An introduction to

structural equation modeling

Hans Baumgartner
Smeal College of Business
The Pennsylvania State University

Structural equation modeling

Structural equation modeling (SEM)

also known as latent variable modeling, latent variable
path analysis, (means and) covariance (or moment)
structure analysis, causal modeling, etc.;

a technique for investigating relationships between latent
(unobserved) variables or constructs that are measured
by (multiple) manifest (observed) variables or indicators;

can be thought of as a combination of regression analysis
(including systems of simultaneous equations) and factor
analysis;

special cases are confirmatory factor analysis and
manifest variable path analysis;

in recent years, SEM has been extended in many ways;

Structural equation modeling

SEM (contd)

two primary advantages of SEM:

SEM makes it possible to study complex patterns of

relationships among the constructs in a conceptual
model in an integrative fashion;
the measurement of unobserved (latent) variables by
observed fallible indicators can be modeled explicitly,
and the effect of measurement error (both random
and systematic) on structural relationships can be
taken into account;

Structural equation modeling

Explaining the usage of coupons

for grocery shopping
(cf. Bagozzi, Baumgartner, and Yi 1992)
Rewards
1

Inconveniences
2

Encumbrances
3

Attitudes
1

Intentions
2

Coupon usage
3

Structural equation modeling

Goal-directed emotions
(Bagozzi, Baumgartner, and Pieters 1998)
61

dieting
volitions

11

31

dieting
behaviors

53

positive
anticipated
emotions

65

12

positive
goal-outcome
emotions

goal
attainment

21

negative
anticipated
emotions

22

75
54
exercising
volitions

42

exercising
behaviors

72

negative
goal-outcome
emotions

Structural equation modeling

The relationship between observed

measurements and constructs of interest

Brand
loyalty

The observed single-item brand

loyalty score is a perfect
measure of true brand loyalty.

All of the variability in observed
scores is trait (substantive)
variance.

Measure of
brand loyalty

(e.g., I think of myself as a

brand-loyal consumer.)

Structural equation modeling

The relationship between observed

measurements and constructs of interest (contd)
Brand
loyalty

Measure of
brand loyalty

The observed brand loyalty

score is contaminated by
random measurement error.

If only a single measure is
available, random
measurement error cannot
be taken into account.

Structural equation modeling

The relationship between observed

measurements and constructs of interest (contd)

The total variability of observed
scores consists of both trait
(substantive) variance and
random error variance.

This results in unreliability of

measurement and the
attenuation of observed
correlations.
E1

T1

T2

E2

Structural equation modeling

The relationship between observed

measurements and constructs of interest (contd)
Solution: Use multiple indicators to measure the focal
construct, in which case we can assess reliability and
correct for attenuation.
Brand
loyalty

Brand loyalty
measure 1

Brand loyalty
measure 2

Brand loyalty
measure 3

Structural equation modeling

The relationship between observed

measurements and constructs of interest (contd)

The total variability of observed
scores consists of trait
(substantive), random error, and
systematic error (method)
variance.

T
M
E

T1

This is likely to confound the

assessment of reliability and
relationships with other
constructs.

T2
E2

E1
M1

M2

It also complicates the

comparison of means.

Structural equation modeling

A comprehensive model of measurement error

yijt = ijt + ijt jt + ijt + ijt

systematic
error

yijt
jt

ijt
ijt
ijt
ijt

random
error

a persons observed score on the ith measure

of construct j at time t
a persons unobserved score for construct j at
time t
systematic error score
random error score
coefficient (factor loading) relating yijt to jt
intercept term (additive bias)

Structural equation modeling

Attitude toward using coupons

(measured at two points in time)

AAt1

x11

x21

AAt2

x31

x41

x12

x22

x32

x42

Structural equation modeling

Attitude toward using coupons

(measured at two points in time)

AAt1

x11

x21

AAt2

x31

x41

x12

x22

x32

x42

Structural equation modeling

Factor correlations
Original correlation

Corrected correlation

Exploratory factor
analysis (PFA with
Promax rotation)

.75

n.a.

Confirmatory factor
analysis

.90

.90

Correlation of
unweighted linear
composites at t1, t2

.82

.819
= .91
.882 .911

.63

.626
= .91
.654 .719

Average correlation of
individual t1, t2 measures

Structural equation modeling

Adoption of managerial innovations

(Bagozzi and Phillips 1982)

Forecasting

Training

y1

y2

y3

y4

Financial
analysis

Accounting

y5

y6

y7

y8

Structural equation modeling

Adoption of managerial innovations (contd)

Forecasting

Training

y1

y2

y3

CEO

y4

Financial
analysis

Accounting

y5

y6

y7

Subordinate

y8

Structural equation modeling

Variance partitioning
Trait

Method

Error

Training-CEO (y1)

.78

.07

.15

Training-Sub (y2)

.25

.23

.53

Forecasting-CEO (y3)

.90

.09

.00

Forecasting-Sub (y4)

.25

.51

.23

Accounting-CEO (y5)

.68

.14

.17

Accounting-Sub (y6)

.93

.04

.03

Financial analysis-CEO (y7)

.62

.38

.00

Financial analysis-Sub (y8)

.74

.10

.15

Structural equation modeling

Graphical specification of a
(congeneric) measurement model
21
1

11 21

31 41

52 62

72 82

x1

x2

x3

x4

x5

x6

x7

x8

55

66

77

88

11

22

33

44

Structural equation modeling

Need for Touch (NFT) scale

(Peck and Childers 2003)
Instrumental touch:

I place more trust in products that can
be touched before purchase.

I feel more comfortable purchasing a
product after physically examining it.

If I can't touch a product in the store, I
am reluctant to purchase the product.

I feel more confident making a
purchase after touching a product.

The only way to make sure a product
is worth buying is to actually touch it.

There are many products that I would
only buy if I could handle them before
purchase.

Autotelic touch:

When walking through stores, I
can't help touching all kinds of
products.

Touching products can be fun.

When browsing in stores, it is
important for me to handle all kinds
of products.

I like to touch products even if I
have no intention of buying them.

When browsing in stores, I like to
touch lots of products.

I find myself touching all kinds of
products in stores.

Structural equation modeling

Graphical specification of an integrated

measurement/latent variable model
11

11

x1

22

x2

x21

x3

31

44

x4

55

x5

66

x6

77

x7

33

22
12

21

42
32

22

11
21

33

11

1
x

63

x73

41

y21

13

y31

y1

y2

y3

y4

11

22

33

44

33

32

2
1

y62

y5

y6

55

66

3
1

y7

Structural equation modeling

Measurement model specification issues:

Reflective vs. formative measurement models
focal
construct

focal
construct

focal
construct

focal
construct

Structural equation modeling

Measurement model specification issues:

Number of indicators per construct

in principle, more indicators are better, but there are
practical limits;

question of how explicitly single-item measures are
modeled:

total aggregation model

partial aggregation model (item parcels)
total disaggregation model
focal
construct

focal
construct

focal
construct

Structural equation modeling

Latent variable model specification issues

recursive vs. nonrecursive models
1
1
2
2
3

specification of plausible alternative models

problem of equivalent models

Structural equation modeling

The problem of equivalent models

3
1

3
1

4
2

4
2

1
4

4
2

Structural equation modeling

Model identification

question whether the parameters in the model are
uniquely determined so that the conclusions
derived from the analysis arent arbitrary;

a necessary condition is that the number of
parameters to be estimated doesnt exceed the
number of unique elements in the (co)variance
matrix of the observed variables;

for relatively simple models, rules of identification
are available; for more complex models, empirical
heuristics may have to be used;

Structural equation modeling

Model estimation

Covariance-based SEM:

estimate the model parameters in such a way that the

covariance matrix implied by the estimated
parameters is as close as possible to the sample
covariance matrix;

e.g., for a factor model

21
1

x = +
= +

11

21

31

41

x1

x2

x3

x4

11

22

33

52

62

72

x5

x6

x7

x8

44

55

66

77

88

Variance-based SEM (PLS):

estimate the parameters so as to maximize the

explained variance in the dependent variables;

82

Structural equation modeling

Model testing

Global fit measures:

2 goodness of fit test

alternative fit indices

Local fit measures:

parameter estimates, standard errors and z-values

measurement model:

reliability and discriminant validity

latent variable model:

R2 for each structural equation

Model modification:

modification indices and EPCs

residuals

Structural equation modeling

Testing the overidentifying restrictions

of a model
1
11
1
2

12

13
3

2
21

3
32

Structural equation modeling

Testing the overidentifying restrictions

of a model (contd)
1
11
1
2

12

2
21

3
32

13
3

There are 21 distinct elements in the covariance matrix of the 6 latent variables,
we estimate 14 parameters, so there are 7 overidentifying restrictions.

Structural equation modeling

Testing the overidentifying restrictions

of a model (contd)
11

11

x1

22

x2

x21

x3

31

44

x4

55

x5

66

x6

77

x7

33

22
12

21

42
32

22

11
21

33

11

1
x

63

x73

41

y21

13

y31

y1

y2

y3

y4

11

22

33

44

33

32

2
1

y62

y5

y6

55

66

3
1

y7

Structural equation modeling

Testing the overidentifying restrictions

of a model (contd)

x1

x2

x3

x4

11 22

33 44

x5

x6

55 66

x7

7
77

y1

y2

y3

y4

11 22

y5

y6

y7

55 66

33 44

There are 105 distinct elements in the covariance matrix of the 14 observed
variables, we estimate 42 parameters, so there are 63 overidentifying restrictions.

Structural equation modeling

Testing the overidentifying restrictions

of a model (contd)
11

11

x1

22

x2

x21

x3

31

44

x4

55

x5

66

x6

77

x7

33

22
12

21

42
32

22

11
21

33

11

1
x

63

x73

41

y21

13

y31

y1

y2

y3

y4

11

22

33

44

33

32

2
1

y62

y5

y6

55

66

3
1

y7

There are 105 distinct elements in the covariance matrix of the 14 observed variables, we estimate 35 parameters, so there are 70 (63+7) overidentifying restrictions.

Structural equation modeling

Problems with the 2 test

it is not a robust test;

it is based on the accept-support logic of testing:

a model is more likely to get support when the sample

size is small and power is low (even though it is an
asymptotic test);
since most models are unlikely to be literally true in
the population, in large samples the model is likely to
be rejected;

thus, many alternative fit indices have been

suggested;

Structural equation modeling

Classification of overall fit indices

Overall fit indices

Stand-alone fit indices

2 test and
variations
minimum fit
function 2
(C1)
normal theory
WLS 2 (C2)
S-B scaled 2
(C3)
2 corrected
for nonnormality
(C4)
2/df
minimum fit
function f
Scaled LR

Incremental fit indices

Noncentrality- Information
based
theory-based Others
measures
measures

Type I indices

Type II indices

NCP

AIC

(S)RMR

NFI

[2 or f]

IFI

Rescaled
NCP (t)

SBC

GFI

RFI

[2/df]

TLI

CIC

PGFI

ECVI

AGFI

CFI
[2-df]

Gamma
hat

TLI
[(2-df)/df]

RMSEA
MC

CN

Structural equation modeling

Goodness of fit statistics for the coupon data:
Degrees of Freedom = 70
Minimum Fit Function Chi-Square = 93.63 (P = 0.031)
Normal Theory Weighted Least Squares Chi-Square = 92.60 (P = 0.037)
Estimated Non-centrality Parameter (NCP) = 22.60
90 Percent Confidence Interval for NCP = (1.60 ; 51.68)
Minimum Fit Function Value = 0.38
Population Discrepancy Function Value (F0) = 0.091
90 Percent Confidence Interval for F0 = (0.0064 ; 0.21)
Root Mean Square Error of Approximation (RMSEA) = 0.036
90 Percent Confidence Interval for RMSEA = (0.0096 ; 0.054)
P-Value for Test of Close Fit (RMSEA < 0.05) = 0.89
Expected Cross-Validation Index (ECVI) = 0.65
90 Percent Confidence Interval for ECVI = (0.57 ; 0.77)
ECVI for Saturated Model = 0.84
ECVI for Independence Model = 12.16
Chi-Square for Independence Model with 91 Degrees of Freedom = 2999.42
Independence AIC = 3027.42
Model AIC = 162.60
Saturated AIC = 210.00
Independence CAIC = 3090.72
Model CAIC = 320.85
Saturated CAIC = 684.75
Normed Fit Index (NFI) = 0.97
Non-Normed Fit Index (NNFI) = 0.99
Parsimony Normed Fit Index (PNFI) = 0.75
Comparative Fit Index (CFI) = 0.99
Incremental Fit Index (IFI) = 0.99
Relative Fit Index (RFI) = 0.96
Critical N (CN) = 268.08

Root Mean Square Residual (RMR) = 0.13

Standardized RMR = 0.049
Goodness of Fit Index (GFI) = 0.95
Adjusted Goodness of Fit Index (AGFI) = 0.92
Parsimony Goodness of Fit Index (PGFI) = 0.63

Structural equation modeling

Model testing

Global fit measures:

2 goodness of fit test

alternative fit indices

Local fit measures:

parameter estimates, standard errors and z-values

measurement model:

reliability and discriminant validity

latent variable model:

R2 for each structural equation

Model modification:

modification indices and EPCs

residuals

Structural equation modeling

Estimation results for the coupon model

11

11

x1

Incon

22

x2

x21

x3

x4

Rewards

55

x5

66

x6

77

x7

12

21

Att

32

BI

42
32

33

22

31

44

22

11
21

33

11

1
x

63

Encumb

x73

41

y21

13

y31

y1

y2

y3

y4

11

22

33

44

33

y62

y5

y6

55

66

y7

Structural equation modeling

Measurement model results for coupon data

Construct

parameter

parameter
estimate

standardized
parameter
estimate

z-value

individualitem
reliability

Inconveniences

composite
reliability
(average variance
extracted)
.88 (.78)

x11
x21
11
22

1.00
0.98
0.56
0.61

0.89
0.88
0.21
0.23

-11.32
3.32
3.71

0.79
0.77
---

Rewards

.76 (.61)
x32
x42
33
44

1.00
0.82
0.45
0.96

0.86
0.70
0.25
0.52

-6.89
2.55
6.63

0.75
0.48
---

Encumbrances

.70 (.45)
x53
x63
x73
55
66
77

1.00
1.73
1.48
2.78
1.85
1.92

0.49
0.77
0.71
0.76
0.41
0.50

-6.30
6.30
9.97
5.49
6.87

0.24
0.59
0.50
----

Structural equation modeling

Measurement model results for coupon data (contd)

Construct

parameter

parameter
estimate

standardized
parameter
estimate

z-value

individualitem
Reliability

composite
reliability
(average variance
extracted)
.88 (.66)

Attitudes
y11
y21
y31
y41
11
22
33
44

1.00
1.04
0.85
1.10
0.68
0.44
0.76
0.59

0.80
0.86
0.73
0.84
0.37
0.26
0.47
0.29

-14.97
12.14
14.58
9.06
7.70
9.82
8.20

0.63
0.74
0.53
0.71
-----

Intentions

.91 (.84)
y42
y52
44
55

1.00
1.09
0.97
0.25

0.87
0.97
0.25
0.07

-18.91
7.04
1.95

0.75
0.93
---

y63
66

1.00
0.00

1.00
0.00

---

1.00
--

Behavior

Structural equation modeling

Discriminant validity
Correlation Matrix of ETA and KSI

aact
bi
bh
inconv
rewards
encumbr

(.81)
aact
-------(.81) 1.00
(.92) 0.70
(---) 0.40
(.88) -0.44
(.78) 0.52
(.67) -0.35

(.92)
bi
--------

(--)
bh
--------

(.88)
inconv
--------

(.78)
rewards
--------

(.67)
encumbr
--------

1.00
0.58
-0.31
0.36
-0.25

1.00
-0.18
0.21
-0.14

1.00
-0.10
0.49

1.00
-0.27

1.00

Note: The latent variable correlations are corrected for attenuation.

Structural equation modeling

Latent variable model results for coupon data

Structural Equations
AACT =

- 0.28*INCONV + 0.44*REWARDS - 0.050*ENCUMBR, Errorvar.= 0.69 , R2 = 0.42

(0.058)
(0.081)
(0.097)
(0.11)
-4.77
5.42
-0.51
6.52

BI = 1.10*AACT, Errorvar.= 1.53 , R2 = 0.48

(0.11)
(0.20)
10.04
7.73
BH = 0.49*BI, Errorvar.= 1.41 , R2 = 0.34
(0.049)
(0.13)
10.10
10.78

Structural equation modeling

Model testing

Global fit measures:

2 goodness of fit test

alternative fit indices

Local fit measures:

parameter estimates, standard errors and z-values

measurement model:

reliability and discriminant validity

latent variable model:

R2 for each structural equation

Model modification:

modification indices and EPCs

residuals

Structural equation modeling

Modification indices

a modification index (MI) refers to the predicted
decrease of the 2 statistic when a fixed parameter
is freely estimated or an equality constraint is
relaxed;

associated with each MI is an expected parameter
change (EPC), which shows the predicted value of
the freely estimated parameter;

data-based model modifications have to be done
carefully;

Structural equation modeling

Modification indices for coupon data

Modification Indices for BETA

AACT
BI
BH

AACT
-------- - 2.34

BI
-------11.05
- - -

BH
-------1.52
2.34
- -

Modification Indices for GAMMA

AACT
BI
BH

INCONV
-------- 5.57
1.61

REWARDS
-------- 3.07
12.67

ENCUMBR
-------- 5.15
2.78

Structural equation modeling

Two-step approach to model modification

(Anderson and Gerbing 1988)

specify a measurement model in which the latent
variable model is saturated and purify the
measurement model;

once the measurement model is in place, attend to
the latent variable model;

Structural equation modeling

Saturated latent variable model

for the coupon data

x1

x2

x3

x4

11 22

33 44

x5

x6

55 66

x7

7
77

y1

y2

y3

y4

11 22

y5

y6

55 66

33 44
2(63)=62.90

y7

Structural equation modeling

Modified latent variable model

11

11

x1

22

x2

x21

x3

31

44

x4

55

x5

66

x6

77

x7

33

22
12

21

42
32

22

11
21

33

11

1
x

63

x73

41

y21

13

y31

y1

y2

y3

y4

11

22

33

44

33

32

2
1

y62

y5

y6

55

66

3
1

y7

2(70)=92.60
vs.
2
(69)=79.21

Structural equation modeling

Multi-sample analysis:
Known population heterogeneity

SEMs can be specified for several populations
simultaneously;

this also allows the estimation of mean structures;

multi-sample models are particularly useful for
assessing measurement invariance (e.g., in crosscultural research);

mediation, moderation, moderated mediation and
mediated moderation can be assessed in a
straightforward fashion;

Structural equation modeling

A factor model with a mean structure

21
11

22
1

2
0

3
2

1
4

6
6

7
7

8
8

x1

x2

x3

x4

x5

x6

x7

x8

66

11

22

33

44

55

77

88

Structural equation modeling

Assessing measurement invariance:

Configural invariance
G1:

x1

x2

G2:

x3

x2

x5

x4

x6

x1

x3

x4

x7

x8

x7

x8

x5

x6

Structural equation modeling

Assessing measurement invariance:

Metric invariance
G1:

x1

x2

G2:

x3

x2

x5

x4

x6

x1

x3

x4

x7

x8

x7

x8

x5

x6

Structural equation modeling

Assessing measurement invariance:

Scalar invariance
G1:

x1

x2

G2:

x3

x2

x5

x4

x6

x1

x3

x4

x7

x8

x7

x8

x5

x6

Structural equation modeling

Linking the types of invariance required

to the research objective
(Steenkamp and Baumgartner 1998)
Configural
invariance

Metric
invariance

Exploring the basic

structure of the construct
cross-nationally

Examining structural
relationships with other
constructs crossnationally

Conducting crossnational comparisons of

means

Scalar
invariance

Structural equation modeling

Satisfaction with Life in the US and AUT:

Final partial scalar invariance model
Factor loadings

Item intercepts

AUT

US

AUT

US

ls1

.92

.92

-.03

-.03

ls2

.90

.90

.12

.12

ls3

1.00

1.00

0.00

0.00

ls4

.80

.80

.72

.72

ls5

1.10

.83

-1.00

.06

Structural equation modeling

Satisfaction with Life in the US and AUT:

Final partial scalar invariance model
Factor loadings

Item intercepts

AUT

US

AUT

US

ls1

.92

.92

-.03

-.03

ls2

.90

.90

.12

.12

ls3

1.00

1.00

0.00

0.00

ls4

.80

.80

.72

.72

ls5

1.10

.83

-1.00

.06

Latent means

AUT: 3.91

US: 3.26

Structural equation modeling

Goal-directed emotions: Results

0.31 (7.4)

.14 (3.6)

positive
anticipated
emotions

dieting
volitions

.20 (7.4)
.36 (6.8)d

dieting
behaviors

.16 (4.4)

.07 (.6)
.54 (4.9)b

.61 (7.7)a,b
.29 (3.1)c,d

positive
goal-outcome
emotions

goal
achievement

negative
anticipated
emotions
.07 (1.9)

.16 (4.0)

.08 (.9)
.56 (3.7)a

.18 (2.3)a,c negative

.46 (8.7)b,d goal-outcome
emotions

exercising .24 (8.7) exercising

volitions
behaviors

0.11 (3.4) .33 (5.3)b

a
b
c
d

men wanting to lose weight

women wanting to lose weight
men wanting to maintain their weight
women wanting to maintain their weight

2(110)=150.51
RMSEA=.061
CFI=.94
TLI=.92

Structural equation modeling

Mediation and moderation

a mediator Me is a variable that accounts for the relation
between a predictor A and a criterion C (i.e., it channels
at least some of the total effect of A on C);
A

Me

a moderator Mo is a variable that affects the direction

and/or strength of the relation between a predictor A and
a criterion C;
Mo
A

Structural equation modeling

Mediation and moderation (contd)

0.31 (7.4)

.14 (3.6)

positive
anticipated
emotions

dieting
volitions

.20 (7.4)
.36 (6.8)d

dieting
behaviors

.16 (4.4)

.07 (.6)
.54 (4.9)b

.61 (7.7)a,b
.29 (3.1)c,d

positive
goal-outcome
emotions

goal
achievement

negative
anticipated
emotions
.07 (1.9)

.16 (4.0)

.08 (.9)
.56 (3.7)a

.18 (2.3)a,c negative

.46 (8.7)b,d goal-outcome
emotions

exercising .24 (8.7) exercising

volitions
behaviors

0.11 (3.4) .33 (5.3)b

a
b
c
d

men wanting to lose weight

women wanting to lose weight
men wanting to maintain their weight
women wanting to maintain their weight

2(110)=150.51
RMSEA=.061
CFI=.94
TLI=.92

Structural equation modeling

What are the effects of positive anticipated emotions on

goal achievement for people who desire to lose weight,
and do these effects differ by gender?

Males

Direct effect

Indirect effect

Total effect

---

.019*

.019*

via dieting

-.002

via exercising

.021*
s.

Females

---

s.

.017*

via dieting

.014*

via exercising

.003

.017*

Structural equation modeling

Hierarchical models
12

11

Construct B

Note: Covariance between u0j and

u1j not shown for simplicity.

Construct A

r2

Structural equation modeling

Latent curve models

21
2

Intercept

Slope

mean 1
variance 11

mean 2
variance 22

1
1

-5
-3

-1 1

1 1

y1

y2

y3

y4

y5

y6

Structural equation modeling

Mixture modeling:
Unobserved population heterogeneity
r

Construct B

Construct A

ri2

Ai 2

LC
i = 1 or 2
Note: The parameters are the mixing probabilities.

Structural equation modeling

Background readings

Kline, Rex B. (2011), Principles and practice of structural
equation modeling, 3rd ed., New York: The Guilford
Press.

Bollen, Kenneth A. (1989), Structural equations with
latent variables, New York: Wiley.

Byrne, Barbara M. (1998), Structural Equation Modeling
with LISREL, PRELIS, and SIMPLIS: Basic Concepts,
Applications, and Programming, Mahwah, NJ: Erlbaum.

Structural equation modeling

Computer programs for SEM

LISREL 9.1 (Jreskog & Srbom)

http://www.ssicentral.com/lisrel/index.html

Mplus Version 7 (Muthen)

http://www.statmodel.com/

EQS 6.2 (Bentler)

http://www.mvsoft.com/eqs60.htm

PROC CALIS in SAS, AMOS in SPSS, special packages

in R, Stata, etc.

Structural equation modeling

Criteria for distinguishing between

reflective and formative indicator models

Are the indicators manifestations of the underlying
construct or defining characteristics of it?

Are the indicators conceptually interchangeable?

Are the indicators expected to covary?

Are all of the indicators expected to have the same
antecedents and/or consequences?

Based on MacKenzie, Podsakoff and Jarvis,

JAP 2005, pp. 710-730.

Structural equation modeling

Consumer Behavior
Attitudes

Aad as a mediator of advertising effectiveness:

Four structural specifications (MacKenzie et al. 1986)
Affect transfer hypothesis
Cad

Aad

Cb

Ab

BI

Reciprocal mediation hypothesis

Cad

Aad

Cb

Ab

BI

Dual mediation hypothesis

Cad

Aad

Cb

Ab

BI

Independent influences hypothesis

Cad

Aad

Cb

Ab

BI

Structural equation modeling

Reliability for congeneric measures

individual-item reliability (squared correlation between a construct

j and one of its indicators xi):
ii = ij2var(j)/[ ij2 var(j) + ii]

composite reliability (squared correlation between a construct and

an unweighted composite of its indicators x = x1 + x2 + ... + xK):
c = (ij)2 var(j)/[ (ij)2 var(j) + ii]
average variance extracted (proportion of the total variance in all
indicators of a construct accounted for by the construct; see Fornell
and Larcker 1981):
ave = (ij2) var(j)/[ (ij2) var(j) + ii]

Structural equation modeling

SIMPLEX specification
Title
A general structural equation model (explaining coupon usage)
Observed Variables
id be1 be2 be3 be4 be5 be6 be7 aa1t1 aa2t1 aa3t1 aa4t1 bi1 bi2 bh1
Raw Data from File=d:\m554\eden2\sem.dat
Latent Variables
INCONV REWARDS ENCUMBR AACT BI BH
Sample Size 250
Relationships
be1 = 1*INCONV
be2 = INCONV
be3 = 1*REWARDS
be4 = REWARDS
be5 = 1*ENCUMBR
be6 = ENCUMBR
be7 = ENCUMBR
aa1t1 = 1*AACT
aa2t1 = AACT
aa3t1 = AACT
aa4t1 = AACT
bi1 = 1*BI
bi2 = BI
bh1 = 1*BH
AACT = INCONV REWARDS ENCUMBR
BI = AACT
BH = BI
Set the Error Variance of bh1 to zero
Options sc rs mi wp
Path Diagram
End of Problem

Modeling random and systematic

measurement error (Baumgartner and Steenkamp 2006)
Stable
(item-)subset
error

SSE

Transient
(item-)subset
error

1TSE

2TSE

3TSE

4TSE

y11 y21 y31 y41 y51

y12 y22 y32 y42 y52

y13 y23 y33 y43 y53

y14 y24 y34 y44 y54

1SIE

SIE

3SIE

4SIE

5SIE

Stable
item-specific
error