Differential Effects of Omitting Formative Indicators: A Comparison of Techniques

Research examining the formative specification of constructs has highlighted the need for researchers to capture all relevant causes of a construct of interest. However, the consequences of omitting a formative indicator have not been thoroughly examined. Given that one of the commonly employed techniques for modeling formatively specified constructs, Partial Least Squares, implicitly assumes that all relevant causes of a construct have been modeled, the consequences of omitting one of those are of prime importance. In this research we compare latent variable and PLS techniques on this issue based on theoretical arguments and results from Monte Carlo simulations. In particular, we focus on the presence or absence of estimation bias in the relationships between formative indicators and the formatively specified construct, and between the latter and other constructs in the research model. Our results highlight differences in how these two techniques cope with the omission of formative indicators, and discuss why those differences occur.

[1]  Detmar W. Straub,et al.  Validating Instruments in MIS Research , 1989, MIS Q..

[2]  Ryad Titah,et al.  Information System Use - Related Activity: An Expanded Behavioral Conceptualization of Individual-Level Information System Use , 2007, Inf. Syst. Res..

[3]  Detmar W. Straub,et al.  Specifying Formative Constructs in Information Systems Research , 2007, MIS Q..

[4]  George M. Marakas,et al.  Revisiting Bias Due to Construct Misspecification: Different Results from Considering Coefficients in Standardized Form , 2012, MIS Q..

[5]  Thomas R. Carretta,et al.  The Role of Measurement Error in Familiar Statistics , 2006 .

[6]  R. Mauro.  Understanding L.O.V.E. (left out variables error): A method for estimating the effects of omitted variables. , 1990 .

[7]  A. Satorra,et al.  Power of the likelihood ratio test in covariance structure analysis , 1985 .

[8]  Lutz Hildebrandt,et al.  Formative Measurement Models in Covariance Structure Analysis: Specification and Identification , 2006 .

[9]  S. Riegg,et al.  Causal Inference and Omitted Variable Bias in Financial Aid Research: Assessing Solutions , 2008 .

[10]  T. Raykov.  Estimation of Composite Reliability for Congeneric Measures , 1997 .

[11]  Ronald T. Cenfetelli,et al.  Interpretation of Formative Measurement in Information Systems Research , 2009, MIS Q..

[12]  Phil Wood.  Confirmatory Factor Analysis for Applied Research , 2008 .

[13]  C. Fornell,et al.  Evaluating structural equation models with unobservable variables and measurement error. , 1981 .

[14]  Jacob Cohen,et al.  Problems in the Measurement of Latent Variables in Structural Equations Causal Models , 1990 .

[15]  C. Schriesheim.  Causal Analysis: Assumptions, Models, and Data , 1982 .

[16]  Richard D. Johnson,et al.  The Evolving Nature of the Computer Self-Efficacy Construct: An Empirical Investigation of Measurement Construction, Validity, Reliability and Stability Over Time , 2007, J. Assoc. Inf. Syst..

[17]  H. Costner.  Theory, Deduction, and Rules of Correspondence , 1969, American Journal of Sociology.

[18]  Cheryl Burke Jarvis,et al.  The problem of measurement model misspecification in behavioral and organizational research and some recommended solutions. , 2005, The Journal of applied psychology.

[19]  Kenneth A. Bollen,et al.  Causal Indicator Models: Identification, Estimation, and Testing , 2009 .

[20]  A. Boomsma,et al.  The robustness of LISREL modeling revisted. , 2001 .

[21]  R. Lennox,et al.  Conventional wisdom on measurement: A structural equation perspective. , 1991 .

[22]  R. P. McDonald,et al.  Structural Equations with Latent Variables , 1989 .

[23]  Adamantios Diamantopoulos,et al.  Advancing formative measurement models , 2008 .

[24]  K. Bollen.  Multiple indicators: Internal consistency or no necessary relationship? , 1984 .

[25]  Ritu Agarwal,et al.  Through a Glass Darkly: Information Technology Design, Identity Verification, and Knowledge Contribution in Online Communities , 2007, Inf. Syst. Res..

[26]  Lawrence R. James,et al.  The unmeasured variables problem in path analysis. , 1980 .

[27]  K. Jöreskog,et al.  Intraclass Reliability Estimates: Testing Structural Assumptions , 1974 .

[28]  R. P. McDonald,et al.  Path Analysis with Composite Variables. , 1996, Multivariate behavioral research.

[29]  David F. Larcker,et al.  Structural Equation Models with Unobservable Variables and Measurement Error: Algebra and Statistics: , 1981 .

[30]  K. Bollen.  Interpretational confounding is due to misspecification, not to type of indicator: comment on Howell, Breivik, and Wilcox (2007). , 2007, Psychological methods.

[31]  E. F. Jackson,et al.  Multiple Indicators in Survey Research , 1962, American Journal of Sociology.

[32]  Cheryl Burke Jarvis,et al.  A Critical Review of Construct Indicators and Measurement Model Misspecification in Marketing and Consumer Research , 2003 .

[33]  Nick Lee,et al.  Improper use of endogenous formative variables , 2013 .

[34]  A. Tomarken,et al.  Potential problems with "well fitting" models. , 2003, Journal of abnormal psychology.

[35]  Adamantios Diamantopoulos,et al.  The error term in formative measurement models: interpretation and modeling implications , 2006 .

[36]  Paul W. Holland,et al.  The sensitivity of linear regression coefficients' confidence limits to the omission of a confounder , 2009, 0905.3463.

[37]  H. Winklhofer,et al.  Index Construction with Formative Indicators: An Alternative to Scale Development , 2001 .

[38]  T. Dijkstra.  Latent Variables and Indices: Herman Wold’s Basic Design and Partial Least Squares , 2010 .