Presented by: Seema Praliya, Pooja Singh, Sunitha Ashok Menon, Shashank Raj Dixit

Definition
Correlation research is a form of analysis in which you correlate one variable with another to determine if there is a relationship between them.

Goals of Correlation Research:

Helps us to understand related events, conditions, and behaviors.
e.g. Is there a relationship between blood pressure and cholesterol level?

To make predictions of how one variable might predict

another
e.g. Can height of be used to predict the self esteem of people?

Nature of correlation Research

The relationship between two or more factors are explored, without manipulating any of the variables being investigated.

It seeks to describe the existing relationship between variables.

The degree of relationship is usually expressed using the correlation coefficient. Note: With correlation research we can only show the magnitude or
degree of relationship or association between variables, not the cause of relationship

Types of Correlation
Positive Correlation: means that high score in one variable is associated with high score in another variable Negative Correlation: means high score in one variable is associated with low score in another variable.

Correlation Research process

Variables to be studied are identified Questions and/ or hypotheses are stated A sample is selected (a minimum of 30 is needed) Data are collected

Surveys Observation Techniques: Mapping Sorting Archives Tactics to analyse the data Multivariate analysis Multiple regression Factor analysis Multidimensional Scaling

Correlations are calculated Results are reported

General characteristics
Focus on naturally occurring Pattern
Clarify the relationship among a complex set of real world variables that vary with the circumstances or setting being studied and are also likely to affect the dynamics of socio-physical interaction. By variables we mean the range of characteristics( of physical features, of people, of activities, or of meanings) e.g: ways in which two variables A & B may be casually related

B A B

A causes B ( either directly or indirectly)

B causes A ( either directly or indirectly)

C causes both A & B

The measurement of specific variable:

Data in correlation research is quantified and tallied through the following: 1. Categorical measurements: the variable of interest is sorted into discrete categories, based on verbal and nominal terms. (e.g. walk, car, bus, metro, train etc are mode of transportation used to get to work.) 2. 3. Ordinal Scales: Here variables are ordered on some basis (a range from lesser to greater) Interval ratio Scales: Measurement of exact intervals; any established and consistent unit of measurement (dollars, feet, degrees, etc). These should not be used for abstract data (attitudes, feelings, etc.)

The use of statistics to clarify patterns of relationship:

The researcher develops either : Descriptive statics : simply presents or describes, important relationships among variables. Correlational statistics: measures the strength and direction of relationship between two or more variables. note: all calculated correlation coefficients fall within a rage of 1.00(negative correlation) to +1.00 (positive correlation). Correlation coefficient close to 0 indicates no consistent linear relationship b/w variables, i.e. relationship cannot be graphed as a straight line.

CORRELATIONAL RESEARCH STRATEGY

Relationship studies- how particular events are related
Causal-comparative research- intermediate position between predictive orientation of relationship studies and the focus on causality that characterizes experimental research.

RELATIONSHIP STUDIES
1)

RELATIONSHIP STUDIES: Although all correlational studies seek to discribe relationships among key variables, relationship studies focus more specifically on the nature and predictive power of such relationships. Eg:Oscar Newmans study of public housing in New York city which produce the theory of DIFENSIBLE SPACES

Records of 169 public housing projects managed by the New York City Housing Authority were examined b) Research team conducted an exhaustive investigation of the complex relationship between user demographics and the physical variables of the housing /site design, and incidence of crime c) The emense research made it possible to determine exactly where is the most dangerous areas of buildings are, as well as to compare crime rates of different building types and project layout
a)

Not only this theory of DEFENSIBLE SPACE define a relationship between environmental variable and behavioral consequences, but also offers a predictive capacity that can be articulated as design guidelines, specifically low income housing that incoporates real and symbolic barriers, defined area of influence, and opportunities for surveillance

CAUSAL-COMPARATIVE RESEARCH
2) CAUSAL- COMPARATIVE STUDIES:

Causal- comparative studies stake out an intermediate position between the predictive orientation of relationship studies and focus on casuality that characterizes experimental research In causal comparative studies, the researcher select comparable groups of people or comparable physical environments and then collects data on a variety of relevant variables. The causal- comparative design can only ascribe cases in a provisional or hypothetical way. This is because causal- comparative research relies on natural occurring variable, as do all correlational studies To make causal-comparative design persuasive, the researcher must establish the essential comparability of the examples studied. Unfortunately, there are often many obstacles to establishing the equivalence of the examples/groups in naturally occurring circumstances .

TACTICS: COLLECTING DATA

Survey
Observation Mapping

Sorting
Archives

SURVEY METHODS
Sampling Stratified random sampling Sample is a part of parent population. Sample is any part of a defined population, which is selected or to be selected for questioning, observation and from which information can be obtained and statistical inferences are drawn or prediction made about entire population. A statistical sample is thus a miniature picture or cross section of the entire group or aggregate from which the sample is taken.

POPULATION
In a statistical population, every member of a group

possesses the same basic and define characteristics but varying in amount of quality from one member to another, for eg. Income tax payers in India.

METHODS OF DESIGNING SAMPLES

Deliberate or purposive or judgment or direct sampling. Random/ lottery method Stratified random sampling Systematic sampling Cluster sampling Quota sampling- sample fixed, allotted numbers Note- systematic sampling is one in which every k^th item is selected in a list representing a population or a stratum. The number k is called sampling intervals. K= N/n where k= sampling intervals N= universe size, n= sample size

DATA
Secondary sources

(published and unpublished literature)

Data sources Primary sources

(one to one)

SURVEY AT GRASS-ROOT LEVEL

Reconnaissance survey
House listing survey Pilot/pre testing survey

Household survey
a. Comprehensive household survey b. Household survey at the selected samples

TOOLS FOR DATA COLLECTION

Schedules
Questionnaire

Survey questionnaire
Survey is perhaps the most frequently employed tool. It allows the researcher to cover an extensive amount of information across a large number of people in a limited amount of time Questionnaire are drafted through the following procedure. 1. Determine main topic to be covered.(orient ques. towards that.) 2. Chose either open ended or close ended format(as well as categorical, ordinal and interval ratio)

Closed and multiple choice question Q. how often do your parents ask you about your home work? Never 1 1-2 times a week 2 3-4 times a week 3 Nearly every day 4

The main advantage of closed questions are: The respondent is restricted to a finite (and therefore more manageable) set of responses, They are easy and quick to answer, They have response categories that are easy to code, they permit the inclusion of more variables in a research study because the format enables the respondent to answer more questions in the same time required to answer fewer openended questions. Open-ended or free-response questions are not followed by any choices
[Link] are your favourite TV programmes? (Please specify their titles) [Link] do you like most about school? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..........................................................

The main advantages of open-ended questions are:

They

allow respondents to express their ideas spontaneously in their own language, They are less likely to suggest or guide the answer than closed questions because they are free from the format effects associated with closed questions, and They can add new information when there is very little existing information available about a topic.

The main disadvantages of open-ended questions are:

1. They may be difficult to answer and even more difficult to analyze, 2. Require time and efforts 3. Require some degree of writing ability. 4. Handwriting can be illegible.

Phrase question clearly(use short sentences, avoid making multiple query in single question) Q. Do you think it is good idea for children to study geography and
history in primary school?

Avoid framing question in negative Avoid hypothetical question?

Q. Would you use the resources in your class if it were available?

Avoid using ambiguous wording, employ nonthreatening language Use a logical sequence of topic(start with non challenging and interesting issue but dont place imp. In the end. Use appealing and simple graphics Introduce the survey properly(provide a reason for the survey, provide a description of what a researcher is expected to do. state provisions for keeping individuals responses confidential

Observation
Most important aspect figuring out what to look for Needs to be systematic As it makes possible to establish justified design objectives and simplifies complexities Essential to devise a thorough observation program To get the direct rather than random approach and not to lose its representativeness Critical self-questioning is very essential A great advantage even a simple numbers count can provide a detailed and powerful view of the human ecology of a particular setting

Observation
Range of variables that might be observed Demographic characteristics
study of a day in life of the ledge at Seagram; no. of people in

time interval(charted number count in specific time intervals)

Specific activities Whytes study of urban plazas (used time-lapse camera to observe behaviour of activity in pedestrian areas) User reactions Mark Fredericksonss study of design juries, with special focus on possibility of gender or minority basis(time-frequency measures and content process categories)
There can be other variables too.

Mapping
To get clear and simplified view of distinct features of an

area
To determine the imageable features of a place To help making design guidelines for that place

Sketching and drafting of maps of areas under

observations
Sketching through visual or verbal experience, better when

difference is nil Derivations of certain features that re-occur in all sample maps; establishing distinguished points of the area Eg., Kevin Lynchs study image of the city derivations of five major urban features; path, edge, node, landmark and district

Mapping
Marking out or pointing the different physical features

over an area layout

Eg., Anne Lusks study of greenway bicycle paths as a basis for

formulating design guidelines Making aesthetically pleasing features on the way prominent on paper by notifying them Used different stickers as notifications Measurement of distances between two distinct points by odometer Was able to determine major destination points along greenway occur every two miles

HABITUAL USER MAPS - Lakefront trail, Chicago

TRACE OVERLAY MAP - Lakefront trail, Chicago

Sorting
Highly effective both in research and practice

situations Sorting of cards having either words or pictures on them Types

Directed sort : sorting in categories(eg. , 5- or 7-point

scale from highly preferred to least preferred) Free sort : sorting in whatever categories suit to respondent (eg., buildings according to functions, styles, etc.) Multiple free sort : sorting can be done as many times as possible

Sorting
Visual exercise for simplifying complexities and

determining difference in priorities and thereby approaches A creative foundation for dialogue between client and architect Ease of expression of motives

Archives
Use of existing database and records

Eg., Oscar Newmans study for defensible spaces He took help of records of New York Housing Authority and also police force databank for demographic data; and thereby correlating them with physical properties of various housing projects Was able to determine exactly where the most dangerous areas of buildings are, and comparison of crime rates in different building types and project layouts

Tactics : Multivariate analysis

Experienced researcher employ such complex statistical method to analyze the collected data to come to the conclusion

Multiple Regression Factor Analysis Multidimensional Scaling

Multiple Regression
This is more complex data analysis, used to describe the

strength and direction of variable relationship . The mathematical equations indicate the amount of variance contributed by each variables.

Linear Multiple Regression Analysis

Well known statistical technique for fitting mathematical

relationship b/w independent and dependent variables.

Assumption 1. All the variables are independent of each other. 2. All the variables are normally distributed. 3. All the variables the continuous. 4. A linear relationship exist between the dependent and 5.

independent variable. The influence of the independent variable is additive. It is easy adaptability to computer programming when handling voluminous data

Factor Analysis
Depends on interval or ratio data. Instead of using key variables to predict the outcomes of other variables, factor analysis aims to articulate an overall pattern. Variables that share similar patterns are grouped into clusters known as factors.

What is factor analysis ?

Factor analysis is a general name denoting a class of Procedures primarily used for data reduction and summarization. Variables are not classified as either dependent or independent. Instead, the whole set of interdependent relationships among variables is examined in order to define a set of common dimensions called Factors.

Purpose of Factor Analysis

To identify underlying dimensions called Factors, that

explain the correlations among a set of variables. e.g. lifestyle statements may be used to measure the psychographic profile of consumers.
To identify a new, smaller set of uncorrelated variables to

replace the original set of correlated variables for subsequent analysis such as Regression or Discriminant Analysis. e.g. psychographic factors may be used as independent variables to explain the difference between loyal and non loyal customers.

Multidimensional Scaling
Offers more flexibility
Uses both nominal data as well as interval or ration data Outcome of the analysis is represented graphically. Its goal is to reveal an underlying pattern or structure

among the variables analyzed, similar to factor analysis, but allowing a greater degree of interpretative flexibility. Results in a graphical plot that locates relationship between all variables spatially. Two variables plotted in close proximity means they represent similar patterns of observations and vice versa.

CORRELATIONAL RESEARCH
Applications of the correlational strategy Prediction

Predictor variable (e.g. GPA) Criterion variable to be predicted (e.g. grad school success)

Reliability and validity Calculating correlations in test-retest reliability and concurrent validity
Evaluating theories E.g. Heredity and IQ correlations between twins

STRENGTHS AND WEAKNESSES

strengths nonintrusive natural behaviors Can clarify relationships among two or more naturally occurring variables. Well suited to studying the breadth of a setting or a phenomenon Can establish predictive relationships high external validity weaknesses cannot assess causality (e.g. TV and aggressive behavior) Research can not control levels or degrees of variables directionality problem low internal validity; less suited to exploring the setting or phenomenon in depth.