Abstract
Econometrics textbooks generally conclude that in regression, because the calculation of path estimate variances includes avariance inflation factor (VIF) that reflects correlations between âindependentâ constructs, multicollinearity should not causefalse positives except in extreme cases. However, textbook treatments of multicollinearity assume perfect measurement ârare in behavioral research. VIF is based on apparent correlations between constructs -- always less than actual correlationswhen measurement error exists. A brief review of recent articles in the MIS Quarterly suggests that the conditions forexcessive false positives are present in published research. In this paper we show (analytically and with a series of MonteCarlo simulations) that multicollinearity combined with measurement error presents greater than expected dangers from falsepositives in IS research when regression or PLS is used. Suggestions for how to address this situation are offered.
Keywords
MulticollinearityVariance inflation factorFalse positive paradoxVariance (accounting)EconometricsStatisticsMonte Carlo methodFalse positives and false negativesComputer scienceRegression analysisMathematicsEconomicsAccounting
Affiliated Institutions
Related Publications
Multicollinearity and misleading statistical results
Multicollinearity represents a high degree of linear intercorrelation between explanatory variables in a multiple regression model and leads to incorrect results of regression a...
Role of range and precision of the independent variable in regression of data
Abstract Regression of the experimental data of one independent variable, y vs . a linear combination of functions of an independent variable of the form y = Σβ j f j (x) is con...
Multidimensionality and Structural Coefficient Bias in Structural Equation Modeling
In this study, the authors consider several indices to indicate whether multidimensional data are “unidimensional enough” to fit with a unidimensional measurement model, especia...
Evaluating Structural Equation Models with Unobservable Variables and Measurement Error
The statistical tests used in the analysis of structural equation models with unobservable variables and measurement error are examined. A drawback of the commonly applied chi s...
Factor Analysis as a Tool for Survey Analysis
Factor analysis is particularly suitable to extract few factors from the large number of related variables to a more manageable number, prior to using them in other analysis suc...
Publication Info
- Year
- 2011
- Type
- article
- Citations
- 10
- Access
- Closed
External Links
Citation Metrics
10
OpenAlex
Cite This
Dale L. Goodhue,
William W. Lewis,
Ronald L. Thompson
(2011).
A Dangerous Blind Spot in IS Research: False Positives Due to Multicollinearity Combined With Measurement Error.
AIS Electronic Library (AISeL) (Association for Information Systems)
.