Special Session Summary Theory Testing By the Use of Structural Equation Modeling

Citation:

Lutz Hildebrandt (1998) ,"Special Session Summary Theory Testing By the Use of Structural Equation Modeling", in E - European Advances in Consumer Research Volume 3, eds. Basil G. Englis and Anna Olofsson, Provo, UT : Association for Consumer Research, Pages: 125.

European Advances in Consumer Research Volume 3, 1998      Page 125

SPECIAL SESSION SUMMARY

THEORY TESTING BY THE USE OF STRUCTURAL EQUATION MODELING

Lutz Hildebrandt, Humboldt University of Berlin, Germany

During the last 20 years the methodology Structural Equation Modeling" and the parallel diffusion of computer software, e.g. LISREL (J÷reskog/S÷rbom 1993), EQS (Bentler 1996) and PLS (Lohm÷ller 1984) has become the most powerful tool in consumer research to test theories about all facets of consumer behavior. Also it has generally become accepted in all other areas of empirical marketing research. The approach of structural equation modeling has the advantage that beside the possibility to test (resp. assess the fit of model) a theory represented by a so called causal model it enables the researchers to evaluate the measurements and the substantive theory either by an overall validationBin nomological validityBor by a stepwise assessment of different kinds of validity criteria. In addition to that, the methodology is directly related to the concepts of theory of science like the theory of two languages by Hempel and Carnap, distinguishing explicitely between empirical data and theoretical concepts in the model structure.

Although the approach is widely accepted for theory testing especially in behavioral science, the discussion about the general appropriateness of structural equation modeling for theory testing is still going on. In general the critique is related to the assumptions of the structural equation model and the way to interpret the tested models, relations and coefficients. During the past years most of the critique however has led to improvements in the existing software tools and also to new developments in the way to analyze data, to collect data and to specify and test new types of models. The aim of the special session therefore was to provide an insight into the recent developments in structural equation modeling. The session also gave some advices how to test behavioral theories and how the results of testing a structural model can be interpreted. The contributions also presented some new ways to apply structural equation modeling analyzing time series or panel data.

Three presentations have contributed to the special session.

"RECENT DEVELOPMENTS IN STRUCTURAL EQUATION MODELING"

by Christian Homburg / Lutz Hildebrandt (WHU Koblenz / HU Berlin)

 

"THE USE OF STRUCTURAL EQUATION MODELS IN MARKETING AND CONSUMER RESEARCH"

by Hans Baumgartner / Christian Homburg (KU Leuven / WHU Koblenz)

 

"THE USE OF LISREL TO ANALYZE PANEL-DATA WITH UNOBSERVABLE EFFECTS"

by Lutz Hildebrandt / Dirk Annacker (both HU Berlin)

The first presentation gave a review of recent developments in structural equation computer programs, possible types of relaxation of model assumptions, the ne implemented tests in existing programs and an evaluation of the features and power of different software programs. The presentation also discussed recent and expected improvements of the existing programs LISREL, EQS and PLS. Special attention was given to the graphical capabilities of the different programs and a comparison was given to newer releases of other programs which are part of larger statistical software packages such as SAS or STATISTICA. Another part of the presentation discussed the possibilities to use the Internet for the estimation and testing of structural equation models in economics and behavioral science.

The second presentation gave an insight into the existing status of structural equation modeling in consumer and marketing research. The paper reviewed prior applications of structural equation modeling in four major marketing journals (the Journal of Marketing Research, Journal of Marketing, International Journal of Research in Marketing, and the Journal of Consumer Research) between 1977 and 1994. The research documented and characterized the number of applications over time, discussed important methodological issues related to structural equation modeling and assessed the quality of previous applications in terms of three aspects: issues related to the initial specification of theoretical models of interest; issues related to data screening prior to model estimation and testing; and issues related to the estimation and testing of theoretical models on empirical data. On the basis of the findings, problem areas were identified, and avenues for improvements suggested.

The third presentation focused on the application of structural equation models to the analysis of time series or panel data. In economic and behavioral research these data are now widely available from consumer panels, household panels or retail panels. In single source data bases, scanner panel data are combined with behavioral and attitudinal data and also data from marketing and retailing. The study presented a panel model based on time series data from strategic marketing. The specified model tested a substantive theory and controls for systematic unobservable effects. Systematic unobservable effects sometimes may be the true forces to cause the change of a target variable, e.g. buying behavior. The focus of the study was to discover the effects of marketing instruments on the increase of market share and profitability, simultaneously controlling for other impact variables and unobservable variables like marketing skills and reputation. Different hypotheses from marketing theory about the lag structure of the marketing and advertising effects were tested against each other, using the LISREL-approach. On the basis of business unit data a fixed effects serial correlation model was estimated. It was shown that there exists no lagged effect of marketing on market share, but there might be a lagged effect of marketing on profitability.

----------------------------------------