# Structural Equation Modeling (SEM) Marketing Research Help

In the department store project, SEM can be used to develop a model that explains store preference in terms of respondents’ evaluations of the store on the eight factors of the choice criteria, The endogenous construct will be store preference, measured by 10 indicator variables; each indicator will be preference for one of the 10 stores considered in the study. The exogenous construct will be store evaluation, which is a second-order construct with evaluation of stores on the eight factors of the choice criteria as the eight first-order factors.

Thus the eight first-order factors are evalualions of stores on quality of merchandise, variety and assortment of merchandise. returns snd adjustment policy, service of store personrel, prices, convenience of location. layout of store, and credit and billing policies, Each first-order factor is measured by 10 indicator variables, with each indicator representing the evaluation of one of the 10 stores on a particular aspect of the choice criteria.

Project Activities

Download the SPSS or SAS data file Sears Data 17 from the Web site for this book. This file contains the evaluation of Sears on the eight factors of the choice criteria (quality, variety and assortment. return policy, service of store personnel, fair prices, convenience of location, store layout. and credit and billing policies). the preference for Sears, importance attached to eight factors of the choice criteria, and agreement with the 21 lifestyle statements, The measurement of these variables is described in Chapter 1, The remaining variables have not been included to keep the number of variables below 50.

1.Formulate the measurement model. Draw a path diagram,
2. Estimate the measurement model on the data contained in Sears Data 17.
3. Asses the reliability and validity of the measurement model.
4. Formulate the structural model. Draw a path diagram.
5. Estimate the structural model on the data contained in Sears Data 17.
6. Asses the validity of the structural model.
7. Interpret the results, What are your conclusions and what recommendations will you make to Sears management?

Path Analysis

Path analysis’ (PA) can be viewed as a special case of structural equation modeling (SEM). We could think of PA as SEM with only single indicators for each of the variables in the causal model. In other words. path analysis is SEM with a structural model, but no measurement model. Other terms used to refer to path analysis include causal modeling. analysis of covariance structures; and latent variable models.

Path analysis may also be viewed as an extension of the regression model. The PA model is depicted in a rectangle-and-arrow figure in which single-headed arrows indicate causation. A regression is done for each variable in the model as a dependent on others. which to model indicates are causes. The regression w~jlbts estimated by the model are compared with the observed correlation matrix for the v~es. and a goodness-of-fit statistic is calculated. Path an~ysis calculates the strength of each relationship using only a correlatiori or covariance matrix as input. We illustrate PA with an example.

The Path to International Success

A recent study sought to explain the internationalization of professional service firms. The underlying theoretical framework was based on attribution theory. The structural model posited that three causal constructs or factors (i.e., uniqueness of offering, financial resources, and competitive pricing) influenced cognitive social consequences (i.e., expectation of success) and the resulting behavioral consequence (i.e., international success). This model is shown in the accompanying diagram. The data were obtained from a sample of 152 U.S.-based professional service firms via a mail survey. First, a measurement model was specified and estimated using CFA. The results showed acceptable composite reliability and convergent and discriminant validity. Then, the structural model (see diagram) was estimated and found to be valid. The results of the structural model provided support for all the four hypotheses (H I to H4), thereby providing support for the attribution theory framework. Several managerial implications were drawn. For example, the

Attribution Model of Internationalization of Professional Service Firms

influence of cognitive psychological consequences on behavioral consequences means that managers who expect international success tend to be successful. Hence, professional service firms wishing to make greater inroads into the international marketplace may wish to search out and promote such managers

Personal Values, Ethical Ideology, and Ethical Beliefs

The relationships among an individual’s personal values, ethical ideology, and ethical beliefs were investigated using structural equation modeling. The data were collected in an Internet survey, and a final sample of 609 completed surveys was obtained. First, a measurement model was tested for key validity dimensions. Then. the hypothesized causal relationships were examined in several path models. The results indicated that individual differences in values directly and indirectly (through idealism) influence the Judgment of ethically questionable consumer practices. The findings not only contributed to the theoretical understanding of ethical decision making but also had managerial implications. For example, understanding what underlies people’s unethical attitudes and behavior would enable retailers to positively influence appropriate behavior of shoppers by appealing to personal values. This should result in a reduction of unethical behaviors such as shoplifting

Statistical Software

Several computer programs are available for SEM. The most widely used program is LISREL, an acronym for LInear Structural RElations. LISREL is a flexible program that can be used in a range of SEM applications normally encountered in marketing research. AMOS (Analysis of Moment Structures) is another program that is gaining in popularity because it is user-friendly and is available as an added module to SPSS. SAS offers the CALIS program. Another widely available program is EQS, an abbreviation for equations. LISREL, AMOS, and EQS are available with point-and-click interface. Mplus is another program that is also used for SEM. For applications commonly encountered in marketing research, these programs should produce similar results. Hence, the selection of a specific computer program for SEM should be made based on availability and the user’s preference

LlSREl

Because LISREL is a popular program, we give detailed instructions for conducting SEM using this software for the first-order factor model application that we have discussed. The data file can be downloaded from the Web site for this book

!CFA-Lisrel
Raw Data from File CH 22 TAM.PSF
Latent Variables: PU PE INT
Relationships:
PU -> PU \ PUl PU3
PE -> PEl PE2 PE3
INT -> INTI INT2 INT3
Method of Estimation = Maximum Likelihood
Number of Decimals = 2
path diagram
Lisrel Output: SE TV Ml EF FS RS SS SC AD=offlT= 1300
End of Program
!SEM-Lisrel
Raw Data from File CH 22 TAM.PSF
Latent Variables: PU PE INT
Relationships:
PUI = I*PU
PU2 = PU
PU3 = PU
PEl = 1* PE
PE2 = PE
PE3 = PE
INTI = l*INT
INT2 = INT
INT3 = INT
INT = PUPE

Method of Estimation = Maximum Likelihood

Number of Decimals = 2
path diagram
Lisrel Output: SE TV MI EF FS RS SS SC AD = off IT= 1300
End of Program

SPSS Windows

AMOS provides you with powerful and easy-to-use structural equation modeling (SEM) software. Using AMOS, you specify, estimate, assess, and present your model in an intuitive path diagram to show hypothesized relationships among variables

There are some distinctions of note when comparing the various SEM software packages. For example, AMOS varies from’ LISm bt how it approaches exogenous variables. LISREL (correctly) asks researchers to make if distinction between exogenous and endogenous variables prior to the testing of a model. AMOS simply treats any variables that do not have paths going to them as exogenous, As a result, LISREL by default allows all exogenous variables to freely Bovary with one another, while researchers using AMOS have to be proactive in establishing this set of relationships,

We give detailed instructions for conducting SEM using AMOS for the first-order factor model application that we have discussed, The data file can be downloaded from the Web site for this book

‘CFA
Sub Main 0
Dim sem As New ArnosEngine
sem.Standardized
sem.TexiOutput
sem.BeginGroup “CH 22 TAM.sav”
sem.Strueture “PUl = PU + (1) EI”
sem.Structure “PU2 = PU + (1) E2”
sem.Structure “PU3 = PU + (1) E3”
sem.Structure “PEl = PE + (1) E4”
sem.Strscture “PE2 = PE + (1) ES”
sem.54ructure “PE3 = PE + (1) E6”
sem.Structure “INTI = INT + (1) E7”
sem.Structure “INT2 = INT + (I) ES”
sem.Structure “INT3 = INT + (1) E9”
sem.Structure “PU (1)”
sem.Si:ructure “PE (I)”
sem.Structure “!NT (1)”
sem.Structure “PU – PE”
sem.Snuctere “PU -INT’
sem.Snucture “PE – INT’
End Sub
Sub Main 0
Dim sem As New AmosEngine
sem.Standardized
sem.TextOutput
sem.BeginGroup “CH 22 TAM.sav”
sem.Structure “PUI = (I) PU + (I) EI”
sern.Structure “PU2 = PU + (I) E2”
sem.Structure “PU3 = PU + (I) E3”
sem.Structure “PEl = (I) PE + (1) £.4”
sem.Structure “PE2 = PE + (I) £5”
sem.Structure “PE3 = PE + (I) E6”
sem.Structure “INTI = (I) INT + (I) E7”
sem.Structure “INTI = INT + (I) E8”
sem.Structure “INT3 = INT + (I) E9”
sem.Structure “lNT = PU + PE + (I) ERR”
End Sub

SAS Learning Edition

The instructions given here and in all the data analysis chapters (14 to 22) will worle with the SAS Learning Edition as well as with the SAS Enterprise Guide. SAS offers the CALIS procedure. We give detailed instructions for conducting SEM usinC the SAS Learning Edition for the first-order factor model application that we have discussed. The data file can be downloaded from the Web site for this book.

1. Select File>New>Code. (Note that SEM is not available as an SAS Learning Edition task and that SAS code needs 10 be submitted in order to run this analysis.)
2. Type ‘Ie following SAS statements into the Code window and enter the physical location of ca-aset in the Iibname statement

The Situation

The soft drink market is one of the most popular but slowly growing markets in the United States. From 2002-2008, the average annual growth rate was estimated to be around 4.7 percent. Carbonated beverages hold the largest share of the market in soft drinks, followed by bottled water and then juices. Diet drinks as a whole have grown in popularity

When Diet Coke was first introduced in 1982. consumers saw advertisements using celebrities to promote the new product. As time went on. a more body-conscious approach was taken in Diet Coke’s advertising image. Consumers began to see advertisements of a construction worker being ogled by women working in an office, promoting the image of a healthier alternative to non-diet soft drinks. As competition has increased and Pepsi launched their advertising campaign of being the choice of the younger and “hipper” consumer, the Coca-Cola Company must reevaluate its current advertising campaign. Muhtar Kent is wondering whether a more product-focused advertising campaign that stresses the factors influencing consumers’ choice of diet soft drink brands would be more effective:

The Marketing Research Decision

1. What theoretical framework can be used to explain consumers’ altitudes, preferences, and consumption of diet soft drink brands?
2. What data should be collected and how should they be analyzed in determining what factors drive consumers’ consumption of soft drink brands? Can SEM be used? If so how?

The Marketing Management Decision

1. Should Muhtar Kent change the advertising for Diet Coke to focus more on the factors influencing consumers’ choice of diet soft drink brands?
2. Discuss how the marketing management decision action that you recommend to Muhtar Kent is influenced by the data analysis that you suggested earlier and by the likely findings.

Posted on December 2, 2015 in Structural Equation Modeling and Path Analysis