1.1 Cronbach's Alpha reliability analysis .........................................................67
6.1.2 All factors that are considered important by churned subscribers on the
scale do not have the same mean.............................................................68
6.1.3 Churned subscribers were consistent in their rating of factors that they
considered when choosing a mobile-service provider ...............................68
7. Results analysis ........................................................................................70
7.1 Factors that were considered by churned prepaid subscribers when
choosing a mobile-service provider ...........................................................70
7.1.1 Relationships between factors that were considered important by churned
subscribers when choosing a mobile-service provider...............................71
7.1.2 Test assumptions for Analysis of Variance ................................................72
7.1.3 Mean rating of factors were not equal for churned subscribers when
choosing a mobile-service provider ...........................................................74
7.1.4 Factors that were significantly different from each other when churned
subscribers changed a mobile-service provider.........................................75
7.1.4.1 Importance of customer service quality and when choosing a mobile-
service provider.........................................................................................75
7.1.4.2 Importance of mobile tariffs when choosing a mobile-service provider ......76
7.1.4.3 The importance of prepaid airtime and starter- pack availability when
choosing a mobile-service provider ...........................................................76
7.1.4.4 The importance of promotion when choosing a mobile-service provider....77
7.1.4.5 The importance of different airtime recharge options when choosing a
mobile-service provider .............................................................................77
7.1.4.6 The importance of loyalty programmes when choosing a mobile-service
provider .....................................................................................................77

ix
 7.1.4.7 Importance of mobile tariff plans when choosing a mobile-service provider
..................................................................................................................78
7.1.4.8 Importance of different types of prepaid product bundles when choosing a
mobile-service provider .............................................................................78
7.1.4.9 Network quality‟s importance when choosing a mobile-service provider....79
7.1.4.10 Importance of brand reputation and when choosing a mobile-service
provider .....................................................................................................79
7.1.4.11 Importance of access to mobile content and location services when
choosing a mobile-service provider ...........................................................80
7.2 Discriminant Analysis ................................................................................80
7.2.1 Discriminant model assumptions ...............................................................80
7.2.1.1 Predictors are not highly correlated ...........................................................80
7.2.1.2 The mean and variance of a given factor are not correlated ......................81
7.2.1.3 The group variance-co-variance matrices are equivalent ..........................82
7.2.1.4 Group membership is assumed to be mutually exclusive and collectively
exhaustive .................................................................................................83
7.2.1.5 The values of each predictor have a normal distribution ............................84
7.2.2 Discriminant model....................................................................................85
7.2.2.1 Canonical discriminant functions ...............................................................85
7.2.2.2 Wilks‟ lambda ............................................................................................86
7.2.2.3 Standardised canonical discriminant function coefficients .........................86
7.2.2.4 Structure matrix .........................................................................................87
7.2.2.5 Canonical discriminant function .................................................................88
7.2.3 Discriminant classification model...............................................................88
7.2.3.1 Classification processing summary ...........................................................88
7.2.3.2 Prior probabilities for groups .....................................................................89
7.2.3.3 Classification function coefficients .............................................................90
7.2.3.3.1 Less than R15,00 a month spent on prepaid service .................................90
7.2.3.3.2 More than R15,00 a month spent on prepaid service ................................91
7.2.3.4 Function coefficients between groups .......................................................92
7.2.3.5 Classification results .................................................................................94
8. Conclusion ................................................................................................98
8.1 Limitations of the study ...........................................................................106
8.2 Future research recommendations ..........................................................107
9. References..............................................................................................109
10. Appendixes .............................................................................................114
10.1 Churn reduction financial model ..............................................................114

x
10.2 Questionnaire..........................................................................................115

xi
List of Tables

Table 1.1: Differences and similarities between the water, electricity and mobile-
telecommunication industries in the South African regulated environment ..............1

Table 1.2: Direct acquisition cost of a prepaid mobile subscriber, August 2011 .......................3

Table 4.1: Characteristics of a mobile-service provider‟s prepaid offerings in South Africa,

June 2011 ........................................................................................................... 31

Table 4.2: Mobile-service providers per second mobile prepaid tariffs in South Africa, June
2011 .................................................................................................................... 31

Table 4.3: Sample size per segment of subscriber usage behaviour ..................................... 35

Table 5.1: Reliability analysis case-processing summary for 419 selected cases used in
Analysis of Variance and Discriminant Analysis ................................................... 39

Table 5.2: Reliability statistics for 419 selected cases for 419 selected cases used in Analysis
of Variance and Discriminant Analysis ................................................................. 39

Table 5.3: Mean scores of factors considered by churned prepaid mobile subscribers when
choosing a mobile-service provider, n 419 .......................................................... 39

Table 5.4: Means and variances item summary statistics of 11 factors that are considered
important by churned prepaid subscribers when choosing a mobile-service
provider, n 419 .................................................................................................... 40

Table 5.5: Scale statistics of 11 factors that are considered important by churned prepaid
subscribers when choosing a mobile-service provider, n 419 .............................. 40

Table 5.6: Hotelling‟s T-squared test of 11 factors that are considered important by churned
prepaid subscribers when choosing a mobile-service provider, n 419.................. 40

Table 5.7: Intraclass correlation coefficient two-way mixed model for consistency, n 419...... 41

Table 5.8: Test of homogeneity of variances between 11 factors that are considered important
by churned subscribers when choosing a mobile-service provider, n 419 ............. 41

Table 5.9: Analysis of Variance of 11 factors that are considered by churned subscribers
when choosing a mobile-service provider, n 419 .................................................. 42

Table 5.10: Robust tests of equality of means of 11 factors that are considered by churned
subscribers when choosing a mobile-service provider when equal variance is not
assumed, n 419................................................................................................... 42

Table 5.11: Tamhane‟sT2 multiple comparison post-hoc test of mean difference of 11 factors
that are considered by churned subscribers when choosing a mobile-service
provider, n 419 .................................................................................................... 42

Table 5.12: Descriptive statistics of the importance of call quality, n 419 ................................ 47

Table 5.13: Descriptive statistics of the importance of drop calls, n 419 .................................. 47

xii
Table 5.14: Descriptive statistics of the importance of geographical coverage, n 419.............. 48

Table 5.15: Descriptive statistics of the importance of price of voice calls, n 419 .................... 48

Table 5.16: Descriptive statistics of the importance of the price of data, n 419 ........................ 48

Table 5.17: Descriptive statistics of the importance of price of an SMS, n 419 ........................ 49

Table 5.18: Descriptive statistics of the importance of per-second tariffs, n 419 ...................... 49

Table 5.19: Descriptive statistics of the importance of a per-minute tariff plan, n 419 .............. 49

Table 5.20: Descriptive statistics of the importance of a peak and off peak tariff plan, n 419 ... 50

Table 5.21: Descriptive statistics of the importance of a flat-tariff plan, n 419 .......................... 50

Table 5.22: Descriptive statistics of the importance of prepaid service with SIM only offering, n
419 50

Table 5.23: Descriptive statistics of the importance of prepaid service with SIM and phone
offering, n 419 ..................................................................................................... 51

Table 5.24: Descriptive statistics of the importance of prepaid service with SIM and free air-
time offering, n 419.............................................................................................. 51

Table 5.25: Descriptive statistics of the importance of prepaid service with SIM bundle offering,
n 419 ................................................................................................................... 52

Table 5.26: Descriptive statistics of the importance of prepaid service with data bundle offering,
n 419 ................................................................................................................... 52

Table 5.27: Descriptive statistics of the importance of prepaid service with music download, n
419 52

Table 5.28: Descriptive statistics of the importance of prepaid service with access to mobile
video, n 419 ........................................................................................................ 53

Table 5.29: Descriptive statistics of the importance of prepaid service with access to mobile
banking, n 419..................................................................................................... 53

Table 5.30: Descriptive statistics of the importance of prepaid service with access to music
download, n 419 .................................................................................................. 54

Table 5.31: Descriptive statistics of the importance of prepaid service with access to ring tones,
n 419 ................................................................................................................... 54

Table 5.32: Descriptive statistics of the importance of prepaid service with access to instant
messaging, n 419 ................................................................................................ 54

Table 5.33: Descriptive statistics of the importance of prepaid airtime and starter- pack
availability through food retail stores, n 419 ........................................................ 55

Table 5.34: Descriptive statistics of the importance of Prepaid airtime and starter- pack
availability through clothing retail store, n 419 ..................................................... 55

xiii
Table 5.35: Descriptive statistics of the importance of prepaid airtime and starter- pack
availability through local neighbour stores, n 419 ................................................ 56

Table 5.36: Descriptive statistics of the importance of prepaid airtime and starter- pack
availability through mobile-service provider stores, n 419 .................................... 56

Table 5.37: Discriminant Analysis case processing summary, n 419 ...................................... 57

Table 5.38: Discriminant Analysis group statistics for factors that are considered by churned
subscribers when choosing a mobile-service provider, n 419 ............................... 57

Table 5.39: Tests of equality of group means for factors that are considered by churned
subscribers when choosing a mobile-service provider, n 419 ............................... 59

Table 5.40: Pooled within group matrices for factors that are considered by churned
subscribers when choosing a mobile-service provider, n 419 ............................... 59

Table 5.41: Box‟s test of equality of covariance matrices for nine factors and two groups that
are considered by churned subscribers when choosing a mobile-service provider, n
419 60

Table 5.42: Box‟s test of equality of covariance matrices for nine factors and two groups that
churned subscribers consider when choosing a mobile-service provider, n 419... 61

Table 5.43: Descriptive statistics for nine factors and two groups that are considered by
churned subscribers when choosing a mobile-service provider, n 419.................. 61

Table 5.44: Discriminant function Eigen values for nine factors and two groups that are
considered by churned subscribers when choosing a mobile-service provider, n
419 62

Table 5.45: Wilks‟ Lambda test for nine factors and two groups that are by considered by
churned subscribers when choosing a mobile-service provider, n 419.................. 62

Table 5.46: Standardised canonical discriminant function coefficients for nine factors and two
groups that considered by churned subscribers when choosing a mobile-service
provider, n 419 .................................................................................................... 62

Table 5.47: Structure matrix for nine factors and two groups that are considered by churned
subscribers when choosing a mobile-service provider, n 419 ............................... 62

Table 5.48: Canonical discriminant function coefficient for nine factors and two groups that are
considered by churned subscribers when choosing a mobile-service provider, n
419...................................................................................................................... 63

Table 5.49: Classification processing summary for nine factors and two groups that are
considered by churned subscribers when choosing a mobile-service provider, n
419 64

Table 5.50: Prior probabilities for groups for nine factors and two groups that are considered by
churned subscribers when choosing a mobile-service provider, n 419.................. 64

Table 5.51: Classification function coefficient for nine factors and two groups that are
considered by churned subscribers when choosing a mobile-service provider, n
419 64

xiv
Table 5.52: Classification group results for nine factors and two groups that are considered by
churned subscribers when choosing a mobile-service provider, n 419.................. 65

Table 10.1: Churn reduction financial model using Vodacom information for financial years
2009 and 2010 .................................................................................................. 114

List of Figures

Figure 8.1: Four level of factors that churned subscribers consider important when choosing a
mobile-service provider .................................................................................... 102

Figure 8.2: Acquisition and retention factors that prepaid subscribers consider when choosing
a mobile-service provider in decreasing order of importance ............................. 104

xv
1. Research formulation

1.1 Research problem

Unlike the water and electricity industries, which are mostly monopolistic in

nature, the mobile-telecommunication industry, although regulated, allows

subscribers the opportunity to choose and change a service provider at will.

While similar to the water and electricity industries, consumers usually have a

choice of contractual, post-paid or prepaid payment options for the services

provided.

The following table provides a summary of the differences and similarities

between the water, electricity and mobile-telecommunication industries in the

South African regulated environment:

Table 1.1: Differences and similarities between the water, electricity and mobile-
telecommunication industries in the South African regulated environment

Water Industry Electricity Mobile

Contractual, and prepaid Contractual, and prepaid Contractual, and prepaid

Product payment options
offerings offerings offerings

Monopolistic and Monopolistic and

Industry structure Oligopolistic
oligopolistic oligopolistic

Acquisition Costs Minimal Minimal High

Limited by price, but high

Product offerings Limited Limited variations in mobile tariffs
and mobile tariff plans

Source: (Eberhard, 2002), (Conradie, Goldin, Leiman, Standish, & Visser, 2001), (MTN Group Limited, 2011),
(Vodacom Group Limited, 2010)

In the mobile-telecommunication industry, post-paid subscribers have

contractual obligations, which impose financial penalties on subscribers if the

1
contracts are prematurely terminated (Lee, Murphy & Dickringer, 2006). These

penalties for early terminations of contracts allow service providers the

opportunity to recover the subscriber acquisition expenses incurred in acquiring

the subscriber.

Prepaid subscribers on the other hand are not bound by a contract, and can

therefore discontinue (churn) the services of the mobile-service provider almost

at their convenience and without notification (Radosavljevik, Putten & Larsen,

2010).

Prepaid subscribers‟ ability to churn at their convenience is a major source of

financial operational loss for service providers because service providers have

no means of recovering the direct subscriber acquisition costs incurred to

acquire a prepaid subscriber.

On the contrary, other regulated utilities, like electricity and water, have minimal

customer-acquisition costs, and they do not have service configuration costs, as

is the case in mobile-telecommunication industry.

Furthermore, as noted in Table 1.1, the electricity and water industries are

mostly monopolistic, and as such have a very high probability of recovering any

direct acquisition cost incurred from the increased lifetime of the prepaid

subscriber.

It is generally much more expensive to attract new customers than it is to retain

existing ones, and sell to them (Kotler & Keller, 2009). Jaworski,

Stathakopoulos and Krishnan (1993) also note that, owing to increased

purchases, referrals price premiums and reduced servicing costs, the

2
profitability of customers tend to increase over the lifetime of the retained

customer.

Livne, Simpson, and Talmor (2011) further argue that successful investment in

customer acquisition is capable of saving future operational expenses, and

hence improve the net profitability of a service provider. Furthermore, Livne,

Simpson and Talmor (2011) note that there is a positive statistical association

with increasing customer retention levels that will result in increases in the

future, net profits and market capitalisation of service providers.

Considering the above, it confirms the need for research to determine the

factors that influence the churned prepaid subscriber‟s choice of service

provider in a regulated mobile industry. Having established the factors that are

important to churned subscribers, focus on these factors can increase

subscriber retention.

Govender (2011), noted that on average the direct subscriber acquistion cost

per prepaid subscriber is in the region of R45.00. This is made up of the

following cost elements:

Table 1.2: Direct acquisition cost of a prepaid mobile subscriber, August 2011

Cost element Cost (including VAT)

Cost of prepaid starter pack, including the cost of SIM card R15

Dealer acquisition bonus paid to a dealer for signing up a prepaid subscriber R20

Supply -chain cost R5

Preload airtime R5

Total R45

Source: (Govender, 2011)

3
Given the direct acquisition cost of a prepaid subscriber as in Table 1.2, a

mobile-service provider is at risk of losing R45.00 per prepaid acquisition. The

provider has no means of recovering the above cost if the prepaid subscriber

decides not to use the mobile service once the service is acquired.

For the period ending 31 March 2010, Vodacom Group Limited (2010) reported

gross prepaid connections of 8.52 million, a churn rate of 43.7%, and an

average of R70,00 revenue per subscriber (ARPU). The net subscriber growth

for the period was minus 1.91 million which implies that Vodacom Group limited

churn 10.44 million subscribers for the reported financial year.

If Vodacom Group managed to reduce their churn percentage by 4% to 40%.

This means that they only churned 9.49 million subscribers instead of 10.44

million subscribers. They would have increased their gross revenue by 2.8%,

while increasing their EBITDA margin by 0.3%. The increase in their EBITDA

margin assumes that they did not need to add 0.947 million subscribers to their

network to compensate for the churned subscribers at a direct acquistion cost of

R45.00 per subscriber (see Table 10.1, Appendix, for calculations and

assumptions).

Given that the mobile-telecommunication industry in South Africa is oligopolistic

and regulated, with two dominating players, a weak third player and a new

fourth entrant, subscriber churn rates should arguably be low. However, the

Vodacom Group Limited (2010) reported a prepaid subscriber churn rate of

43.7% for the year ending 31 March 2010, while the MTN Group Limited (2010)

reported a prepaid subscriber churn rate of 43.5% for the year ending 31

December 2009.

4
The high prepaid churn rates of Vodacom Group Limited (2010) and MTN

Group Limited (2010) would indicate strong competition between the service

providers. However, owing to the oligopolistic nature of the market, and no

incentive to compete aggressively on mobile tariffs, it is argued that there are

other factors apart from price that cause subscribers to churn from their

respective service providers. These factors similarly drive subscribers to join the

service providers that meet subscribers‟ expectations of the factors that they

consider important.

Thus, with just under 50% churn rate for prepaid service in South Africa in 2009,

arguably that there are other factors or a combination of factors, including

mobile-telecommunication tariffs that influence the churned prepaid subscribers‟

choice of a service provider in the South Africa.

A high yearly churn rate thus warrants a need for research into factors that

influence the churned prepaid subscriber‟s choice of a service provider in the

South African regulated mobile-telecommunication industry to understand which

factors are important, and perceived to be significantly different by churned

prepaid subscribers when they change their mobile-telecommunication service

provider.

With reference to the growth potential, the World Cellular Information Services

(2011) estimated that there were 40 473 350 registered prepaid subscriber

identification module (SIM) users in South Africa with a potential growth of 24%

in the next five years.

5
However, there are only 29 148 000 unique prepaid subscribers. World Cellular

Information Services (2011) define these subscribers as the number of

individuals who own SIM cards. The difference of 39% between unique

subscribers and registered SIM subscribers draws attention to the high churn

rate and the multiple SIM usage behaviour in the market, among other things. In

accordance with the above, a subscriber might have several SIMs registered

with different networks, and alternate between different service providers.

Multiple factors could be the reason for this. For example, to take advantage of

differing call rates at different times of the day, as well as to exploit multiple

electronic identities.

In summary, the research problem is about the high prepaid subscriber churn in

an oligopolistic and regulated service industry that still has potential growth.

High prepaid subscriber churn impacts on the service provider‟s profitability,

financial evaluation and resource utilisation. Simultaneously it deprives

subscribers of improved service because service providers are committing

resources to subscriber acquisition and the management of subscriber retention

instead of committing these resources to improving the value propositions

offered to subscribers.

1.2 Research objective

The main objective of this research is to determine the factors that churned

prepaid mobile-telecommunication subscribers considered important when they

choosing their service provider in the context of the South African regulated

mobile-telecommunication industry.

6
In meeting the objective of the research, the benefits from the research are as

follows:

 Adding to the knowledge on prepaid subscriber retention in the high

mobile churn market of South Africa.

 Adding to the knowledge on prepaid subscriber acquisition in the high

mobile churn market of South Africa.

 Adding to the knowledge on optimal resource allocation of limited

business resources.

The above benefits are dependent on reaching the following research results:

 Determining the factors that are statistically significant when churned

prepaid subscribers choose a mobile-service operator.

 Determining the factors that churned prepaid subscribers consider being

equal when choosing a mobile-service provider, and similarly, the factors

that they do not consider to be equal.

 Determining a discriminate model to predict and classify which groups

churn prepaid subscribers will belong to by using the factors that they

consider important when choosing a mobile-service provider.

7
2. Literature review

2.1 Network quality

Research by Paulrajan and Rajkumar (2011) indicated that communication and

price are the key and influencing factors that motivate subscriber‟s preference

of service provider in the Indian mobile-telecommunication market.

Communication was defined as call quality, call drop rate and geographical

coverage, which in essence, amount to network quality.

Seo, Ranganathan and Babad (2008) further note that geographical coverage

and voice clarity (call quality) are the fundamental quality characteristics of a

mobile-telecommunication service which affect the subscriber‟s choice of a

mobile-service provider. In addition, research conducted by Birke and Swann

(2006) also found that network geographical coverage influences the

subscriber‟s choice of a mobile-service provider.

Research results by Turki (2010) further show that more than a third of the

participants in the UK mobile market claimed that poor mobile signal, which

usually results from poor network coverage, is the main cause of subscribers

switching to another mobile-service provider.

Rahman, Haque and Ahmad (2010) also noted that network quality was one of

the important factors of overall service quality of mobile-service providers in

Malaysia.

8
Similarly, Min and Wan (2009) noted from research undertaken in the Korean

mobile market that 13,7% of subscribers had churned their mobile-service

provider owing to network coverage problems in the form of weak signals.

2.2 Mobile tariffs

Sensitivity of demand refers to the extent to how volume-sensitive a product, or

a service, is to price changes. Thus, sensitivity represents a valuable strategic

tool in pricing (Tucker, 1966). Although the principle is relatively old, it holds true

for modern-day price and demand that a small change in price could result in

higher subscriber churn.

Thus, mobile tariffs are argued to be an influential factor for prepaid mobile

subscribers when choosing a mobile-service provider (Haque, Rahman &

Rahman, 2010). Haque, Rahman and Rahman (2010) also argue that mobile

tariffs are one of the factors that subscribers consider when deciding to churn

from a mobile-service provider.

However, Munnukka (2008) argues that the subscriber‟s price perception

relates to perception of quality, value and other beliefs. Thus, arguably, if a

service provider was rated as high in network quality, the subscriber might be

less sensitive to changes in mobile tariffs. The opposite would be true if the

network quality was poor. In other words, changes in mobile tariffs could result

in subscriber churn if network quality was poor.

With reference to mobile tariffs, a company that offers lower charges would be

able to attract more subscribers who would be willing to commit themselves to

the mobile telephone networks, and hence, a significant number of “call

9
minutes” might be achieved (Rahman, Haque & Ahmad, 2010). Thus, arguably

mobile tariffs do play a vital part in the choices that subscribers make in terms

of a service provider.

In addition, research conducted by Paulrajan and Rajkumar (2011) found that

mobile tariffs was one of the key and influencing factors that motivated

subscribers to prefer a specific service provider. This was further supported by

Rahman, Haque and Ahmad (2010) in their research findings that mobile tariff

was an important factor for subscribers when selecting a mobile-service

provider.

Min and Wan (2009) also identified mobile tariffs as one of the four main factors

that affect the switching behaviour of subscribers in the Korean mobile market.

The other factors included customer satisfaction, switching cost and customer

loyalty. From their research results 27,4% of subscribers churned their mobile-

service provider because of more competitive deals in the market, while 14,7%

of subscribers churned their mobile-service provider owing to high-priced calls

tariffs.

2.3 Mobile tariff plan types

Mobile tariffs and mobile tariff plan types are closely related. However, mobile

tariff plans reflect subscribers‟ preferences that minimise their billing rates,

given their expected usage (Raghuram, Jedidi, & Kohli, 2008). The empirical

results of Lambrecht and Skiera (2007) indicate that prepaid tariffs are driven by

the underestimation of the subscriber‟s usage, and indicate the subscriber‟s risk

of overspending.

10
Considering that the prepaid subscribers at the lower end of the market are

driven by cost, and tend to make short calls, changes in the billing unit can

make a significant difference in overall mobile expenditure (Barrantesa &

Galperin, 2008). In addition, Barrantesa and Galperin (2008) note that per-

second tariffs will increase usage among lower-end prepaid subscribers, and

will prevent them from churning.

Thus mobile tariff plans of mobile-service providers such as per-second tariffs,

flat tariffs or two-part tariffs would drive certain usage behaviour (Kollmann,

2000), and, depending on the market, will be perceived differently by

subscribers.

However, the risk of having many different prepaid mobile tariffs plans in an

oligopolistic or competitive mobile-telecommunication market could encourage

multiple mobile-service provider SIM usage behaviour (Kollmann, 2000). A

subscriber could use any SIM of the different service providers at a specific

point that provides the maximum benefit to the subscriber at the expense of

another service provider.

2.4 Distribution and availability

For most service organisations, consumer-marketing and industrial-marketing

firms, the distribution channel, or inter-organisational network of institutions,

comprising agents, wholesalers, distributors and retailers play a significant role

in the flow of goods from producers to consumers (Gorchels, Marien, & West,

2004).

11
In commoditised industry, such as the mobile-telecommunication industry,

subscribers will expect product availability through both formal and informal

channels, and it to be easily accessible (Heggde & Kumar, 2011). Thus, lack of

product availability will in turn result in subscribers churning as their needs and

wants of product are not met (Ireland, Hoskisson & Michael, 2011).

Lambart, Stock and Ellram (1998), argued that when consumers perceived all of

the top brands as substitutes for each other, this would lead to a lower brand

loyalty, which in turn would decrease the manufacturer‟s power. However, this

actually increases the distributor‟s power because sales are then determined by

what is in stock, and most often by what is recommended by the distributor, and

not by what a particular brand is offering.

Considering the high prepaid churn in South Africa, it would appear that the

subscribers are not significantly influenced by what the brand offers. By making

use of the argument of Lambart, Stock and Ellram (1998), this highlights the

power that distributors have in the South Africa prepaid market. This in turn

highlights the importance of product availability for subscribers‟ purchase

decision (Kotler & Keller, 2009).

2.5 Promotions

The objectives of any promotional strategy are to increase sales, maintain or

improve market share, create or improve brand recognition, create a favourable

climate for future sales, inform and educate the market, create a competitive

advantage that is relative to competitor‟s products or market position; and

improve promotional efficiency (Rowley, 1998). With these objectives, the

12
importance of promotions as a factor in the choice of a mobile-service provider

is noted.

Alvarez and Casielles (2005) further noted that promotion was a set of stimuli

that were offered sporadically, and these stimuli reinforced publicity actions to

promote the purchasing of a certain product. Moreover promotional offers

consist of several different objects for creating a better sales impact, for

example, coupons, samples, premiums, discounts, contests, point-of-purchase

displays and frequent-buyer programmes.

Every promotion technique was intended to have a direct impact on buying

behaviour and perception about the company or service providers (Alvarez &

Casielles, 2005). Alvarez and Casielles (2005) further argued that the objectives

of promotion would be reached largely when it was done sporadically; in other

words, when the consumer did not expect it. Furthermore, they note that

promotional actions had to be well planned, systematically organised, and

commonly integrated into the subject corporation‟s strategic marketing plan.

Rahman, Haque and Ahmad (2010), further noted that promotion had significant

impact on subscriber perception in selecting a mobile-telecommunication

service provider in Malaysia, since it was used to communicate with the

subscribers in terms of product offerings. Thus, promotion, and the perceived

value it creates for the subscriber are factors that could significantly influence

the prepaid subscriber‟s choice of a service provider.

13
Similarly, in the Bangladesh mobile-telecommunication market, Haque, Rahman

and Rahman (2010), found that statistically promotions had a positive impact on

the subscribers‟ perceptions when choosing a mobile-service provider.

2.6 Loyalty programmes

A loyalty programme is a supplier‟s structural effort to increase the customer‟s

attitudinal and behavioural commitment to the supplier‟s marketing offer

(Noordhoff, Pauwels & Odekerken, 2006). Yi and Jeon (2003) noted that in

times of serve competition, a loyalty programme is usually introduced to build

customer loyalty through a planned reward scheme that is based on a

customer‟s purchase history.

Furthermore, loyalty programmes enable firms to build stronger relationships,

enhance customer retention, encourage customers‟ recommendations, and

increase the number of products and services sold to their clients (Steers,

2007).

Leenheer, Bijmolt, Heerde and Smidts (2002), noted that loyalty has three types

of designs, namely, a saving component, price discounts and a multivendor

structure.

- A saving component is a saving programme, which gives saving points to

customers, dependent on the monetary amount spent by the customer at

the company. A programme member can redeem his or her points for a

reward, such as a free product, after reaching the minimal redeeming

threshold.

14
 - Price discounts are loyalty programmes that give price discounts on

promoted items.

- A multivendor structure is a loyalty programme that a company forms in

cooperation with companies from other industries with non-overlapping

product offerings.

Loyalty programmes are similar for mobile-telecommunication users with the

objective of rewarding subscribers for their usage (Kim, Park, & Jeong, 2004).

According to Kotler and Keller (2009), frequency programmes are designed to

provide rewards to customers who buy frequently. These programmes are also

evident in the mobile-telecommunication industry. However, the relevance and

the impact of influencing customer retention and acquisition have been

highlighted as possible factors that will influence a prepaid subscriber‟s choice

of a mobile-service operator.

2.7 Customer service quality

Customer service is a system of activities that comprises customer-support

systems, complaint-processing, speed of complaint processing, ease of

reporting a complaint and friendliness when reporting complaint (Kim, Park, &

Jeong, 2004). According to Soderlund and Rosengren (2008) the friendly

attitude and courteous behaviour of the service workers at service firms leave a

positive impression on the customer, which leads to customer satisfaction,

whilst the opposite behaviour will result in subscribers churning as of a result of

poor customer service.

15
According to Leisen and Vance (2001), customer service quality helps to create

the necessary competitive advantage by being an effective differentiating factor.

However, competitive advantage by firms is a value-creating strategy, which is

not simultaneously implemented by any existing or potential competitors (Barny,

1991). As a result, service quality can be used as a competitive advantage,

which is related to the customer‟s satisfaction, and leads to consumer loyalty

and future purchase (Johnson & Sirikit, 2002).

In particular, consumers prefer service quality when the price and other cost

elements are held in a constant state (Boyer & Hult, 2005). It has become a

distinct and important aspect of the product and service offering (Wal, Pampallis

& Bond, 2002). Moreover, according to them, a competitive advantage is also

sustained when other companies are unable to duplicate the benefits of this

strategy.

Service quality is essential and important for a telecommunication provider to

ensure the quality service for establishing and maintaining a loyal and profitable

customer (Zeithaml, 2000; Leisen & Vance, 2001). Rahman, Haque and Ahmad

(2010) noted that this might be a major problem for the telecommunication

service providers, especially for the mobile-telecommunication service

providers, to deliver quality service consistently as changes in market

compositions and competing characteristics have been surfacing incessantly.

Omotayo and Joachim (2008), through research results indicated a strong

relationship between customer service, satisfaction and retention in the mobile-

telecommunication industry in Nigeria.

16
Research conducted by Rahman, Haque and Ahmad (2010) on Malaysian

mobile-telecommunication subscribers also found that customer service quality

directly affected the subscriber‟s perceptions in selecting mobile-

telecommunication service providers. Furthermore, among all the significant

variables from the study, they also found that Malaysian telecommunication

subscribers perceived that price or call rate was the most important issue

followed by customer service quality, service availability and promotion.

In the Korean mobile-telecommunication market, Min and Wan (2009), noted

that 11,6% of subscribers churned their mobile-service provider because of

poor customer services.

2.8 Product bundles

Consumer relation to product form is dependent on consumer‟s personal

characteristics, surrounding products, utilities, experience, enjoyment of use

and the contribution to the fulfilment of the object‟s purpose (Horvath & Sajtos,

2002). Therefore, suppliers should increase the quality and quantity of utilitarian

reinforcements provided to both existing and potential new subscribers in order

to satisfy their needs, and encourage their retention behaviour (Ferguson &

Hlavinka, 2006).

Product bundles are a reflection of utilitarian reinforcements that could

encourage consumer retention. Thus, the research proposition is that by

offering different types of product benefits it would increase the utilitarian

reinforcements, and would influence the subscriber‟s choice of a mobile-service

provider.

17
Bundling is defined as the sale of two or more separate products or services in

a package at a discount (Stremersch & Tellis, 2002). A premise of bundling is

that consumers have asymmetric valuations of separate products. The larger

the dispersion and the higher the uncertainty in such valuations, the bigger the

gains are from the bundling of these products (Rautio, Anttila & Tuominen,

2007).

Among other things, benefits of bundling include the following:

- Bundling is likely to transfer surplus from consumers to firms, but it also

encourages products to be offered that might not be available under an a

la carte pricing system (Crawford & Cullen, 2007).

- They (Crawford & Cullen) further stated that bundling is an effective

entry-deterrent strategy, especially in an oligopolistic market.

- Bundling also changes competitive structure, and reduces the threat of

focused specialists (Nalebuff, 2004).

- Bundling reduces both average and fixed costs, as average costs are

reduced by the demand increase and by shared fixed costs (Rautio,

Anttila & Tuominen, 2007).

Stremersch and Tellis (2002), identify the following two categories of bundling:

1. “Price bundling”, which is the sale of two or more separate products in a

package at a discount without any integration of the products.

2. “Product bundling”, which is the integration and sale of two or more

separate products in a discounted package with benefit to some

consumers because of the benefit brought about by the integration.

18
Whereas price bundling is a pricing and promotional tool, product bundling is

regarded as strategic and of long-term value, and involves new product

development.

Chan-Olmsted and Guo (2011), note that in a competitive market, a mixed-

product bundling strategy, one that offers both bundled and individual services,

would dominate pure bundling strategy. In a sense, when necessary the mixed-

product bundling system offers a firm more opportunity for differentiation.

Furthermore, Chan-Olmsted and Guo (2011), argue that in order to differentiate

their products in a competitive marketplace, mobile-telecommunication service

providers will adopt a “mixed-product” bundling strategy.

2.9 Content and location-based services

Messaging services, mobile commerce, music and sound download, photo

download, mobile TV, mobile games and location-based services are becoming

important product additions to mobile-service providers (Kuo & Yen, 2009).

Value-added services have an impact on subscribers' usage patterns, and

become a significant differentiator across telecommunication service providers

(Kargin, Basoglu, & Daim, 2009). Thus, the research proposition is that content

and location-based services will be factors that influence the choice of a mobile-

service provider.

2.10 Brand perception

The concept of brand loyalty has strategic importance in terms of a firm‟s ability

to obtain sustainable competitive advantage and growth (Ersoy & Calık, 2010).

19
Ersoy and Calık, (2010) further noted that brand-loyal consumers are more

profitable, and that the costs of marketing to them are lower, than customers

who are not brand-loyal. In fact, it is suggested that the cost of recruiting a new

customer is five times greater than the cost of retaining an existing customer

(Wood, 2004) because of the following:

- Loyalty reduces customer acquisition costs.

- Positive word of mouth from loyal customers saves on marketing costs to

get new customers.

- Loyal customers‟ demand elasticity is lower, based on the degree or type

of loyalty.

- Brand-loyal customers increase the chances that a brand extension will

succeed and lower the risk of new product failure.

- Loyalty rates are connected to market share.

Arguably, consumers who develop a positive mental schema of a brand will

tend towards higher satisfaction and loyalty (Brodie, Whittome & Brush, 2009;

Hartmank & Spiro, 2005). Therefore, a positive corporate image appears to

encourage customer loyalty to the service provider.

In addition, research findings by Kim and Lee (2010) indicate that corporate

image and brand awareness are also strong antecedents for establishing

customer loyalty in mobile-communications service markets. These findings

support the results of the two previous studies (Andreassen & Lindestad B,

1988; Kwon & Lennon, 2009) that corporate image impacts on customer loyalty

with a varying degree of service expertise. In addition, corporate image and

20
brand awareness are also strong antecedents for establishing customer loyalty

in mobile-communications service markets (Kim, Park, & Jeong, 2004).

21
3. Research propositions

The following proposed factors influences churn prepaid mobile subscribers in

choosing their service providers in the South African regulated mobile-

telecommunication industry:

F1. Network quality significantly influences the choice of a mobile-service

provider

F11 Call quality significantly influences the choice of a mobile-service provider.

F12 Call drop rate significantly influences the choice of a mobile-service provider.

F13 Geographical coverage significantly influences the choice of a mobile-service

provider.

F2. Mobile tariffs significantly influences the choice of a mobile-service

provider

F21 The price of voice calls significantly influences the choice of a mobile-service

provider.

F22 The price of a megabit of data significantly influences the choice of a mobile-

service provider.

F23 The price of an SMS significantly influences the choice of a mobile-service

provider.

F3. Mobile tariffs plan types significantly influences the choice of a mobile

service-provider.

22
 F31 A per-second tariff plan significantly influences the choice of a mobile-service

provider.

F32 A per-minute tariff plan significantly influences the choice of a mobile-service

provider.

F33 A peak and off-peak tariff plan significantly influences the choice of a mobile-

service provider.

F34 A flat tariff plan significantly influences the choice of a mobile-service provider.

F4. Prepaid airtime and starter-pack availability significantly influences the

choice of a mobile-service provider.

F41 Access to prepaid airtime and starter- pack availability via food-retail channels

significantly influences the choice of a mobile-service provider.

F42 Access to prepaid airtime and starter- pack via clothing-retail channels significantly

influences the choice of a mobile-service provider.

F43 Access to prepaid airtime and starter- pack availability via informal retails

significantly influences the choice of a mobile-service provider.

F44 Access to prepaid airtime and starter- pack availability via mobile-operator specific

stores significantly influences the choice of a mobile-service provider.

F5. Different airtime recharge options significantly influences s the choice of a

mobile-service provider.

F6. Promotions significantly influence the choice of a mobile-service provider.

F7. Loyalty programmes significantly influences the choice of a mobile-service

provider.

23
F8. Customer service quality significantly influences the choice of a mobile-

service provider.

F9. Different types of prepaid product bundles significantly influence the

choice of a mobile-service provider.

F91 Prepaid service-offering SIM only, significantly influences the choice of a mobile-

service provider.

F92 Prepaid service offering SIM and phone, significantly influences the choice of a

mobile-service provider.

F93 Prepaid service offering SIM and free airtime significantly influences the choice of a

mobile-service provider.

F94 Prepaid service offering an SMS bundles significantly influences the choice of a

mobile-service provider.

F95 Prepaid service offering a data bundle significantly influences the choice of a

mobile-service provider.

F96 Prepaid service offering music downloads significantly influences the choice of a

mobile-service provider.

F10. Access to mobile content and location services significantly influences the

choice of a mobile-service provider.

F11. Brand reputation significantly influences the choice of a mobile-service

provider.

24
4. Methodology

4.1 Statement of method

4.1.1 Type of research study

Having considered the research problem, the research objectives and research

propositions, namely identify the factors that are significant in influencing

churned prepaid subscribers choosing a mobile-service provider, the researcher

decided on following a quantitative approach. Blumberg, Cooper & Schindler

(2008) describe quantitative studies as studies that rely on quantitative

information, in other words numbers and figures.

The research will determine those factors that churned prepaid subscribers

consider being important when choosing a mobile-service provider by making

use of Analysis of Variance and Discriminant Analysis.

Once these factors have been defined as important through Descriptive

statistics and Analysis of Variance analysis, Discriminant Analysis will be used

to predict and classify churn subscribers into groups, based on the average

monthly spend on prepaid service, by using the factors that they consider as

been important.

A qualitative research approach will not be suitable for achieving the objectives

of the research neither will it be in a position to test the research propositions.

Qualitative research studies base their accounts on qualitative information that

is words, non-quantified causalities, semantics, sentences and narratives

(Blumberg, Cooper & Schindler, 2008). The research propositions called for

25
testing the significance of the factors that were achieved by statistical analysis.

For this purpose, words, sentences and narratives will not suffice.

4.1.2 Degree of research question crystallisation

The study will be a formal in nature in that it begins with a descriptive account of

the current situation, namely the high rate of subscriber churn, and an

exploration of factors that are important in driving churned subscriber‟s

choosing a mobile-service provider followed by research propositions. The

former involves precise procedures and data-source specification.

The goal of the research will be to provide a valid representation of the current

state, namely the research problem that is the high prepaid subscriber churn in

an oligopolistic and regulated mobile service industry. In addition, the research

will test the proposed factors that influence churned subscribers choosing a

mobile-service provider which in turn will –

- improve subscriber acquisition;

- improve subscriber retention and thereby reduce subscriber acquisition

costs;

- allow service providers to focus their resources adequately to ensure

optimal utilisation of resources;

- Reduce subscriber churn, while increasing the lifetime of subscribers,

and thereby increasing the revenue-generating potential of the mobile

service.

4.1.3 Method of data collection

26
Blumberg, Cooper and Schindler (2008) highlight two methods of data

collection, namely, monitoring and interrogation or communication. Monitoring

includes studies in which the researcher inspects the activities of a subject or

the nature of certain material without attempting to elicit responses from

anyone. This type of method will not suffice for the purposes and objective of

the researcher since actual information about respondents gathered from them

are required.

The interrogation or communication method allows the researcher to question

the subjects, and to collect their responses by personal or impersonal means.

This method will achieve the objectives and purpose of the research, and is the

method that is recommended for data collection.

4.1.4 Control of variables

The research will follow an ex-post facto design, in that there will be no control

over the variables, and for that reason, no manipulation can take place.

The research will only report on what has happened, or what is happening, and

in doing so factors will be held constant by judicious selection of subjects in

accordance with strict sampling procedures and statistical manipulation of

findings. In addition, the research will be limited to churned subscribers who

have been actively using the mobile network within a six-month period, and who

have been active in the last month of the six-month period.

4.1.5 The purpose of the study

27
Of the various types of studies that are available for the purpose, namely

descriptive, causal and/or predictive, the descriptive study will be the most

appropriate for the research objectives and propositions of this study.

According to Blumberg, Cooper and Schindler (2008) descriptive studies serve

the following research objectives:

- Description of phenomena or characteristics that are associated with a

subject population (such as the significant factors that influence the choice

of a mobile-service provider).

- Estimates of the proportion of a population that have similar characteristics.

- Discovering associations among different independent variables and a

dependent variable.

4.1.6 The time dimension

The proposed research will be a cross-sectional study, in the sense that the

study will be carried out only once, and will represent a snapshot of a single

point in time. This is unlike longitudinal studies, which are repeated over an

extended period.

However, it is anticipated that this research will form the reference for

longitudinal studies through which the relevance of factors can be evaluated to

establish whether the factors that churned subscribers regard as determining

their choice of a mobile-service provider change over time.

4.1.7 The topical scope

28
The proposed research will be a statistical study, which, according to Blumberg,

Cooper and Schindler (2008), is designed for breadth.

In essence, the study will attempt to capture a population‟s characteristics

(characteristics of churned prepaid subscribers in South Africa) by making

inferences from a sample‟s characteristic.

Hypotheses, or research propositions, are to be tested quantitatively, and

generalisation about the findings will be presented. This will be subject to

constraining hypotheses that are based on the representativeness of the

sample and the validity of the research design.

4.1.8 The research environment

Given the nature of the study, the research environment will be similar to that of

field conditions, in that the research will be conducted under actual

environmental conditions. Respondents will be interrogated in their usual or

current environment, such as in their homes or workplaces.

4.2 Unit of analysis

In accordance with the research topic and research proposition, the unit of

analysis will be churned prepaid mobile subscribers who have been active

users of the mobile network within a six-month period, and who have been

active in the last month of the six-month period. This unit of analysis will be the

level at which the research is performed and the objects are researched, these

being churned prepaid mobile subscribers.

4.3 Population

29
According to Blumberg, Cooper and Schindler (2008), the population is the total

collection of elements from which some inferences are made. In the proposed

research, the population will be prepaid registered subscribers who have been

active on the 8ta mobile network during a six-month period ending 31

December 2010. On 31 December 2010 there were 296 059 of these

subscribers on the 8ta mobile network.

It is argued that although there are four mobile network service providers in

South Africa, with 8ta entering the market only in 2010, the prepaid subscribers

who use 8ta are no different from those of the other three mobile operators

because of the following:

- Vodacom Group Limited (2010) reported a prepaid subscriber churn rate

of 45.4% between 1 April 2008 and 31 March 2009 while MTN Group

Limited (2010) reported a prepaid subscriber churn rate of 43.5%

between 1 January 2009 and 31 December 2009. These relatively similar

churn rates would imply that the factors that cause prepaid subscribers to

change and leave a mobile-service provider are not unique to an

individual service provider.

- The review of the prepaid offering in June 2011, in the South African

market, in Table 4.1, indicates that there is no significant difference

between the products offered by the different mobile-service providers.

- Mobile tariffs of the four mobile-service providers (see Table 4.2) are of a

similar nature.

30
Table 4.1: Characteristics of a mobile-service provider’s prepaid offerings in South Africa, June
2011

MTN VODACOM Cell C 8ta

Per-second and per-minute products Yes Yes Yes Yes

Peak and off-peak tariff rates Yes Yes Yes Yes

Flat-rate tariffs Yes Yes Yes Yes

SMS, MMS, content services, location services

Yes Yes Yes Yes
internet browsing
Loyalty campaigns Yes Yes Yes Yes

Bonus-recharge campaigns Yes Yes Yes Yes

Source: (Cell C, 2011), (MTN, 2011), (Vodacom, 2011), (8ta, 2011)

Table 4.2: Mobile-service providers per second mobile prepaid tariffs in South Africa, June 2011

Cell C Vodacom MTN MTN 8ta

Easy-chat prepaid 4U prepaid Call per sec Call per sec peak Per second

*Tariff rates during peak hours (07:00 and 20:00)

Own net R 2.85 R 2.85 R 2.89 R 2.39 R 2.75

Other net R 2.85 R 2.99 R 2.89 R 2.39 R 2.75

Fixed R 2.85 R 2.85 R 2.89 R 2.39 R 0.65

MMS R 0.90 R 0.80 R 0.50

SMS R 0.80 R 0.80 R 0.75 R 0.75 R 0.50

*Tariffs rates during non-peak hours (20:00 and 07:00)

Own net R 1.30 R 1.12 R 1.19 R 1.59 R 1.12

Other net R 1.30 R 1.30 R 1.19 R 1.59 R 1.12

Fixed R 1.30 R 1.12 R 1.19 R 1.59 R 0.65

MMS R 0.90 R 0.80 R 0.50

SMS R 0.34 R 0.80 R 0.35 R 0.35 R 0.50

Source: (Cell C, 2011), (MTN, 2011), (Vodacom, 2011), (8ta, 2011)

4.4 Sampling

The strength of a sample design lies in how well it represents the characteristics

of the population it purports to represent (Blumberg, Cooper & Schindler, 2008).

In the proposed research, the sample is expected to be representative of the

31
296 059 prepaid registered subscribers that have been active within a six-month

period ending 31 December 2010 on the 8ta mobile network and who have

churned in the last 12 months from a mobile-service provider.

4.4.1 Relevant population

As noted above, the relevant population will be all prepaid registered

subscribers on the 8ta mobile network who have been active within a six-month

period ending 31 December 2010. On 31 December 2010 there were 296 059

prepaid registered subscribers to the above network.

4.4.2 Parameters of interest

Population parameters are summarised statistic descriptors, which in the

proposed research are factors that are significant in subscribers choosing a

mobile-service provider. Examples of these will be incidence proportions,

means and variance.

Sample statistics that are gathered in essences will be descriptive statistics,

Analysis of Variance statistics and statistics generated from Discriminant

Analysis.

4.4.3 Sample frame

The sample frame is very similar to the population. However, it differs at the

theoretical level because a complete and correct list of the population is rarely

found (Blumberg, Cooper & Schindler, 2008). In other words, there will be a

difference between the population and the sample frame for the proposed

research.

32
For the purpose of the proposed research the sample frame will be the 296 059

prepaid registered subscribers on the 8ta mobile network who have been active

within a six-month period ending 31 December 2010. The sample frame and

population will be the same, as there is a complete and correct list of the

population.

However, considering that not all registered subscribers remain active users

over a prolonged period, the 296 059 prepaid registered subscribers referred to

above have been further reduced to 106 122 who were active on the 8ta

network for the month of December 2010. Thus the sample frame for the study

is the above list of 106 122 subscribers. From these 106 122 subscribers a

sample of 861 were chosen who completed the questionnaire that will be used

for the research.

Furthermore, of the sample of 861 subscribers above, only subscribers who

churned from their previous service provider in the last 12 months will be used

in the analysis, as churned subscribers are the focus sample of the research.

4.4.4 Sample type

The research will follow a probability sample, in that probability-based

confidence estimates of various parameters to the population will be made of

prepaid subscribers. In this study, as noted in section 4.4.3, the population will

be the 106 122 subscribers as explained above.

In addition, all 106 122 of the population elements make up the sample frame.

The sample frame is a list of elements that will have an equal chance of being

selected to be included in the sample (Blumberg, Cooper & Schindler, 2008).

33
Furthermore, considering that an actual sampling frame exists, sampling will be

done with replacement. In selecting the sample elements for the research, the

following predetermined rules will be used:

- A three-process callback will be followed, after which the subscriber will

be regarded as not active, and will be removed from the sample frame.

- Once a subscriber has been removed from the sample frame, a new

subscriber will be randomly selected as a replacement.

- A subscriber non-participation in the research is considered as no-

response, and will be replaced with a new randomly selected subscriber.

- If most of his or her answers are irrelevant, or if a respondent refuses to

answer, this respondent will be replaced.

4.4.5 Sample size

With an error margin of 5%, a 95% confidence level, a population size of 106

122 subscribers and a response distribution probability of 50%, the

recommended sample size of the proposed research will be 385 units

(http://www.raosoft.com, 2011). However, considering that the research focus is

on subscribers‟ preference-based decisions, which generally do not follow

normal distributions, the sample size will be increased to 861 completed and

meaningful answers. Furthermore, it is anticipated that of the 861 selected

subscribers at least half would have churned in the last 12 months.

4.4.5.1 Stratified sampling

Having considered, that there could be different subscriber usage behaviour in

the sample frame of 106 122, the frame was broken down into segments.

34
Table 4.3 lists the two segments of subscribers in accordance with their

average spend amount and sample size. However to ensure that those

subscribers who spend less are well represented in the sample, the sample is

disproportionate to subscribers who spend less as they are most likely to

demonstrate the characteristic of subscribers that churn.

Table 4.3: Sample size per segment of subscriber usage behaviour

Count Sample size

Count of subscribers who spend less than R15.00 per month

on their prepaid service 44 738 760

Count of subscribers who spend more than R15.00 per month

on their prepaid service 61 384 101
Total 106 122 861
Source: (8ta, 2011)

4.4.6 Communication method

Considering that the research will be conducted by means of a survey, the

options available for conducting the research are personal interviews,

telephonic interviews, mail, computer or a combination of all four (Blumberg,

Cooper & Schindler, 2008). Having considered the advantages and

disadvantages, the telephonic interview process was chosen to be the proposed

survey method.

Further preference for the choice of the telephonic survey is that it will allow the

sampling method as described in sections 4.4.2, 4.4.3 and 4.4.4 to be utilised.

4.4.7 Data-collection method

35
The data will be collected by means of a predefined questionnaire. Since the

survey will be conducted via the telephone, the results will be captured

immediately on the computer terminal.

The data will be coded into electronic format for statistical analysis.

4.4.8 Measurement Instruments

Properties are characteristics of the objects (Blumberg, Cooper & Schindler,

2008). In the proposed research, the social properties of the individuals will be

measured.

The research will focus strongly on the social properties of individuals and on

subscribers‟ perception of factors that are significant when choosing a service

provider. In addition, the research will explain a virtual preference for the service

provider, and not necessarily be followed by an action. The proposed research

will also measure the subscribers‟ decisions and decision confidence in

choosing a mobile-service provider.

4.4.8.1 Data types

The research will utilise all four types of data, namely nominal, ordinal, interval

and ratio.

4.4.8.2 Error sources

Blumberg, Cooper and Schindler (2008) identify the following four major error

sources that may contaminate the research results:

- Participant

36
 - Situational factors

- Measurer

- Data-collection instruments

4.4.8.3 Characteristics of sound measurement

There are three major criteria for evaluating a measurement tool, namely,

validity, reliability and practically (Blumberg, Cooper & Schindler, 2008).

Regarding validity, Blumberg, Cooper and Schindler (2008) suggest defining

what constitutes relevant evidence in terms of the nature of the research

problem. The researcher‟s judgement is to organise the answer in accordance

with measure-relevant types.

Reliability draws reference to consistency; a measure is reliable to the degree

that it supplies consistent results.

Stability has three measurement coefficients, namely stability, equivalence and

internal consistency

In evaluating the measurement tool for the research, validity, reliability, and

practically will be measured by using the appropriate statistical as necessitated.

4.4.8.4 Rating scales

See questionnaire in appendix for rating scales used in the question.

4.4.9 Data analysis

37
In addition to descriptive statistics, Analysis of Variance and Discriminant

Analysis will be used together with the associated analysis. These methods are

referenced further under the research results section.

38
5. Results

All statistical tests and subsequent results were obtained by using the IBM

SPSS Statistics 19 version with the researched data as the source data.

5.1. Reliability analysis

5.1.1 Case-processing summary

Table 5.1: Reliability analysis case-processing summary for 419 selected cases used in

Analysis of Variance and Discriminant Analysis

N %

Cases Valid 419 100.0

Excluded 0 .0

Total 419 100.0

5.1.2 Reliability statistics

Table 5.2: Reliability statistics for 419 selected cases for 419 selected cases used in Analysis of

Variance and Discriminant Analysis

Cronbach's Alpha Cronbach's Alpha based on standardised Items No of Items

.711 .713 11

Table 5.3: Mean scores of factors considered by churned prepaid mobile subscribers when

choosing a mobile-service provider, n 419

Mean Std deviation N

Network quality 7.25 2.452 419

Mobile tariffs 9.54 .936 419

39
Mobile tariffs plans 7.89 1.889 419

Prepaid airtime and starter- pack availability 9.29 1.711 419

Different airtime recharge options 8.16 2.185 419

Promotions 8.79 1.822 419

Loyalty programmes 8.05 2.125 419

Customer service quality 9.62 .826 419

Different types of prepaid product bundles 7.54 2.613 419

Access to mobile content and location services 6.61 2.131 419

Brand reputation 6.96 1.993 419

Table 5.4: Means and variances item summary statistics of 11 factors that are considered

important by churned prepaid subscribers when choosing a mobile-service provider,

n 419

Maximum No of
Mean Minimum Maximum Range Variance
/Minimum Items

Item
8.155 6.609 9.616 3.007 1.455 1.083 11
means

Item
3.819 0.682 6.828 6.145 10.009 3.54 11
variances

Table 5.5: Scale statistics of 11 factors that are considered important by churned prepaid

subscribers when choosing a mobile-service provider, n 419

Mean Variance Std deviation No of Items

89.71 101.562 10.078 11

5.1.3 Hotelling‟s T-squared test

Table 5.6: Hotelling’s T-squared test of 11 factors that are considered important by churned

prepaid subscribers when choosing a mobile-service provider, n 419

40
 Hotelling's T-Squared F df1 df2 Sig

1395.050 136.501 10 409 .000

5.1.4 Intraclass correlation coefficient

Table 5.7: Intraclass correlation coefficient two-way mixed model for consistency, n 419

F-test with true value F-test with true value

95% confidence interval
Intraclass 0 0

correlation Lower
Upper bound Value df1 df2 Sig
bound

Single
b
.142 0.117 0.171 2.817 418 4180 0
measures

Average
c
.711 0.612 0.723 2.817 418 4180 0
measures

Two-way mixed effects model where people effects are random and measures effects are fixed.

a. Type C intraclass correlation coefficients using a consistency definition –the between-measure variance is excluded
from the denominator variance.
b. The estimator is the same, whether the interaction effect is present or not.
c. This estimate is computed assuming the interaction effect is absent, because it is not estimable otherwise.

5.2. Analysis of Variance between factors that are considered important by

prepaid mobile subscribers when choosing a mobile-service provider

5.2.1 Test of homogeneity of variances

Table 5.8: Test of homogeneity of variances between 11 factors that are considered important

by churned subscribers when choosing a mobile-service provider, n 419

Levene statistic df1 df2 Sig.

69.142 10 4598 .000

5.2.2 Analysis of Variances

41
Table 5.9: Analysis of Variance of 11 factors that are considered by churned subscribers when

choosing a mobile-service provider, n 419

Sum of squares df Mean square F Sig.

Between groups 4537.461 10 453.746 118.807 .000

Within groups 17560.621 4598 3.819

Total 22098.081 4608

5.2.3 Robust tests of equality of means

Table 5.10: Robust tests of equality of means of 11 factors that are considered by churned

subscribers when choosing a mobile-service provider when equal variance is not

assumed, n 419

a
Statistic df1 df2 Sig.

Welch 194.280 10 1821.902 .000

Brown-Forsythe 118.807 10 3766.851 .000

a. Asymptotically F distributed

5.2.4 Post-hoc test of means

Table 5.11: Tamhane’sT2 multiple comparison post-hoc test of mean difference of 11 factors

that are considered by churned subscribers when choosing a mobile-service

provider, n 419

Mean 95% confidence interval

difference (I-

(I) Category (J) Category J) Std. Error Sig. Lower bound Upper bound

*
Network quality Mobile tariffs -2.296 .128 .000 -2.72 -1.87

*
Mobile tariffs plans -.642 .151 .001 -1.14 -.14

*
Prepaid airtime and -2.038 .146 .000 -2.52 -1.55

starter- pack availability

42
 *
Different airtime recharge -.916 .160 .000 -1.45 -.38

options

*
Promotions -1.539 .149 .000 -2.04 -1.04

*
Loyalty programmes -.804 .158 .000 -1.33 -.28

*
Customer service quality -2.368 .126 .000 -2.79 -1.95

Different types of prepaid -.296 .175 .995 -.88 .29

product bundles

*
Access to mobile content .640 .159 .003 .11 1.17

and location services

Brand reputation .284 .154 .977 -.23 .80

*
Mobile tariffs Network quality 2.296 .128 .000 1.87 2.72
*
Mobile tariffs plans 1.654 .103 .000 1.31 2.00

Prepaid airtime and .258 .095 .320 -.06 .57

starter- pack availability

*
Different airtime recharge 1.379 .116 .000 .99 1.77

options
*
Promotions .757 .100 .000 .42 1.09
*
Loyalty programmes 1.492 .113 .000 1.11 1.87

Customer service quality -.072 .061 1.000 -.27 .13

*
Different types of prepaid 2.000 .136 .000 1.55 2.45

product bundles
*
Access to mobile content 2.936 .114 .000 2.56 3.31

and location services

*
Brand reputation 2.580 .108 .000 2.22 2.94
*
Mobile tariffs plans Network quality .642 .151 .001 .14 1.14
*
Mobile tariffs -1.654 .103 .000 -2.00 -1.31
*
Prepaid airtime and -1.396 .124 .000 -1.81 -.98

starter- pack availability

Different airtime recharge -.274 .141 .947 -.74 .19

options
*
Promotions -.897 .128 .000 -1.32 -.47

Loyalty programmes -.162 .139 1.000 -.62 .30

*
Customer service quality -1.726 .101 .000 -2.06 -1.39

Different types of prepaid .346 .158 .794 -.18 .87

product bundles

43
 *
Access to mobile content 1.282 .139 .000 .82 1.74

and location services

*
Brand reputation .926 .134 .000 .48 1.37
*
Prepaid airtime and Network quality 2.038 .146 .000 1.55 2.52

starter- pack availability Mobile tariffs -.258 .095 .320 -.57 .06
*
Mobile tariffs plans 1.396 .124 .000 .98 1.81
*
Different airtime recharge 1.122 .136 .000 .67 1.57

options
*
Promotions .499 .122 .003 .09 .90
*
Loyalty programmes 1.234 .133 .000 .79 1.68
*
Customer service quality -.329 .093 .023 -.64 -.02
*
Different types of prepaid 1.742 .153 .000 1.24 2.25

product bundles
*
Access to mobile content 2.678 .134 .000 2.23 3.12

and location services

*
Brand reputation 2.322 .128 .000 1.90 2.75
*
Different airtime recharge Network quality .916 .160 .000 .38 1.45

options Mobile tariffs -1.379

*
.116 .000 -1.77 -.99

Mobile tariffs plans .274 .141 .947 -.19 .74

*
Prepaid airtime and -1.122 .136 .000 -1.57 -.67

starter- pack availability

*
Promotions -.623 .139 .000 -1.08 -.16

Loyalty programmes .112 .149 1.000 -.38 .61

*
Customer service quality -1.451 .114 .000 -1.83 -1.07
*
Different types of prepaid .621 .166 .011 .07 1.17

product bundles
*
Access to mobile content 1.556 .149 .000 1.06 2.05

and location services

*
Brand reputation 1.200 .144 .000 .72 1.68
*
Promotions Network quality 1.539 .149 .000 1.04 2.04
*
Mobile tariffs -.757 .100 .000 -1.09 -.42
*
Mobile tariffs plans .897 .128 .000 .47 1.32
*
Prepaid airtime and -.499 .122 .003 -.90 -.09

starter- pack availability

*
Different airtime recharge .623 .139 .000 .16 1.08

options

44
 *
Loyalty programmes .735 .137 .000 .28 1.19
*
Customer service quality -.828 .098 .000 -1.15 -.50
*
Different types of prepaid 1.243 .156 .000 .73 1.76

product bundles
*
Access to mobile content 2.179 .137 .000 1.72 2.63

and location services

*
Brand reputation 1.823 .132 .000 1.39 2.26
*
Loyalty programmes Network quality .804 .158 .000 .28 1.33
*
Mobile tariffs -1.492 .113 .000 -1.87 -1.11

Mobile tariffs plans .162 .139 1.000 -.30 .62

*
Prepaid airtime and -1.234 .133 .000 -1.68 -.79

starter- pack availability

Different airtime recharge -.112 .149 1.000 -.61 .38

options
*
Promotions -.735 .137 .000 -1.19 -.28
*
Customer service quality -1.563 .111 .000 -1.93 -1.19

Different types of prepaid .508 .165 .108 -.04 1.06

product bundles
*
Access to mobile content 1.444 .147 .000 .96 1.93

and location services

*
Brand reputation 1.088 .142 .000 .62 1.56
*
Customer service quality Network quality 2.368 .126 .000 1.95 2.79

Mobile tariffs .072 .061 1.000 -.13 .27

*
Mobile tariffs plans 1.726 .101 .000 1.39 2.06
*
Prepaid airtime and .329 .093 .023 .02 .64

starter- pack availability

*
Different airtime recharge 1.451 .114 .000 1.07 1.83

options
*
Promotions .828 .098 .000 .50 1.15
*
Loyalty programmes 1.563 .111 .000 1.19 1.93
*
Different types of prepaid 2.072 .134 .000 1.63 2.52

product bundles
*
Access to mobile content 3.007 .112 .000 2.64 3.38

and location services

*
Brand reputation 2.652 .105 .000 2.30 3.00

Different types of prepaid Network quality .296 .175 .995 -.29 .88

45
 *
product bundles Mobile tariffs -2.000 .136 .000 -2.45 -1.55

Mobile tariffs plans -.346 .158 .794 -.87 .18

*
Prepaid airtime and -1.742 .153 .000 -2.25 -1.24

starter- pack availability

*
Different airtime recharge -.621 .166 .011 -1.17 -.07

options
*
Promotions -1.243 .156 .000 -1.76 -.73

Loyalty programmes -.508 .165 .108 -1.06 .04

*
Customer service quality -2.072 .134 .000 -2.52 -1.63
*
Access to mobile content .936 .165 .000 .39 1.48

and location services

*
Brand reputation .580 .161 .018 .05 1.11
*
Access to mobile content Network quality -.640 .159 .003 -1.17 -.11
*
and location services Mobile tariffs -2.936 .114 .000 -3.31 -2.56
*
Mobile tariffs plans -1.282 .139 .000 -1.74 -.82
*
Prepaid airtime and -2.678 .134 .000 -3.12 -2.23

starter- pack availability

*
Different airtime recharge -1.556 .149 .000 -2.05 -1.06

options
*
Promotions -2.179 .137 .000 -2.63 -1.72
*
Loyalty programmes -1.444 .147 .000 -1.93 -.96
*
Customer service quality -3.007 .112 .000 -3.38 -2.64
*
Different types of prepaid -.936 .165 .000 -1.48 -.39

product bundles

Brand reputation -.356 .143 .507 -.83 .12

Brand reputation Network quality -.284 .154 .977 -.80 .23

*
Mobile tariffs -2.580 .108 .000 -2.94 -2.22

*
Mobile tariffs plans -.926 .134 .000 -1.37 -.48

*
Prepaid airtime and -2.322 .128 .000 -2.75 -1.90

starter- pack availability

*
Different airtime recharge -1.200 .144 .000 -1.68 -.72

options

*
Promotions -1.823 .132 .000 -2.26 -1.39

*
Loyalty programmes -1.088 .142 .000 -1.56 -.62

*
Customer service quality -2.652 .105 .000 -3.00 -2.30

46
 *
Different types of prepaid -.580 .161 .018 -1.11 -.05

product bundles

Access to mobile content .356 .143 .507 -.12 .83

and location services

*. The mean difference is significant at the 0.05 level.

5.3. Descriptive statistics of sub factors relating to the 11 factors that are

considered by churned subscribers when choosing a mobile-service

provider

5.3.1 Descriptive statistics of 3 factors that influence network quality

5.3.1.1. Descriptive statistics of the importance of call quality

Table 5.12: Descriptive statistics of the importance of call quality, n 419

N Valid 394

No response 25

Mean 1.38

Mode 1

Std deviation .656

5.3.1.2. Descriptive statistics of the importance of drop calls

Table 5.13: Descriptive statistics of the importance of drop calls, n 419

N Valid 389

No response 30

Mean 2.49

Mode 3

Std deviation .664

47
5.3.1.3. Descriptive statistics of the importance of geographical coverage

Table 5.14: Descriptive statistics of the importance of geographical coverage, n 419

N Valid 389

No response 30

Mean 2.11

Mode 2

Std deviation .735

5.3.2 Descriptive statistics of 3 factors that influence mobile tariffs

5.3.2.1 Descriptive statistics of the importance of price of calls

Table 5.15: Descriptive statistics of the importance of price of voice calls, n 419

N Valid 419

No response 0

Mean 1.54

Mode 1

Std deviation .739

5.3.2.2 Descriptive statistics of the importance of the price of data

Table 5.16: Descriptive statistics of the importance of the price of data, n 419

N Valid 416

No response 3

Mean 2.37

Mode 3

Std deviation .735

48
5.3.2.3 Descriptive statistics of the importance of price of an SMS

Table 5.17: Descriptive statistics of the importance of price of an SMS, n 419

N Valid 419

No response 0

Mean 2.10

Mode 2

Std deviation .750

5.3.3 Descriptive statistics of 4 factors that influence mobile-tariff plan types

5.3.3.1 Descriptive statistics of the importance of per-second tariffs

Table 5.18: Descriptive statistics of the importance of per-second tariffs, n 419

N Valid 327

No response 92

Mean 1.76

Mode 1

Std deviation .814

5.3.3.2 Descriptive statistics of the importance of a per-minute tariff plan

Table 5.19: Descriptive statistics of the importance of a per-minute tariff plan, n 419

N Valid 307

No response 112

Mean 2.02

Mode 3

Std deviation .826

49
5.3.3.3 Descriptive statistics of the importance of a peak and off peak tariff plan

Table 5.20: Descriptive statistics of the importance of a peak and off peak tariff plan, n 419

N Valid 334

No response 85

Mean 2.06

Mode 2

Std deviation .775

5.3.3.4 Descriptive statistics of the importance of a flat tariff plan

Table 5.21: Descriptive statistics of the importance of a flat-tariff plan, n 419

N Valid 197

No response 222

Mean 2.27

Mode 3

Std deviation .773

5.3.4 Descriptive statistics of 6 factors of different prepaid product bundles

5.3.4.1 Descriptive statistics of the importance of prepaid service with SIM only

offering

Table 5.22: Descriptive statistics of the importance of prepaid service with SIM only offering, n

419

N Valid 92

No response 327

Mean 2.12

Mode 3

50
Std deviation .823

5.3.4.2 Descriptive statistics of the importance of prepaid service with SIM and

phone offering

Table 5.23: Descriptive statistics of the importance of prepaid service with SIM and phone

offering, n 419

N Valid 215

No response 204

Mean 2.00

Mode 2

Std deviation .800

5.3.4.3 Descriptive statistics of the importance of prepaid service with SIM and

free air-time offering

Table 5.24: Descriptive statistics of the importance of prepaid service with SIM and free air-

time offering, n 419

N Valid 336

No response 83

Mean 1.68

Mode 1

Std deviation .771

5.3.4.4 Descriptive statistics of the importance of prepaid service with SIM

bundle offering

51
Table 5.25: Descriptive statistics of the importance of prepaid service with SIM bundle offering,

n 419

N Valid 118

No response 301

Mean 2.31

Mode 3

Std deviation .790

5.3.4.5 Descriptive statistics of the importance of prepaid service with data

bundle offering

Table 5.26: Descriptive statistics of the importance of prepaid service with data bundle offering,

n 419

N Valid 211

No response 208

Mean 2.10

Mode 3

Std deviation .821

5.3.4.6 Descriptive statistics of the importance of prepaid service with music

download

Table 5.27: Descriptive statistics of the importance of prepaid service with music download, n

419

N Valid 187

No response 232

Mean 2.19

Mode 2

52
Std deviation .744

5.3.5 Descriptive statistics of 5 factors that influence access to mobile content

and location services

5.3.5.1 Descriptive statistics of the importance of prepaid service with access to

mobile video

Table 5.28: Descriptive statistics of the importance of prepaid service with access to mobile

video, n 419

N Valid 207

No response 212

Mean 2.43

Mode 3

Std deviation .685

5.3.5.2 Descriptive statistics of the importance of prepaid service with access to

mobile banking

Table 5.29: Descriptive statistics of the importance of prepaid service with access to mobile

banking, n 419

N Valid 307

No response 112

Mean 1.63

Mode 1

Std deviation .771

5.3.5.3 Descriptive statistics of the importance of prepaid service with access to

53
 music download

Table 5.30: Descriptive statistics of the importance of prepaid service with access to music

download, n 419

N Valid 266

No response 153

Mean 2.06

Mode 2

Std deviation .758

5.3.5.4 Descriptive statistics of the importance of prepaid service with access to

ring tones

Table 5.31: Descriptive statistics of the importance of prepaid service with access to ring tones,

n 419

N Valid 120

No response 299

Mean 2.39

Mode 3

Std deviation .714

5.3.5.5 Descriptive statistics of the importance of prepaid service with access to

instant messaging

Table 5.32: Descriptive statistics of the importance of prepaid service with access to instant

messaging, n 419

N Valid 259

No response 160

54
Mean 1.85

Mode 1

Std deviation .822

5.3.6 Descriptive statistics of 4 factors for prepaid airtime and starter- pack

availability

5.3.6.1 Descriptive statistics of the importance of prepaid airtime and starter-

pack availability through food retail stores

Table 5.33: Descriptive statistics of the importance of prepaid airtime and starter- pack

availability through food retail stores, n 419

N Valid 325

No response 94

Mean 2.19

Mode 2

Std deviation .728

5.3.6.2 Descriptive statistics of the importance of prepaid airtime and starter-

pack availability through clothing retail stores

Table 5.34: Descriptive statistics of the importance of prepaid airtime and starter- pack

availability through clothing retail store, n 419

N Valid 294

No response 125

Mean 2.28

Mode 2

Std deviation .698

55
5.3.6.3 Descriptive statistics of the importance of prepaid airtime and starter-

pack availability through local neighbourhood stores

Table 5.35: Descriptive statistics of the importance of prepaid airtime and starter- pack

availability through local neighbour stores, n 419

N Valid 323

No response 96

Mean 1.61

Mode 1

Std deviation .794

5.3.6.4 Descriptive statistics of the importance of prepaid airtime and starter-

pack availability through mobile-service provider stores

Table 5.36: Descriptive statistics of the importance of prepaid airtime and starter- pack

availability through mobile-service provider stores, n 419

N Valid 236

No response 183

Mean 1.91

Mode 1

Std deviation .878

5.4. Discriminant Analysis

5.4.1 Discriminant Analysis descriptive statistics

Discriminant Analysis was done with the following selection inputs:

56
 - Grouping variable: Two grouping categories were used, namely

subscribers that spend less than R15,00 a month on prepaid service, and

subscribers that spend more than R15,00 a month on prepaid service.

- Selection variable: Subscribers who churned during the last 12 months

from date of interview. Subscribers who did not churn during the last 12

months from date of interview were not selected.

- Independent variables: Nine independent variables of 11 variables

selected.

- Analysis was conducted with all variables entered together as opposed

to the stepwise method.

5.4.1.1. Discriminant Analysis descriptive case processing summary

Table 5.37: Discriminant Analysis case processing summary, n 419

Unweighted cases N Percent

Valid 419 48.7

Missing or out-of-range group codes 0 .0

At least one missing discriminating variable 0 .0

Excluded Both missing or out-of-range group codes and at least one 0 .0

missing discriminating
Unselected a variable 442 51.3

Total 442 51.3

Total 861 100.0

a: Unselected are subscribers that did not churn in the last 12 months

5.4.1.2. Discriminant Analysis group statistics

Table 5.38: Discriminant Analysis group statistics for factors that are considered by churned

subscribers when choosing a mobile-service provider, n 419

57
 Valid N (listwise)

Unweighte
Groups Mean Std deviation Weighted
d

Usage less than

Network quality 6.9 2.544 335 335
R15,00 a month

Mobile tariffs 9.65 0.743 335 335

Mobile tariffs plans 7.76 1.989 335 335

Prepaid airtime and starter- pack

9.51 1.597 335 335
availability

Promotions 9.02 1.712 335 335

Loyalty programmes 7.87 2.298 335 335

Customer service quality 9.66 0.811 335 335

Access to mobile content and

6.32 2.195 335 335
location services

Brand reputation 6.7 2.055 335 335

Usage more than

Network quality 8.65 1.313 84 84
R15,00 a month

Mobile tariffs 9.13 1.404 84 84

Mobile tariffs plans 8.43 1.292 84 84

Prepaid airtime and starter- pack

8.38 1.856 84 84
availability

Promotions 7.86 1.958 84 84

Loyalty programmes 8.77 0.91 84 84

Customer service quality 9.45 0.87 84 84

Access to mobile content and

7.75 1.352 84 84
location services

Brand reputation 8.01 1.275 84 84

All groups
Network quality 7.25 2.452 419 419
combined

Mobile tariffs 9.54 0.936 419 419

Mobile tariffs plans 7.89 1.889 419 419

Prepaid airtime and starter- pack

9.29 1.711 419 419
availability

Promotions 8.79 1.822 419 419

Loyalty programmes 8.05 2.125 419 419

Customer service quality 9.62 0.826 419 419

Access to mobile content and 6.61 2.131 419 419

58
 location services

Brand reputation 6.96 1.993 419 419

5.4.1.3. Test of equality of group means

Table 5.39: Tests of equality of group means for factors that are considered by churned

subscribers when choosing a mobile-service provider, n 419

Wilks' Lambda F df1 df2 Sig.

Network quality .917 37.614 1 417 .000

Mobile tariffs .951 21.495 1 417 .000

Mobile tariffs plans .980 8.694 1 417 .003

Prepaid airtime and starter- pack

.930 31.576 1 417 .000
availability

Promotions .934 29.240 1 417 .000

Loyalty programmes .971 12.441 1 417 .000

Customer service quality .990 4.142 1 417 .042

Access to mobile content and location

.928 32.413 1 417 .000
services

Brand reputation .931 31.124 1 417 .000

5.4.1.4. Pooled within group matrices

Table 5.40: Pooled within group matrices for factors that are considered by churned

subscribers when choosing a mobile-service provider, n 419

Prepaid Access to

airtime mobile
Mobile Customer
Network Mobile and Loyalty content Brand
tariffs Promotions service
quality tariffs starter- programmes and reputation
plans quality
pack location

availability services

Network
Correlation 1 0.082 0.018 -0.013 0.022 0.143 0.06 0.243 0.219
quality

59
 Mobile
0.082 1 0.165 0.212 0.079 0.144 0.11 0.08 0.107
tariffs

Mobile
0.018 0.165 1 0.24 0.145 0.235 0.096 0.064 0.166
tariffs plans

Prepaid

airtime and
-0.013 0.212 0.24 1 0.129 0.153 0.125 0.039 0.056
starter- pack

availability

Promotions 0.022 0.079 0.145 0.129 1 0.18 0.181 0.019 0.109

Loyalty
0.143 0.144 0.235 0.153 0.18 1 0.192 0.283 0.266
programmes

Customer

service 0.06 0.11 0.096 0.125 0.181 0.192 1 0.172 0.173

quality

Access to

mobile

content and 0.243 0.08 0.064 0.039 0.019 0.283 0.172 1 0.291

location

services

Brand
0.219 0.107 0.166 0.056 0.109 0.266 0.173 0.291 1
reputation

5.4.2 Discriminant Analysis

5.4.2.1 Box‟s test of equality of covariance matrices

Table 5.41: Box’s test of equality of covariance matrices for nine factors and two groups that

are considered by churned subscribers when choosing a mobile-service provider, n

419

Groups Factors Log determinant

Usage less than R15,00 a month 9 8.247

Usage more than R15,00 a month 9 3.280

Pooled within-groups 9 8.085

The ranks and natural logarithms of determinants printed are those of the group covariance matrices.

60
Table 5.42: Box’s test of equality of covariance matrices for nine factors and two groups that

churned subscribers consider when choosing a mobile-service provider, n 419

Box's M 344.788

sF Approx. 7.354

df1 45

df2 77491.191

Sig. .000

Tests null hypothesis of equal population covariance matrices.

Table 5.43: Descriptive statistics for nine factors and two groups that are considered by

churned subscribers when choosing a mobile-service provider, n 419

N Mean Std deviation Skewness Kurtosis

Statistic Statistic Statistic Statistic Std. Error Statistic Std. Error

Network quality 419 7.25 2.452 -.887 .119 -.278 .238

Mobile tariffs 419 9.54 .936 -2.427 .119 6.272 .238

Mobile tariffs plans 419 7.89 1.889 -1.211 .119 1.899 .238

Prepaid airtime and starter- 419 9.29 1.711 -2.531 .119 5.658 .238

pack availability

Promotions 419 8.79 1.822 -2.013 .119 4.682 .238

Loyalty programmes 419 8.05 2.125 -1.644 .119 3.091 .238

Customer service quality 419 9.62 .826 -2.357 .119 5.386 .238

Access to mobile content 419 6.61 2.131 -.213 .119 -.317 .238

and location services

Brand reputation 419 6.96 1.993 -.841 .119 1.082 .238

Valid N (listwise) 419

5.4.2.2 Summary of canonical discriminant functions

61
Table 5.44: Discriminant function Eigen values for nine factors and two groups that are

considered by churned subscribers when choosing a mobile-service provider, n

419

Function Eigen value % of variance Cumulative % Canonical correlation

a
1 .435 100.0 100.0 .550

a. First 1 canonical discriminant functions were used in the analysis.

Table 5.45: Wilks’ Lambda test for nine factors and two groups that are by considered by

churned subscribers when choosing a mobile-service provider, n 419

Test of function(s) Wilks' Lambda Chi-square Df Sig.

1 .697 148.903 9 .000

Table 5.46: Standardised canonical discriminant function coefficients for nine factors and two

groups that considered by churned subscribers when choosing a mobile-service

provider, n 419

Function

Network quality .338

Mobile tariffs -.373

Mobile tariffs plans .333

Prepaid airtime and starter- pack availability -.397

Promotions -.420

Loyalty programmes .208

Customer service quality -.172

Access to mobile content and location services .259

Brand reputation .292

Table 5.47: Structure matrix for nine factors and two groups that are considered by churned

subscribers when choosing a mobile-service provider, n 419

62
 Function

Network quality .456

Access to mobile content and location services .423

Prepaid airtime and starter- pack availability -.417

Brand reputation .414

Promotions -.402

Mobile tariffs -.344

Loyalty programmes .262

Mobile tariffs plans .219

Customer service quality -.151

Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions

Variables ordered by absolute size of correlation within function.

Table 5.48: Canonical discriminant function coefficient for nine factors and two groups that are

considered by churned subscribers when choosing a mobile-service provider, n

419

Function

Network quality .144

Mobile tariffs -.408

Mobile tariffs plans .178

Prepaid airtime and starter- pack availability -.240

Promotions -.238

Loyalty programmes .099

Customer service quality -.209

Access to mobile content and location services .126

Brand reputation .152

(Constant) 5.089

5.4.3 Discriminant Analysis classification statistics

5.4.3.1 Classification processing summary

63
Table 5.49: Classification processing summary for nine factors and two groups that are

considered by churned subscribers when choosing a mobile-service provider, n

419

Processed 861

Excluded Missing or out-of-range group codes 0

At least one missing discriminating variable 0

Used in output 861

5.4.3.2 Prior probabilities for groups

Table 5.50: Prior probabilities for groups for nine factors and two groups that are considered by

churned subscribers when choosing a mobile-service provider, n 419

Cases used in analysis

Quota Prior Unweighted Weighted

Usage less than R15,00 a month .500 335 335.000

Usage more than R15,00 a month .500 84 84.000

per month

Total 1.000 419 419.000

5.4.3.3 Classification function coefficients

Table 5.51: Classification function coefficient for nine factors and two groups that are

considered by churned subscribers when choosing a mobile-service provider, n

419

Quota

Usage less than R15,00 a Usage more than R15,00 a

month month

Network quality .690 .926

Mobile tariffs 9.388 8.717

64
Mobile tariffs plans .540 .832

Prepaid airtime and starter-

1.381 .986
pack availability

Promotions 1.285 .894

Loyalty programmes -.340 -.177

Customer service quality 12.135 11.792

Access to mobile content and

.128 .335
location services

Brand reputation .026 .275

(Constant) -120.564 -113.011

Fisher's linear discriminant functions

5.4.3.4 Classification results

Table 5.52: Classification group results for nine factors and two groups that are considered by

churned subscribers when choosing a mobile-service provider, n 419

Predicted group membership

Usage less than Usage more than

Group R15,00 a month R15,00 a month Total

Cases selected Original Count Usage less than 271 64 335

R15,00 a month

Usage more than 15 69 84

R15,00 a month

% Usage less than 80.9 19.1 100.0

R15,00 a month

Usage more than 17.9 82.1 100.0

R15,00 a month

Cross- Count Usage less than 269 66 335

a
validated R15,00 a month

Usage more than 17 67 84

R15,00 a month

% Usage less than 80.3 19.7 100.0

R15,00 a month

65
 Usage more than 20.2 79.8 100.0

R15,00 a month

Cases not Original Count Usage less than 377 48 425

Selected R15,00 a month

Usage more than 1 16 17

R15,00 a month

% Us ageless than 88.7 11.3 100.0

R15,00 a month

Usage more than 5.9 94.1 100.0

R15,00 a month

a. Cross validation is done only for those cases in the analysis. In cross validation, each case is classified by the
functions derived from all cases other than that case.
b. 81.1% of selected original grouped cases correctly classified.
c. 88.9% of unselected original grouped cases correctly classified.
d. 80.2% of selected cross-validated grouped cases correctly classified.

66
6. Data quality

6.1 Reliability of scale

Before proceeding with the analysis, it is prudent to review, among other things,

the reliability of the scale that is used in the research.

Reliability refers to the property of a measurement instrument that causes it to

give similar results for similar inputs. In addition, reliability explains the degree

of consistency of a measure (SPSS Inc, 2010).

Using the reliability analysis function in IBM SPSS Statistics 19, a reliability test

was conducted on the key data items that were used in the research.

The model used for the reliability test was Cronbach‟s Alpha. This model is a

model of internal consistency, based on the average inter-item correlation

(SPSS Inc, 2010).

In addition, Hotelling‟s T-squared test and intraclass correlation coefficient tests

were conducted.

6.1.1 Cronbach's Alpha reliability analysis

Cronbach's Alpha is a measure of reliability (Cronbach, 1951). More

specifically, Alpha is a lower bound for the true reliability of the survey.

Mathematically, reliability is defined as the proportion of the variability in the

responses to the survey by respondents (SPSS Inc, 2010).

For the research to be deemed a reliable survey, answers to questions

regarding the importance of factors that are considered by churned subscribers

67
when choosing a mobile-service provider should have differed. Respondents

should have different opinions, not because the survey was confusing or had

multiple interpretations.

Table 5.2 provides the Cronbach‟s Alpha and Cronbach‟s Alpha that is based

on standardised items. The 11 factors tested in the research had coefficients of

0.711 and 0.713 respectively for Cronbach‟s Alpha and Cronbach‟s Alpha that

is based on standardised items.

Values of 0.70 or greater are indicative of good scale reliability (O‟Leary-Kelly &

Vokurka, 1988). Considering that, the coefficients from the analysis are greater

than 0.70 would indicate the scale in the research has good reliability.

6.1.2 All factors that are considered important by churned subscribers on the

scale do not have the same mean.

Another measure to test whether the scale was reliable is to conduct Hotelling‟s

T-square test for the null hypothesis that all the factors have the same means.

Table 5.6 confirms that the null hypothesis that all the factors have the same

mean is not accepted, as the test statistic has a value that is lower than the 0.05

significant value. Thus, the null hypothesis is rejected, and the alternative

hypothesis that factors do not have the same means is accepted, further

indicating that the scale used in the research is reliable.

6.1.3 Churned subscribers were consistent in their rating of factors that they

considered when choosing a mobile-service provider

68
The average measure of intraclass correlation of 0.711, from table 5.7, indicates

that the subscribers were consistent in their ratings of the factors that they

considered when choosing a mobile-service provider.

In addition, at a 95% confidence level there is not much variation between the

lower-bound and upper-bound intraclass correlations. Similar to the Cronbach‟s

Alpha of 0.711, an intraclass correlation of greater than 0.70 is considered a

good indication of scale reliability (McGraw & Wong, 1996).

Results from Cronbach's Alpha reliability analysis, Hotellings T-squared test

and intraclass correlation coefficient test indicate that the scale used in the

survey is reliable.

69
7. Results analysis

7.1 Factors that were considered by churned prepaid subscribers when

choosing a mobile-service provider

Customer service quality was the most important factor that churned

subscribers considered when choosing a mobile-service provider. Customer

service quality had and overall rating of 9.62 out of a scale of 10, with 10 been

extremely important and 0 been totally unimportant, for a sample of 419

subscribers that churned.

Table 5.3 provides the mean scores of factors that were considered by churned

prepaid subscribers when choosing a mobile-service provider. Evident from the

table is that customer service quality, mobile tariffs, prepaid airtime, starter-

pack availability and promotions all had mean scores over 8.5, indicating that

these factors were extremely important to churned subscribers when choosing a

mobile-service provider.

In addition to rating these factors with scores over 8.5, the standard deviation of

these factors were also the lowest out of 11 factors that were considered by

churned subscribers when choosing a mobile-service provider. This would

indicate that subscribers perceived these as important factors, and that the

sample of 419 subscribers consistently rated these factors highly.

Research conducted by Rahman, Haque and Ahmad (2010) on exploring

influencing factors for the selection of mobile-phone service providers in

Malaysia also concluded that service quality, price, availability and promotions

were significant factors in influencing subscribers‟ choice of a mobile-service

70
provider. Similar research conducted by Haque, Rahman and Rahman (2010)

also concluded that service quality, price, availability and Promotions were

significant factors in influencing subscribers‟ choice of a mobile-service provider

in the Bangladesh mobile market.

Access to mobile content and location services, brand reputation and network

quality scored the lowest out of 11 factors that were considered important by

churned subscribers when choosing a mobile-service provider.

Brand reputation had the second lowest mean rating out of the 11 factors,

indicating that although important, churned subscribers did not consider the

brand reputation of a mobile-service provider first when choosing a mobile-

service provider.

In 2009, Vodacom Group Limited and MTN Group Limited reported an annual

prepaid subscriber churn rate of 45.5% and 43.5% respectively for the South

African prepaid mobile market. These figures support the answer to the

question why subscribers rated brand reputation as their second lowest factor

when choosing a mobile-service provider.

Arguably, the two different subscriber groups would have rated factors

differently. As noted previously the two groups were subscribers that spent on

average of less than R15,00 a month on their prepaid mobile service, and the

subscriber group that spent more than R15,00 a month on their prepaid mobile

service. This difference in rating of factors between the two subscriber groups

are explored in the section about Discriminant Analysis.

7.1.1 Relationships between factors that were considered important by

71
 churned subscribers when choosing a mobile-service provider

In order to establish whether churned subscribers rate factors differently, an

Analysis of Variance test was conducted to test the null hypothesis that churned

subscribers rated factors equally.

The alternative hypothesis was that churned subscribers do not rated factors

equally when choosing a mobile-service provider.

In addition to establishing whether there was a difference in mean rating

between factors that were considered by churned subscribers when choosing a

mobile-service provider, a post-hoc multiple comparison test was conducted to

establish which factors were significantly different from one another.

7.1.2 Test assumptions for Analysis of Variance

SPSS Inc (2010) lists the following assumptions that are required to conduct an

Analysis of Variance test:

 Each group is an independent and random sample from a normal

population. Considering the methodology that was used to conduct the

research, the research met this assumption, as the subscribers were

selected using simple random sampling.

In meeting the assumptions of normality, an Analysis of Variance test

was robust to departures from normality, although the data should be

symmetric (SPSS Inc, 2010).

72
 The kurtosis statistic for the data used in the analysis was 1.327, which

indicated that subscribers rated factors very highly and as such does

depart from a normally distributed sample. The high peak of the data

could be explained because of using a scale of 0 to 10 in determining the

importance of factors that were considered by churn subscribers when

choosing a mobile-service provider.

The skewness statistic was -1.297 indicating that data was skewed to the

left, and that subscribers rated factors positively and closer towards the

rating of 10. The mean score of all factors in rating the importance of

factors that were important to churned subscribers when considering a

mobile-service provider was 8.155 (see Table 5.4) which explained why

there was a negative skewness to the data.

Considering that the kurtosis and skewness did indicate that the data did

not necessarily represent a normal distribution, the Analysis of Variance

test was robust to departures from normality, and thus the analysis was

still conducted.

 The groups should come from populations with equal variances. In order

to test this assumption Levene's homogeneity-of-variance test was

conducted. From Table 5.8 the null hypothesis that the data came from a

population with equal variances was rejected. The test statistic was less

than the significant value of 0.05, indicating that the assumption that the

data came from a population of equal variance was rejected.

73
 The analysis-of-variance test did provide unequal variance tests for data

in which equal variance was rejected, these included Brown-Forsythe

and Welch tests to test the null hypothesis that the mean ratings of

factors that churned subscribers considered important when choosing a

mobile-service provider were equal.

7.1.3 Mean rating of factors were not equal for churned subscribers when

choosing a mobile-service provider

As noted in 7.1.2 the assumption of equal variance was rejected. The following

were the tests that were used to test the null hypothesis that mean scores of

factors that were considered by churned subscribers were equal:

 Brown-Forsythe: This statistic was preferable to the F statistic when the

assumption of equal variances did not hold (SPSS Inc, 2010). From

Table 5.10 the test statistic was less than the 0.05 significant level. This

implied that the null hypothesis that mean for the different factors were

equal was rejected. Thus, the alternative hypothesis was accepted that

the mean of the different factors were not equal.

 Welch: This statistic was preferable to the F statistic when the

assumption of equal variances did not hold (SPSS Inc, 2010). From

Table 5.10 the test statistic was less than 0.05, which implied that the

null hypothesis for the different factors was equal, was rejected. Thus,

the alternative hypothesis was accepted that the means of the different

factors were not equal.

74
Considering that both Welch and Brown-Forsythe tests rejected the null

hypothesis and accepted the alternative hypothesis the conclusion was that

churned subscribers rated the importance of different factors differently when

choosing a mobile-service provider.

Thus, customer service quality, mobile tariffs, prepaid airtime and starter- pack

availability and promotions were significant in influencing churned prepaid

subscribers when choosing a service provider. These factors had the highest

mean ratings (see Table 5.3).

The relationships between these factors, as well as the other seven factors are

explored further under the Discriminant Analysis section.

7.1.4 Factors that were significantly different from each other when churned

subscribers changed a mobile-service provider

In order to establish which factors were significantly different from each other

when subscribers changed a mobile-service provider, a post-hoc multiple

comparison test was conducted.

When taking into consideration that equal variance was not assumed for the

Analysis of Variance test, Tamhane's T2 post-hoc test for unequal variance was

used.

Tamhane's T2 post-hoc test is a conservative pairwise comparisons test that is

based on a T test. This test is appropriate when the variances are unequal

(SPSS Inc, 2010). The results from the test were the following:

7.1.4.1 Importance of customer service quality and when choosing a mobile-

75
 service provider

Customer service quality had a mean rating of 9.62 out of 11 factors, and,

based on mean rating, was listed as the most important factor that was

considered by churned subscribers when choosing a mobile-service provider.

In addition, the mean of customer service quality was significantly different from

means of the other nine factors. Churned subscribers rated customer service

quality and mobile tariffs equally when choosing a mobile-service provider.

7.1.4.2 Importance of mobile tariffs when choosing a mobile-service provider

Mobile tariffs had a mean rating of 9.54 out of 11 factors, and, based on mean

rating, was listed as the second most important factor that was considered by

churned subscribers when choosing a mobile-service provider.

In addition, the mean of mobile tariffs was significantly different from the means

of the other eight factors. Churned subscribers rated mobile tariffs, customer

service quality, prepaid airtime and starter- pack availability equally when

choosing a mobile-service provider.

7.1.4.3 The importance of prepaid airtime and starter- pack availability when

choosing a mobile-service provider

Prepaid airtime and starter- pack availability had a mean rating of 9.29 out of 11

factors, and, based on mean rating, was the third most important factor that was

considered by churned subscribers when choosing a mobile-service provider.

76
In addition, the mean of prepaid airtime and starter- pack availability was

significantly different from the means of the other nine factors. Churned

subscribers rated prepaid airtime, starter-pack availability and mobile tariffs

equally when choosing a mobile-service provider.

7.1.4.4 The importance of promotion when choosing a mobile-service provider

Promotion had a mean rating of 8.79 out of 11 factors, and, based on mean

rating, was the fourth most important factor that was considered by churned

subscribers when choosing a mobile-service provider.

In addition, the mean of promotion was significantly different from the means of

the other 10 factors.

7.1.4.5 The importance of different airtime recharge options when choosing a

mobile-service provider

Different airtime recharge options had a mean rating of 8.16 out of 11 factors,

and, based on mean, were the fifth most important factor that was considered

by churned subscribers when choosing a mobile-service provider.

In addition, the mean of different airtime recharge options was significantly

different from the means of the other eight factors. Churned subscribers rated

different airtime recharge options, mobile tariff plans and loyalty programmes

equally when choosing a mobile-service provider.

7.1.4.6 The importance of loyalty programmes when choosing a mobile-service

provider

77
Loyalty programmes had a mean rating of 8.05 out of 11 factors, and, based on

mean rating, was listed as the sixth most important factor that was considered

by churned subscribers when choosing a mobile-service provider.

In addition, the mean of loyalty programmes was significantly different from

means of the other seven factors. Churned subscribers rated loyalty

programmes, mobile tariff plans, different airtime recharge options and different

types of prepaid product bundles equally when choosing a mobile-service

provider.

7.1.4.7 Importance of mobile tariff plans when choosing a mobile-service

provider

Mobile tariff plans had a mean rating of 7.89 out of 11 factors, and, based on

mean, was listed as the seventh most important factor that was considered by

churned subscribers when choosing a mobile-service provider.

In addition, the mean of mobile tariff plans was significantly different from the

means of the other seven factors. Churned subscribers rated mobile tariff plans,

loyalty programmes, different airtime recharge options and different types of

prepaid product bundles equally when choosing a mobile-service provider.

7.1.4.8 Importance of different types of prepaid product bundles when choosing

a mobile-service provider

Different types of prepaid product bundles had a mean rating of 7.54 out of 11

factors, and were listed as the eight most important factors that were

considered by churned subscribers when choosing a mobile-service provider.

78
In addition, the mean for different types of prepaid product bundles was

significantly different from the means of the other seven factors. Churned

subscribers rated different types of prepaid product bundles, network quality,

mobile tariff plans and loyalty programmes equally when choosing a mobile-

service provider.

7.1.4.9 Network quality‟s importance when choosing a mobile-service provider

Network quality had a mean rating of 7.25 out of 11 factors, and, based on

mean, was listed as the ninth most important factor that was considered by

churned subscribers when choosing a mobile-service provider.

In addition, the mean of network quality was significantly different from the

means of the other eight factors. Churned subscribers rated network quality,

different types of prepaid product bundles and brand reputation equally when

choosing a mobile-service provider.

7.1.4.10 Importance of brand reputation and when choosing a mobile-service

provider

Brand reputation had a mean rating of 6.96 out of 11 factors, and, based on

mean rating, was the tenth most important factor that was considered by

churned subscribers when choosing a mobile-service provider.

In addition, the mean of brand reputation was significantly different from the

means of the other eight factors. Churned subscribers rated brand reputation,

network quality and access to mobile content and location services equally

when choosing a mobile-service provider.

79
7.1.4.11 Importance of access to mobile content and location services when

choosing a mobile-service provider

Access to mobile content and location services had a mean rating of 6.61, and,

out of 11 factors, was the eleventh most important factor that was considered by

churned subscribers when choosing a mobile-service provider.

In addition, the mean of access to mobile content and location services was

significantly different from means of the other nine factors. Churned subscribers

rated access to mobile content, location services and brand reputation equally

when choosing a mobile-service provider.

7.2 Discriminant Analysis

7.2.1 Discriminant model assumptions

SPSS Inc (2010) lists the following assumptions that are to be met or that are

required to run a discriminant model:

 The predictors must not be highly correlated with each other

 The mean and variance of a given predictor are not correlated

 The group variance-co-variance matrices are equivalent

 Group membership is assumed to be mutually exclusive and collectively

exhaustive

 The values of each predictor have a normal distribution

7.2.1.1 Predictors are not highly correlated

80
Table 5.40 provides the correlations between the different factors. Evident from

Table 5.40 is that the factors are not highly correlated, and the highest

correlation is 0.291.

Hays (1981) noted that a perfect correlation had an absolute value of 1, thus

0.20 can be considered as an uncorrelated data set. Thus, the assumption

needed to run a discriminant model that the predictors were not highly

correlated with each other, has not been violated.

7.2.1.2 The mean and variance of a given factor are not correlated

To test the assumption that mean and variance of a given predictor are not

correlated, SPSS did a test of equality of group means using Univariate

Analysis of Variance.

Univariate Analysis of Variance performed a one-way analysis-of-variance test

for equality of group means for each independent variable (SPSS Inc, 2010).

The factors that are used in the discriminant model are listed in Table 5.39. As

is evident in the table, all the factors that are used in the model have means and

variances that are not correlated. All factors have a significance level, which is

less than the 0.05 for the null hypothesis that the factors have equal means.

The following factors were removed from the discriminant model:

 Different airtime recharge options

 Different types of prepaid product bundles

81
The above factors were removed from the discriminant model because they had

means and variances that were correlated.

With the removal of the above factors, the assumption for mean and variance of

a given factor not being correlated with another factor has not been violated.

7.2.1.3 The group variance-co-variance matrices are equivalent

Discriminant Analysis is dependent on the assumption that the variance-co-

variance matrices are equivalent across groups.

SPSS uses the Box‟s M tests to validate this assumption by using the null

hypothesis that the covariance matrices do not differ between groups formed by

the dependent.

In addition, for this assumption to hold, the log determinants should be equal

(Burns & Burns, 2008). Log determinants are a measure of the variability of the

groups, and large differences in log determinants indicate groups that have

different covariance matrices (SPSS Inc, 2010).

The log determinants in Table 5.41 differ from each other, and appear not to be

equal. This indicates that the groups have different covariance matrices. In

addition, the Box M test (see Table 5.42) has a test statistic, which is less than

the significance value of 0.05, which indicates that the test is significant in

rejecting the null hypothesis.

The null hypothesis states that the covariance matrices do not differ between

groups formed by the dependent, and as such, the alternative hypothesis is

accepted that the co-variances matrices differ between groups.

82
Since the co-variances differ between groups, the assumption to conduct a

discriminant model is violated. However, with large samples, a significant result

is not regarded as too important (Burns & Burns, 2008). It should be noted that

the research has a sample size of 861, of which 419 subscribers churned in the

last 12 months, and in the light of this, a significant result is not regarded as

important. SPSS Inc (2010) notes that for sufficiently large samples, a non-

significant p value indicates that there is insufficient evidence that the matrices

differ.

However, SPSS Inc (2010) recommends the running of Discriminant Analysis

by using separate group co-variances matrices to determine if the classification

results differ by much. The classification results did not change from the 81.1%

result, which was correctly classified, thus the Discriminant Analysis was

conducted by using the within group co-variances.

7.2.1.4 Group membership is assumed to be mutually exclusive and

collectively exhaustive

Since the research has two categories of prepaid subscribers, namely those

subscribers that spend less than R15,00 a month on prepaid service, and those

that spend more than R15,00 a month on prepaid service, no subscriber can

belong to more than one group, as each subscriber is unique to a specific

group. Thus, the assumption that group membership is assumed to be mutually

exclusive is not violated.

In addition, each subscriber if a member of a group, and the group membership

is collectively exhaustive within the subsample of 419 subscribers who churned

83
in the last 12 months from date of interview. Thus, the assumption is not

violated that group membership is assumed collectively exhaustive.

Furthermore, Burns and Burns (2008) noted that Discriminant Analysis was

most effective when group membership was a truly categorical variable. Group

membership in the research was a truly categorical variable with one binary

variable for two categories.

7.2.1.5 The values of each predictor have a normal distribution

The factors that were considered important by churned subscribers when

choosing a mobile-service provider were obtained by using a rating scale. The

rating scale resulted in the data being skewed and peaked, indicating that the

data was not normally distributed. Since the data was not normally distributed

this violated the assumption that each predictor should have a normal

distribution.

Table 5.43 provides the skewness and kurtosis for each factor. Six factors are

peaked and skewed indicating an abnormal distribution, with the exception of

the following three factors:

- Network quality

- Access to mobile content and location services

- Brand reputation

However, as long as the sample sizes are large, group sizes are about equal,

and outliers are minimal, Discriminant Analysis is robust in tolerating violations

of this assumption (Burns & Burns, 2008).

84
The above statement is true for the research data, since it is a large sample set

of 419 selected cases. In addition, the Discriminant Analysis was conducted

using the function of “all groups equal” which assumes equal prior probabilities

for all groups instead of using observed group sizes in the sample to determine

the prior probabilities of group membership.

While the data does not follow a normal distribution, the sample size, adjusting

for group sizes using equal prior probabilities and the absences of outliers the

Discriminant Analysis was still conducted although the assumption of normality

was violated.

7.2.2 Discriminant model

As noted, Discriminant Analysis is used for predictive Discriminant Analysis and

group classification, which address the question of how to assign new cases to

groups.

The following sections will focus on the discriminant function coefficients that

are used to predict usage groups of churn subscribers by using factors that they

consider important when choosing a mobile-service provider.

7.2.2.1 Canonical discriminant functions

Table 5.44 provides information on the discriminant function that was produced.

The maximum number of discriminant functions produced is one, as there are

only two groups.

The canonical correlation is the multiple correlation between the predictors and

the discriminant function (Burns & Burns, 2008). The canonical correlation

85
provides an index of overall model fit, which is interpreted as being the

proportion of variance, explained by R2 (Burns & Burns, 2008).

From Table 5.44 a canonical correlation of 0.550 suggests that the model

explains 31% of the variation in the grouping variables, and that the model is

reasonable in prediction. A perfect classification model will have a canonical

correlation of 1 (Levine, 1977).

7.2.2.2 Wilks‟ lambda

Wilks‟ lambda indicates the significance of the discriminant function. Table 5.45

indicates a highly significant function as the test statistic is less than the

significant level of 0.05 for the function, and provides the proportion of total

variability not explained, i.e. it is the converse of the squared canonical

correlation (Burns & Burns, 2008).

A significant function confirms the alternative hypothesis that the model

produces a significant discriminant function.

7.2.2.3 Standardised canonical discriminant function coefficients

The interpretation of the discriminant coefficients (or weights) is like that in

multiple regression (Burns & Burns, 2008). Table 5.46 provides an index of the

importance of each factor to discriminant function. The importance of each

factor is like the standardised regression coefficients (beta‟s) in multiple

regression (Burns & Burns, 2008), and the sign indicates the direction of the

relationship.

86
The following are the strongest predictors for the model in descending order of

importance:

1. Promotions

2. Prepaid airtime and starter-pack availability

3. Mobile tariffs

4. Network quality

5. Mobile tariff plans

6. Brand reputation

7. Access to mobile content and location services

8. Loyalty programmes

9. Customer service quality

7.2.2.4 Structure matrix

The structure matrix provides another way of indicating the relative importance

of the factors used by churned subscribers when choosing a mobile-service

provider.

The structure matrix in Table 5.47 shows the correlations of each variable with

each discriminate function. The Pearson coefficients in Table 5.47 are structure

coefficients or discriminant loadings and notwithstanding problem

dimensionality and data scaling 0.30 is the cut-off between important and less

important variables (Burns & Burns, 2008).

In order of importance, as is evident from Table 5.47, factors that have structure

coefficients equal to 0.30 or greater are the following:

87
 1. Network quality

2. Access to mobile content and location services

3. Prepaid airtime and starter-pack availability

4. Brand reputation

5. Mobile tariffs

6. Promotions

7.2.2.5 Canonical discriminant function

The canonical discriminant function coefficients in Table 5.48 provide the

unstandardised coefficients that were used to create the discriminant function or

equation (Burns & Burns, 2008).

From Table 5.48 the following equation is created for the discriminant model to

predict usage groups of churn subscribers by using factors that they consider

important when choosing a mobile-service provider:

Churned subscribers usage group = (0.144 X network quality) + (-0.408 X

mobile tariffs) + (0.178 X mobile tariffs plans) + (-0.240 X prepaid airtime and

starter-pack availability) + (-0.238 X promotions) + (0.099 X loyalty

programmes) + (-0.209 X customer service quality) + (0.126 X access to mobile

content and location services) + (0.152 X brand reputation) + 5.089

7.2.3 Discriminant classification model

7.2.3.1 Classification processing summary

The discriminant classification model processed 861 cases (see Table 5.41).

However, only 419 cases created the discriminant classification model. The 419

88
cases were subscribers who churned within the previous 12 months from the

date of the interview.

The minimum ratio of valid cases to independent variables for Discriminant

Analysis is 5 to 1, with a preferred ratio of 20 to 1 (Burns & Burns, 2008). In the

analysis, there were 419 valid cases and 9 independent variables (after the

reduction that was made in Section 7.2.1.2). Thus the ratio of cases to

independent variables is 46.56 to 1, which satisfies the minimum requirement,

as well as the preferred requirement. Thus, the sample size is more than

adequate to conduct Discriminant Analysis.

7.2.3.2 Prior probabilities for groups

As noted in Section 7.2.1.5, the Discriminant Analysis was conducted with “all

groups equal”, which assumes equal prior probabilities for all groups, as

opposed to using observed group sizes in the sample to determine the prior

probabilities of group membership.

Equal prior probabilities for all groups was used to compensate for the data not

been normal. In Table 5.50 the prior probabilities are split equally for the two

groups.

Discriminant Analysis requires that there be a minimum number of cases in the

smallest group defined by the dependent variable. In addition, the number of

cases in the smallest group must be larger than the number of independent

variables, and must preferably contain 20 or more cases (Burns & Burns, 2008).

From Table 5.50 the smallest group has 84 cases, which is larger than the 9

independent variables and contains more than the preferred 20 cases.

89
7.2.3.3 Classification function coefficients

Fisher's classification function coefficients are used to create a classification

equation that classifies cases. In addition, a separate set of classification

function coefficients is obtained for each group (SPSS Inc, 2010).

Using the function coefficients (see Table 5.51) of the factors that were

considered by churned subscribers when choosing a mobile-service provider

the following equations were created to classify the two groups, which are

based on average monthly spending.

7.2.3.3.1 Less than R15,00 a month spent on prepaid service

Classification equation: Less than R15,00 a month spent on prepaid service =

(0.690 X network quality) + (9.388 X mobile tariffs) + (0.540 X mobile tariff

plans) + (1.381 X prepaid airtime and starter-pack availability) + (1.285 X

promotions) + (-0.340 X loyalty programmes) + (12.135 X customer service

quality) + (0.128 X access to mobile content and location services) + (0.026 X

brand reputation) -120.564

Notably from the above equation is that customer service has the highest

function coefficient. This is not surprising, considering that churned subscribers

rated customer service quality, based on mean, as the most important factor

that was considered when choosing a mobile-service provider. Considering

subscribers who churned within the previous 12 months would indicate that the

customer service quality at the previous mobile-service provider was

substandard that they considered customer service quality as the highest mean

rated factor.

90
Mobile tariffs had the second highest function coefficient , which relates back to

mobile tariffs being rated as the second most important factor that was

considered by churned subscribers when choosing a mobile-service provider.

Taking into consideration that on average this group spent less than R15,00 a

month on their prepaid mobile service, it would indicate that these subscribers

were price sensitive, and would churn out of their current mobile-service

provider if they were offered cheaper tariffs accompanied by good customer

service quality.

Prepaid airtime and starter-pack availability have the third-highest function

coefficient, while promotions have the fourth-highest function coefficients.

These factors respectively were rated third and fourth in importance when they

were considered by churned subscribers for choosing a mobile-service provider.

In summary: Given the relative size of the function coefficients of customer

service quality, mobile tariffs, prepaid airtime and starter-pack availability and

promotions would indicate that mobile subscribers who churned are very

sensitive to these factors, and would most likely churn again if these factors are

not met. Customer service quality and mobile tariffs are especially sensitive,

considering the function coefficients, thus customer service quality and mobile

tariffs that do not meet the expectations of subscribers, and could result in

subscribers churning out of current a mobile-service provider.

7.2.3.3.2 More than R15,00 a month spent on prepaid service

Classification equation: More than R15,00 a month spent on prepaid

service = (0.926 X network quality) + (8.717 X mobile tariffs) + (0.832 X mobile

91
tariffs plans) + (0.986 X prepaid airtime and starter-pack availability) + (0.894 X

promotions) + (-0.177 X loyalty programmes) + (11.792 X customer service

quality) + (0.335 X access to mobile content and location services) + (0.275 X

brand reputation) -113.924

Similar to churned subscribers who spend less the R15,00 a month on their

prepaid service, subscribers that spend more than R15,00 a month on their

prepaid service had high function coefficients for customer service quality,

mobile tariffs, prepaid airtime and starter-pack availability.

7.2.3.4 Function coefficients between groups

Network quality

Subscribers that spend more than R15,00 a month on their prepaid service

(Group 2) had a higher function coefficient than subscribers that spend less

than R15,00 a month on their prepaid service (Group 1). This would indicate

through variance of 0.24 on function coefficient that Group 2 perceived network

quality as a higher factor when choosing a mobile-service provider.

Mobile tariffs

Group 1 had a higher function coefficient than Group 2. This would indicate

through variance of 0.67 on function coefficient that subscribers that spend less

than R15,00 a month on their prepaid service (Group 1) are more sensitive to

mobile tariffs when choosing a mobile-service provider.

Mobile tariff plans

92
Group 2 had a higher function coefficient than Group 1. This would indicate

through variance of 0.29 on function coefficient that subscribers that spend

more than R15,00 a month on their prepaid service (Group 2) are influenced a

lot more by the types of mobile tariff plans that are offered by a mobile-service

provider when choosing a service provider.

Prepaid airtime and starter-pack availability

Group 1 had a higher function coefficient than Group 2. This would indicate

through variance of 0.40 between function coefficients that subscribers that

spend less than R15,00 a month on their prepaid service (Group 1) are

influenced a lot more by the prepaid airtime and starter-pack availability offered

by a mobile-service provider when choosing a service provider.

Promotions

Group 1 had a higher function coefficient than Group 2. This would indicate

through variance of 0.39 between function coefficients that subscribers that

spend less than R15,00 a month on their prepaid service (Group 1) are

influenced a lot more by the promotions offered by a mobile-service provider

when choosing a service provider.

Loyalty programmes

Group 2 had a higher function coefficient than Group 1. This would indicate

through variance of 0.16 between function coefficients that subscribers that

spend more than R15,00 a month on their prepaid service (Group 2) are

93
influenced a lot more by loyalty programmes offered by a mobile-service

provider when choosing a service provider.

Customer service quality

Group 1 had a higher function coefficient than Group 2. This would indicate

through variance of 0.34 between function coefficients that subscribers that

spend less than R15,00 a month on their prepaid service (Group 1) are

influenced a lot more by the customer service quality of a mobile-service

provider when choosing a service provider.

Access to mobile content and location services

Group 2 had a higher function coefficient than Group 1. This would indicate

through variance of 0.21 between function coefficients that subscribers that

spend more than R15,00 a month on their prepaid service (Group 2) are

influenced a lot more by the mobile content and location services offered by a

mobile-service provider when choosing a service provider.

Brand reputation

Group 2 had a higher function coefficient than Group 1. This would indicate

through variance of 0.25 between function coefficients that subscribers that

spend more than R15,00 a month on their prepaid service (Group 2) are

influenced a lot more by the brand reputation of a mobile-service provider when

choosing a service provider.

7.2.3.5 Classification results

94
The classification table is a table in which the rows are the observed categories

of the dependent, and the columns are the predicted categories (SPSS Inc,

2010). When the prediction is perfect all cases will lie on the diagonal and the

percentage of cases on the diagonal is the percentage of correct classifications

(SPSS Inc, 2010).

The cross-validated set of data is a more honest presentation of the power of

the discriminant function than that provided by the original classifications, and

often produces a poorer outcome (SPSS Inc, 2010).

The classification results from Table 5.52 reveal that 81.1% of selected original

grouped cases correctly classified churned subscribers that spend less than

R15,00 monthly, and more than R15,00 monthly, on prepaid service in the

respective groups.

However, as noted above, SPSS Inc (2010) recommends using the cross-

validated set of data. The cross-validated set of data has 80.2% of selected

cross-validated grouped cases correctly classified into churned subscribers that

spend less than R15,00 monthly, and more than R15,00 monthly, on prepaid

service.

Burns and Burns (2008), note that the independent variables could be

characterised as useful predictors of membership in the groups defined by the

dependent variable, if the cross-validated classification accuracy rate were

significantly higher than the accuracy attainable by chance alone. Operationally,

the cross-validated classification accuracy rate should be 25% or more higher

than the proportional by chance accuracy rate (Burns & Burns, 2008).

95
The proportional by chance accuracy rate is calculated by squaring and

summing the proportion of cases in each group (Burns & Burns, 2008). The

proportional by chance accuracy rate is 50% (0.50² + 0.50² = 0.005) for the

classification analysis.

The cross-validated accuracy rate is 80.1%, which is greater than proportional

by chance accuracy criteria of 50%, implying that the criterion for classification

accuracy is satisfied.

In summary Discriminant, analysis was conducted to predict the groups churned

subscribers belong too by using the factors that were considered important by

churned subscribers when choosing a mobile-service provider.

The following were predictor factors:

 Promotions

 Prepaid airtime and starter-pack availability

 Mobile tariffs

 Network quality

 Mobile tariff plans

 Brand reputation

 Access to mobile content and location services

 Loyalty programmes

 Customer service quality

96
Significant mean differences were observed for most of the factors, while the

other factors were removed because there was no significant mean difference

between them.

The log determinants were not very similar and the Box‟s M indicated that the

assumption of equality of covariance matrices was violated. However, given the

large sample of 419 selected cases, the problem was not regarded as serious.

The discriminate function revealed an association (R2) of 0.55 between groups

and all predictors, accounting for 30.3% of between group variability.

The structure matrix revealed six predictors, namely network quality, access to

mobile content and location service, prepaid airtime and starter-pack

availability, brand reputation, mobile tariffs and promotions.

The cross-validated accuracy rate was 80.1%, which was greater than

proportional by chance accuracy criteria of 50%, implying that the criterion for

classification accuracy was satisfied.

97
8. Conclusion

As stated in the research objective, the benefits derived from conducting the

research were the following:

 Adding to the knowledge on prepaid subscriber retention in the high

mobile churn market of South Africa

 Adding to the knowledge on prepaid subscriber acquisition in the high

mobile churn market of South Africa

 Adding to the knowledge on optimal resource allocation of limited

business resources

The above benefits were dependent on reaching the following research results:

 Determining the factors that were statistically significant when churned

prepaid subscribers changed a mobile-service operator

 Determining the factors that churned prepaid subscribers considered to

be equal when choosing a mobile-service provider, and similarly the

factors that they did not consider equal

 Creating a classification discriminant model that is based on monthly

spending categories, to predict and classify the groups to which churned

prepaid subscribers will belong by using the factors that they consider

important when choosing a mobile-service provider

Determining factors that are statistically significant when churned prepaid

subscribers change a mobile-service operator

98
The following factors in order of their decreasing importance were established,

and given in Section 7.1:

1. Customer service quality

2. Mobile tariffs

3. Prepaid airtime and starter-pack availability

4. Promotions

The above were the four factors that were considered the most important by

churned subscribers when choosing a mobile-service provider. These factors

are considered to be similar in importance to what mobile subscribers in

Malaysia (Rahman, Haque, & Ahmad, 2010) and mobile subscribers in

Bangladesh (Haque, Rahman, & Rahman, 2010), consider being important.

Both Malaysia and Bangladesh are developing and unsaturated markets like

South Africa.

By listing, customer service quality as the most influential factor it would imply

that the previous service provider‟s customer service quality was not perceived

as acceptable for churned subscribers. In addition, considering that the top two

mobile-service providers both have very similar churn rates, this implies that

their customer service quality is very similar as perceived from within the survey

group. Thus, there is an opportunity for mobile-service providers to differentiate

themselves by effecting improvements in their customer service quality.

Similarly, improvement in customer service quality, while keeping all other

factors the same, will in turn lead to improvement in subscriber acquisition.

However, it may be argued that customer service quality is a retention factor, as

99
opposed to being a primary acquisition factor, because the survey group listed

customer service quality as the most important factor for choosing a service

provider (Brodie, Whittome, & Brush, 2009).

Mobile tariffs were the second most important factor that was considered

important by churn subscribers when choosing a mobile-service provider.

Considering that subscribers churned from their previous service provider to a

new service provider based on mobile tariffs would indicate that mobile tariffs

are an acquisition factor.

However, using mobile tariffs as an acquisition factor, while keeping all other

factors the same, would only be a cyclic short-term strategy that could lead to

short-term results, as it would be easy for service providers to replicate and

marginally improve on mobile tariffs for the subscribers to churn, and join yet

another new service provider.

Considering that in Section 7.1.4.1, the research statistically concluded that

churned subscribers rated customer service quality and mobile tariffs equally

when choosing a mobile-service provider, it is recommended that these two

factors be leveraged for acquisition and retention. For example, by decreasing

mobile tariffs, a mobile-service provider could acquire subscribers from another

service provider, while simultaneously improving the customer service quality

that these acquired subscribers would receive could result in retaining the

acquired subscribers for a longer period.

Prepaid airtime and starter-pack availability is, among other things, a function of

the distribution strength of a service provider, and, with all other factors

100
remaining the same, the service provider that has the better distribution would

likely acquire more subscribers.

While airtime and starter-pack availability is important to acquire subscribers,

promotions can act both as acquisition and retention factors. However using

promotions for acquisitions is a short-term strategy, and thus benefits will be a

short-lived solution, as subscribers may churn to the service provider that offers

the better promotion in the market, if all other factors are equal and perceived

as such.

In order of decreasing importance, the other 7 factors that were considered

important by churned subscribers are given as follows in Section 7.1:

5. Different airtime recharge options

6. Loyalty programmes

7. Mobile tariff plans

8. Different types of prepaid product bundles

9. Network quality

10. Brand reputation

11. Access to mobile content and location services

Factors that were considered equal by churned prepaid subscribers when

choosing a mobile-service provider, and similarly, factors that were not

considered by them to be equal

Customer service quality, mobile tariffs, airtime and starter-pack availability and

promotions can be summarised as Level 1 Factors (see Figure 8.1) that are

considered by churn subscribers when choosing a mobile-service provider.

101
These were the four most important factors from Section 7.1.2, that churned

subscribers considered, and they can therefore be grouped together as the

primary value proposition that churned subscribers consider when choosing a

mobile-service provider.

Figure 8.1: Four level of factors that churned subscribers consider important
when choosing a mobile-service provider

Source: Adapted from section 7.1 of research

As concluded in Section 7.1.4.5, churned subscribers, rate different airtime

recharge options, mobile tariff plan types and loyalty programmes on a par

when choosing a mobile-service provider. These factors are summarised as

Level 2 Factors (see Figure 8.1) that churn subscribers consider when choosing

a mobile-service provider.

In addition, the above factors are classified as essential product offerings that

are required by churned subscribers. These factors, however, can be replicated,

and do not provide a competitive differentiate. With the exception of loyalty

102
programmes (which is a retention factor), the others are considered acquisition

factors (Kotler & Keller, 2009).

Level 3 Factors (see Figure 8.1) include network quality, different types of

prepaid product bundles and brand reputation. Churned subscribers as

concluded in Section 7.1.4.6 rate them equally. Brand reputation was listed

equally with access to mobile content and location services. Brand reputation

and network quality relate to the competency of the service provider and act

both as retention and acquisition factors. Furthermore, these factors are

differentiating factors, which a service provider can control, and can increasingly

build competencies into.

Churned subscribers rate access to mobile content and location services and

brand reputation similarly as is concluded in Section 7.1.4.10. The above are

grouped as Level 4 Factors. These factors do not provide long-term

differentiation, but act as short-term acquisition factors. The uniqueness of the

product offering will diminish as other competitors build up their competencies,

and offer the same suit of products.

Having determined the factors that are important, the factors that are equal to

each other, and the factors that are not equal when churned prepaid

subscribers change a mobile service operator, the research was able to add to

the knowledge relating to retention and acquisitions of prepaid churned

subscribers. Figure 8.2 summarises those acquisition and retention factors

which are concluded from the survey sample, and which prepaid subscribers

consider when choosing a mobile-service provider

103
Figure 8.2: Acquisition and retention factors that prepaid subscribers consider when choosing a
mobile-service provider in decreasing order of importance

Source: Derived from analysis of the research analysis in Sections 7.1

Discriminant model, based on monthly spending categories, to predict and

classify the groups that churned prepaid subscribers will belong to by using the

factors that they consider important when choosing a mobile-service provider

The research established two classification equations in Section 7.2.3.3 to

classify the groups, based on monthly spending categories that churn

subscribers, will belong to by using the factors that they consider important

when choosing a mobile-service provider.

With a cross-validated accuracy rate of 80.1%, the classification equations

established that the discriminant model correctly classifies churned subscribers

into the two groups (R< R>15). In addition, Section 7.2.3.4 established the

following difference between factors of the two groups:

104
 Network quality: Subscribers who spend more than R15,00 a month on

their prepaid service perceived network quality as a higher factor when

choosing a mobile-service provider

 Mobile tariffs: Subscribers who spend less than R15,00 a month on

their prepaid service are more sensitive to mobile tariffs when choosing a

mobile-service provider

 Mobile tariff plan types: Subscribers who spend more than R15,00 a

month on their prepaid service are influenced a lot more by the types of

mobile tariff plan types offered by a mobile-service provider when

choosing a service provider

 Prepaid airtime and starter-pack availability: Subscribers who spend

less than R15,00 a month on their prepaid service are influenced a lot

more by prepaid airtime and starter-pack availability of a mobile-service

provider when choosing a service provider

 Promotions: Subscribers who spend less than R15,00 a month on their

prepaid service are influenced a lot more by the promotions offered by a

mobile-service provider when choosing a service provider

 Loyalty programmes: Subscribers who spend more than R15,00 a

month on their prepaid service are influenced a lot more by loyalty

programmes offered by a mobile-service provider when choosing a

service provider

 Customer service quality: Subscribers who spend less than R15,00 a

month on their prepaid service are influenced a lot more by customer

service quality of a mobile-service provider when choosing a service

provider

105
  Access to mobile content and location services: Subscribers who

spend more than R15,00 a month on their prepaid service are influenced

a lot more by mobile content and location services offered by a mobile-

service provider when choosing a service provider

 Brand reputation: Subscribers who spend more than R15,00 a month

on their prepaid service are influenced a lot more by the brand reputation

of a mobile-service provider when choosing a service provider

Taking the above into consideration, one has to conclude that given service,

providers have limited resources and competencies will not be able to address

all the factors that are considered important by churned subscribers when

choosing a mobile-service provider. Thus, service providers have to make

decisions about their portfolios of subscriber bases, and how to optimally

address them, given their resources and competencies, by understanding the

factors that are important to them when choosing a mobile-service provider.

The research was able to provide equations on both how to predict and classify

churned subscribers into groups based on monthly spending on usage. Based

on these equations the mobile-service providers will be able to optimally

address the mix of their subscriber base.

Furthermore, apart from addressing the desired subscriber mix, mobile-service

providers can align tactical and strategic actions by positioning themselves to

reach their targeted subscriber mix based on the factors that subscribers

consider important when choosing a mobile-service provider.

8.1 Limitations of the study

106
 - The data used in the research violated assumptions of a normal

distribution. However, the tests that were conducted by using the data

were robust to the violations of a normal distribution

- The research used the subscribers from one service provider for the

sample. However statistical tests were limited to subscribers who

churned to one specific provider in the previous 12 months from the date

of the interview

- The research did not test whether there was a significant difference

between the factors why subscribers churned and the factors considered

for choosing first time a mobile service provider

8.2 Future research recommendations

The research has established the factors that were considered important by

churned mobile subscribers when choosing a mobile-service provider. However,

these factors need to be explored in more detail, since the underlying factors

that drive the primary factors that were established in the research are not

sufficiently understood. Future research should explore the following:

- Customer service quality expectations of prepaid mobile subscribers

- Customer service quality reputation of mobile-service operators

- Mobile tariff expectations of prepaid mobile subscribers

- The influence of mobile tariffs on the reputation of mobile-service

providers

- Prepaid airtime and starter-pack availability, and its influence on prepaid

mobile subscribers

107
- The effective use of promotions to increase the retention of prepaid

mobile subscribers

108
9. References

8ta. (2011, July 21). Retrieved July 21, 2011, from 8ta: http://www.8ta.com

Alvarez, B. A., & Casielles, R. V. (2005). Consumer Evaluations of Sales Promotion: The Effect on
Brand Choice. European Journal of Marketing , 39 (1-2), 54-70.

Andreassen, T. W., & Lindestad B, B. (1988). Customer loyalty and complex services: the
impact of corporate image on quality, customer satisfaction and loyalty for customers with
varying degrees of service expertise. The International Journal of Service Industry , 9 (1), 7–23.

Barny, J. B. (1991). Firm Resources and Sustained Competitive Advantage. Journal of

Managment , 17 (1), 99-120.

Barrantesa, R., & Galperin, H. (2008). Can the poor afford mobile telephony? Evidence from
Latin America. Telecommunications Policy , 32 (8), 521-530.

Birke, D., & Swann, G. M. (2006). Network effects and the choice of mobile phone operator.
Journal of Evolutionary Economics , 16 (1-2), 65-84.

Blumberg, B., Cooper, D. R., & Schindler, P. S. (2008). Business Research Methods, Second
European Edition. Berkshire: McGraw-Hill Higher Education.

Boyer, K. K., & Hult, G. T. (2005). Customer Behavior in an Online Ordering Application: A
Decision Scoring Model. Decision Sciences , 36 (4), 569-598.

Brodie, J. R., Whittome, J. R., & Brush, G. J. (2009). Investigating the service brand: A customer
value perspective. Journal of Business Research , 62 (3), 345-355.

Burns, R., & Burns, R. (2008). Business Research Methods and Statistics Using SPSS. London:
Sage.

Cellc. (2011, July 21). Retrieved July 21, 2011, from Cellc: http://www.cellc.co.za

Chan-Olmsted, S. M., & Guo, M. (2011). Strategic Bundling of Telecommunication Services:

Triple-Play Strategies in the Cable TV and Telephone Industries. Journal of Media Business
Studies , 8 (2), 63-81.

Conradie, B., Goldin, J., Leiman, A., Standish, B., & Visser, M. (2001). Competition Policy and
Privatisation in the South African Water Industry. University of Cape Town, Development Policy
Research Unit. Capetown: University of Cape Town.

Crawford, S. G., & Cullen, J. (2007). Bundling, product choice, and efficiency: Should cable
television networks be offered a` la carte? Information Economics and Policy , 19 (3-4), 379-
404.

Cronbach, L. (1951). Coefficient alpha and the internal structure of tests. Psychometrika , 16
(3), 297-334.

109
Eberhard, A. (2002). Competition and Regulation in the Electricity Supply Industry in South
Africa. Trade and Industrial Policy Secretariat. Capetown: Trade and Industrial Policy
Secretariat.

Ersoy, N. F., & Calık, A. U. (2010). Brand Loyalty: Emotional Devotion or Rational Behavior – A
Study on Mobile Telephones from Eskisehir, Turkey. The Business Review, Cambridge , 15 (1),
212 -219.

Ferguson, R., & Hlavinka, K. (2006). The long tail of loyalty: how personalized dialogue and
customized rewards will change marketing forever. Journal of Consumer Marketing , 23 (6),
357-361.

Gorchels, L., Marien, E., & West, C. (2004). The managers guide to distribution channels. New
York: McGraw-Hill Companies, Inc.

Govender, V. (2011, August 20). Senior Manager, MTN. (E. Mankum, Interviewer)

Haque, A., Rahman, S., & Rahman, M. (2010). Factors Determinants the Choice of Mobile
ServiceProviders: Structural Equation Modeling Approach on Bangladeshi Consumers. Business
and Economics Research Journal , 1 (3), 17-34.

Hartmank, K. B., & Spiro, R. L. (2005). Recapturing store image in customer based-store equity:
a construct conceptualization. 58 (8), 1112-1120.

Hays, L. W. (1981). Statistics (3rd Edition ed.). New York: Holt, Rinehart, and Winston.

Heggde, G., & Kumar, S. (2011). Strategies for Sustainable Channel Relations in Mobile
Telecom Sector. Journal of Economics and Behavioral Studies , 2 (1), 7-18.

Horvath, D., & Sajtos, L. (2002). "How do mobile communicate? The role of product design in
product related consumer response: The case of mobile telephone.". Advanced in Consumer
Research , 29, 237-238.

http://www.raosoft.com. (2011, April 26). Retrieved April 26, 2011, from

http://www.raosoft.com: http://www.raosoft.com/samplesize.html

Ireland, R. D., Hoskisson, R. E., & Michael, H. A. (2011). The Managment of Strategy: Concepts
and Cases (nineth ed.). Canada: South-Western Cengage learning.

Jaworski, B. J., Stathakopoulos, V., & Krishnan, S. H. (1993). Control combinations in marketing:
Conceptual frameworks and emphirical evidence. Journal of marketing , 57-69.

Johnson, W. C., & Sirikit, A. (2002). Service Quality in the Thai Telecommunication Industry: A
Tool for Achieving a Sustainable Competitive Advantage. Management Decision , 40 (7), 693-
701.

Kargin, B., Basoglu, N., & Daim, T. (2009). Factors affecting the adoption of mobile services.
International Journal of Services Sciences , 2, 29-52.

110
Kim, K. M., Park, C. M., & Jeong, H. D. (2004). The effects of customer satisfaction and
switching barrier on customer loyalty in Korean mobile telecommunication services.
Electronics and Telecommunications Research Institute , 305-348.

Kim, Y., & Lee, J. (2010). Relationship between corporate image and customer loyalty in mobile
communications service markets. Africa Journal of Business Management , 4 (18), 4035-4041.

Kollmann, T. (2000). The price/acceptance function: Perspectives of a pricing policy in

European telecommunication markets. European Journal of Innovation Management , 3 (1), 7-
14.

Kotler, P., & Keller, K. L. (2009). A Framework for Marketing Managment, Fourth Edition
(Fourth ed.). New Jesery: Pearson Prentice Hall.

Kuo, Y., & Yen, S. (2009). Towards an understanding of the behavioral intention to use 3G
mobile value-added services. Computers in Human Behavior , 25, 103-110.

Kwon, W. S., & Lennon, S. J. (2009). What induces online loyalty? Online versus offline brand
images. Journal of business research , 62 (5), 557-564.

Lambart, D., Stock, J., & Ellram, L. M. (1998). Fundamentals of Logistics Management. New
York: Irwin and McGraw-Hill.

Lambrecht, A., & Skiera, B. (2007). Paying Too Much and Being Happy About It: Existences,
Causes, and Consequences of Tariff-Choice Biases. Journal of Marketing Research , 43 (2), 212-
223.

Lee, R., Murphy, J., & Dickringer, A. (2006). Investigating the Loyalty of Prepaid and Postpaid
Mobile Customers. International Conference on Business and Information. Singapore.

Leenheer, J., Bijmolt, T. H., Heerde, H. J., & Smidts, A. (2002). Do Loyalty Programs Enhance
Behavioral Loyalty? An Empirical Analysis Accounting for Program Design and Competitive
Effects. Center , 65.

Leisen, B., & Vance, C. (2001). Cross-national Assessment of Service Quality in the
Telecommunication Industry: Evidence from the USA and Germany. Managing Service Quality ,
11 (5), 307-317.

Levine, M. S. (1977). Canonical analysis and factor comparision (Vol. 6). (S. Publications, Ed.)
1977, Thousand Oaks, California.

Livne, G., Simpson, A., & Talmor, E. (2011). Do Customer Acquisition Cost, Retention and Usage
Matter to Firm Performance and Valuation? Journal of Business Finance & Accounting , 38 (2),
35-55.

McGraw, K. O., & Wong, S. P. (1996). Forming inferences about some intraclass correlation
coefficients. Psychological Methods , 1 (1), 30-46.

111
Min, D., & Wan, L. (2009). Switching factors of mobile customers in Korea. Journal of Service
Science , 1 (1), 105-120.

MTN. (2011, July 21). Retrieved July 21, 2011, from MTN: http://www.mtn.co.za

MTN Group Limited. (2010). Integrated Business Report for the year ended 31 December 2009.
Johannesburg: MTN Group Limited.

MTN Group Limited. (2011). Integrated Business Report for the year ended 31 December 2010.
Johannesburg: MTN Group Limited.

Munnukka, J. (2008). Customers’ purchase intentions as a reflection of price perception.

Journal of Product & Brand Management , 17 (3), 188–196.

Nalebuff, B. (2004). Bundling as an entry barrier. Quarterly Journal of Economics , 119 (1), 159-
187.

Noordhoff, C., Pauwels, P., & Odekerken, S. G. (2006). The effect of customer card programs -
A comparative study in Singapore and the Netherlands. International Journal of Service
Industry Managment , 15 (4), 351-364.

O’Leary-Kelly, S., & Vokurka, R. (1988). The empirical assessment of construct validity. Journal
of Operational Management , 16 (4), 387-405.

Omotayo, O., & Joachim, A. A. (2008). Customer service in the retention of mobile phone Users
in Nigeria. African Journal of Business Managment , 2 (2), 026-031.

Paulrajan, R., & Rajkumar, H. (2011). Service Quality and Customers Preference of Cellular
Mobile Service Providers. Journal of Technology Management & Innovation , 6 (1), 38-45.

Radosavljevik, D., Putten, P. v., & Larsen, K. K. (2010). The Impact of Experimental Setup in
Prepaid Churn Prediction for Mobile Telecommunications:What to Predict, for Whom and
Does the CustomerExperience Matter? Transactions on Machine Learning and Data Mining , 3
(2), 80-99.

Raghuram, I., Jedidi, K., & Kohli, R. (2008). A Conjoint Approach to Multipart Pricing. Journal of
Marketing Research , 45 (10), 195-210.

Rahman, S., Haque, A., & Ahmad, M. I. (2010). Exploring influencing factors for the selection of
mobile phone service providers: A structural equational modeling (SEM) approach on
Malaysian consumers. African Journal of Business Management , 4 (10), 2885-2898.

Rautio, T., Anttila, M., & Tuominen, M. (2007). Bundling of information goods: A value driver
for new mobile TV services. International Journal of Revenue Management , 1 (1), 45-64.

Rowley, J. (1998). Promotion and Marketing Communications in the Information Marketplace.

Lib. Rev , 47 (8), 383-387.

Seo, D., Ranganathan, C., & Babad, Y. (2008). Two-level model of customer retention in the US
mobile telecommunications service market. Telecommunications Policy , 32 (3-4), 182-196.

112
Soderlund, M., & Rosengren, S. (2008). Revisiting the smiling service Worker and customer
satisfaction. International Journal of Service Industry Management , 19 (5), 552-574.

SPSS Inc. (2010). Help IBM SPSS Statistics. Help IBM SPSS Statistics . New York, New York,
United States.

Steers, S. (2007). Loyalty Programs Boost Retention, Profits. National Underwriter / Property &
Casualty Risk & Benefits Management , 111 (19), 20-24.

Stremersch, S., & Tellis, J. G. (2002). Strategic bundling of products and prices: A new synthesis
for marketing. Journal of Marketing , 66 (1), 55-72.

Tucker, S. (1966). Pricing for Higher Profit: Criteria, Method Applications. New York: McGraw-
Hill.

Turki, A. M. (2010). Customer Service Retention A Behavioural Perspective of the UK Mobile

Market. Doctoral thesis, Durham University.

Vodacom. (2011, July 21). Retrieved July 21, 2011, from Vodacom: http://www.vodcom.co.za

Vodacom Group Limited. (2010). Preliminary results for year end 31 March 2010. South Africa:
Vodacom Group Limited.

Wal, R. W., Pampallis, A., & Bond, C. (2002). Service quality in a cellular telecommunications
company: a South African experience. Managing Service Quality , 12 (5), 323 - 335.

Wood, L. M. (2004). Dimensions Of Brand Purchasing Behavior: Consumers in The 18-24 Age
Group. Journal of Consumer Behavior , 4 (1), 9-24.

World Celluar Information Services. (2011, April 26). www.wcisplus.com. Retrieved April 26,
2011, from www.wcisplus.com:
http://www.wcisplus.com:80/wcisplus/report/columns/subscription types-measures-
[dates]/rows/regions-countries-[operators]-technologies-networks/annual/mar 2010-mar
2011;regions=africa;countries=south+africa;technologies=-phs;network statii=all+network+st

Yi, Y., & Jeon, H. (2003). Effects of loyalty programmes on Value Perception, Program Loyalty
and Brand Loyalty. Journal of the Academy of Marketing Science , 31 (3), 229-240.

Zeithaml, V. A. (2000). Service quality, profitability and the economic worth of customers:
what we know and what we need to learn. Journal of the Academy of Marketing Science , 28
(1), 67-85.

113
 10. Appendixes

10.1 Churn reduction financial model

Table 10.1: Churn reduction financial model using Vodacom information for financial years
2009 and 2010

Base Case Scenario Case Formula Remarks

Year 2009 2010 2010

Subscribers (thousands) 23 679 j 21 765 21 765 a Reported by Vodacom

Subscriber net adds
(thousands) -1 914 -1 914 b Reported by Vodacom
Subscribers gross adds
(thousands) 8 523 8 523 c Reported by Vodacom

Churn subscribers (thousands) 10 437 9 490 jxd

Churn percentage 44% 40% d Reduced churn by 4%

Subscribers less churn
(thousands) 13 242 14 189 j-(j x d)
Average revenue
generating subscribers
Average revenue generating increased due to reduced
subscribers (thousands) 18 026 l 18 539 f churn
Reported ARPU by
ARPU R 70 R 70 k Vodacom
f x k x 12
(months) x Revenue increases by
Revenue R 15 141 525 000 R 15 572 482 800 1000 2.8%

Average prepaid acquisition

cost R 45 e

Acquisition cost R 42 622 200 (f - l) x e

EBITDA increases by 0.3%
k + ((f - l) x due to reducing churn by
EBITDA R 14 790 000 000 k 14 832 622 200 e) 4%
Source: (Vodacom Group Limited, 2010), (Govender, 2011)

114
10.2 Questionnaire

Good day, my name is Ethesen Mankum. I am a MBA student from GIBS, the business school of the University of
Pretoria, student number 10665392. In completion of my master thesis, I am conducting research to understand the
usage of your prepaid mobile service. Your input into the research will be greatly appreciated.

If you approve, I will ask you a series of questions, which should take about 15 minutes of your time. The completion of
the questionnaire will be conducted telephonically and if at any time, you wish to cancel the interview you may do so.

Please note that your personal details will not be shared during the reporting of the research results, and information
shared will be collectively reported upon. Thus, your responses to the questionnaire will be kept in strict confidence.

In addition, the completion of your questionnaire, storing of results and reporting of results will be done in strict
compliance of the rules and regulations of GIBS.

Yes

No

1 . Is your mobile phone prepaid or contract?

Prepaid

Contract

If contract close interview

Q2A. Have you recently (past 12 months) switched from mobile service provider?

a) Yes

b) No

If no skip to Q3

Q2B. Why did you switch

Reasons for switching

a) Voice Network Performance (e.g. Call quality, Drop calls, Geographical

coverage)
b) Data Network Performance (e.g. Internet browsing – Facebook or receiving
e-mails)
c) Price of Prepaid charges (e.g. The price of voice calls, the price of data & the
price of sms)
d) Different type of tariffs plans (e.g. per second, per minute, flat tariffs, peak and
off peak tariffs)
e) Product availability (e.g. sim purchases)

f) Airtime recharge options (e.g. Pin less, vouchers, USSD, ATM, Internet)

115
 g) Promotions

h) Loyalty programmes

i) Customer service quality

j) Prepaid product bundles (e.g. sms bundles, data bundles, music bundles)

k) Access to content based services (e.g. Video broadcast, Multimedia services,

News, Gaming, Internet browsing, Music download)
l) Access to location-based (e.g. GPS navigation, Weather alerts, Traffic
updates, Restaurant reviews/info, Locating nearby convenience services)
m) Brand perceptions/reputation of service provider

n) Store image

o) Market share

p) Innovative services / products

Q3. How would you rate the overall importance when choosing a mobile service provider regarding the following
statements by using a scale from 0 to 10, where 0 means „Totally Unimportant‟ and 10 means „Extremely Important‟,
you may also choose any number in between:

0 1 2 3 4 5 6 7 8 9 10 D/K N/
A
Totally Extremely
Unimportant Important

a) Network quality (e.g.

Call quality, Drop
calls, Geographical
coverage, Internet
browsing – Facebook
or receiving
e-mails)
b) Mobile tariffs (e.g.
The price of voice
calls, the price of data
& the price of
sms)
c) Different types of
Mobile tariffs plans
(e.g. per second, per
minute, flat tariffs,
peak and off peak
tariffs)
d) Prepaid airtime and
starter pack
availability
e) Different airtime
recharge options

116
 0 1 2 3 4 5 6 7 8 9 10 D/K N/
A
Totally Extremely
Unimportant Important

(e.g. Pin less,

vouchers, USSD,
ATM, Internet)
f) Promotions

g) Loyalty
programmes
h) Customer service
quality
i) Different types of
prepaid product
bundles (e.g. sms
bundles, data
bundles, music
bundles)
j) Access to mobile
content and location
services (e.g. Video
broadcast, Multimedia
services, News,
Gaming, Internet
browsing, Music
download. GPS
navigation, Weather
alerts, Traffic updates,
Restaurant
reviews/info, Locating
nearby convenience
services))
k) Brand
perceptions/reputati
on of service
provider

Q4. Out of the following statements regarding the “Network quality” which of the following statement is most important to
you, second most important to you and third most important to you when choosing a mobile service provider:

117
 1

Most Important,

Second Most Important,

Third Most Important,

4 Do not know
a) Call quality

b) Drop calls

c) Geographical coverage

Q5. Out of the following statements regarding the “Price of prepaid” which of the following statement is most important
to you, second most important to you and third most important to you when choosing a mobile service provider:

Most Important,

Second Most Important,

Third Most Important,

4 Do not know

a) The price of voice calls

b) The price of data calls

c) The price of sms

Q6. Out of the following statements regarding the “Tariff Plans” which of the following statement is most important to
you, second most important to you and third most important to you when choosing a mobile service provider:

118
 1

Most Important,

. Second Most Important,

Third Most Important,

4 Do not know

a) Per second tariff plan

b) Per minute tariff plan

c) Peak and off peak tariff plan

d) Flat tariff plan

Q7. Out of the following statements regarding the “Prepaid Product Bundles” which of the following statement is most
important to you, second most important to you and third most important to you when choosing a mobile service
provider:

Most Important,

Second Most Important,

Third Most Important,

4 Do not know

a) Prepaid service with SIM only

offering
b) Prepaid service with SIM and phone
offering
c) Prepaid service with SIM and free
airtime offering
d) Prepaid service with SIM bundle
offering
e) Prepaid service with data bundle
offering
f) Prepaid service with music
downloads

Q8. Out of the following statements regarding the “Mobile Content/Services” which of the following statement is most
important to you, second most important to you and third most important to you when choosing a mobile service
provider:

119
 1

Most Important,

Second Most Important,

Third Most Important,

4 Do not know

a) Access to mobile video

b) Access to mobile banking

c) Access to music downloads

d) Access to ringtones downloads

e) Access to instant messaging (such

as blackberry, mix it)

Q9. Out of the following statements regarding the “Product availability” which of the following statement is most
important to you, second most important to you and third most important to you when choosing a mobile service
provider:

Most Important,

Second Most Important,

Third Most Important,

4 Do not know

a) Product available through food retail

store

b) Product available through clothing

retail store

c) Product available through local

neighbourhood stores/ corner stores

e) Product available through mobile

service provider branded stores

120
D1. What is your personal monthly income before tax and deductions?

a) Less than R 1 000 Continue

b) Between R1 000 and R3 000 Continue

c) Between R3 001 and R5 000 Continue

d) Between R5 001 and R10 000 Continue

e) Between R10 001 and R20 000 Continue

f) More than R20 000 Continue

g) Refused Continue

D2. Gender of respondent

c) Male

d) Female

D3. How old are you?

a) 16-24 years old

b) 25-29 years old

c) 30-34 years old

d) 35-39 years old

e) 40-44 years old

f) 45-49 years old

g) 50-54 years old

h) 55-59 years old

i) 60 or older

D4. Which Province do you live in?

a) Gauteng

b) Eastern Cape

c) Western Cape

d) Northern Cape

e) Limpopo

f) Mpumalanga

g) North West

h) Free State

i) Kwazulu-Natal

121