Commentary on “Mediation analysis and categorical variables: The final frontier” by Dawn Iacobucci

Abstract

The commentary discusses several topics in Dr. Iacobucci's article on mediation analysis with categorical variables including the lack of equivalency of mediated effect measures, ab, and c–c′, that are algebraically equivalent in ordinary least squares regression models. The commentary raises the issue of correlated mediation coefficients and presents formulas to include the correlation in the test of significance. References to new developments in causal mediation analysis for categorical variables are given and the importance of programs of research to establish a mediation theory is emphasized.

Going Native: Effects of Disclosure Position and Language on the Recognition and Evaluation of Online Native Advertising

“…For hypotheses 6(a) and 6(b), testing mediation effects with both a dichotomous mediator and dichotomous outcome required nonstandard approaches due to the lack of algebraic equivalence of some of the estimators in the model (e.g., Herr n.d. ;Iacobucci 2012;MacKinnon and Cox 2012). We chose to approach this by using weighted least squares regression with means and variance adjusted in Mplus, which treats each observed dichotomous variable as an indicator of an underlying latent variable (MacKinnon and Cox 2012; Hayes 2012).…”

Relationships among Maternal Stress and Depression, Type 2 Responses, and Recurrent Wheezing at Age 3 Years in Low-Income Urban Families

“…This was performed using a four-step approach with several linear and logistic regression models to determine if coping skills mediated the relationship between stress or depression and recurrent wheeze, and if depression mediated the relationship between stress and recurrent wheeze (30). This was followed up with a test for partial mediation, using the techniques described by Iacobucci, which employs both linear and logistic regression models (31). Relationships between cytokine expression and perceived stress score, depression score, and composite stress were examined using linear regression; cytokine values were log transformed due to their skewed distributions.…”

Medical tourists’ emotional and cognitive response to credibility and Servicescape

“…Image has also been analysed as being a mediator between exogenous and endogenous variables (MacKinnon & Cox, 2012), where the exogenous variables are credibility (e.g. Veasna et al, 2013) or store environment or other intrinsic cues (Baker et al, 1994) and behavioural intentions as consequences or even service quality (Gröonroos, 1990;Loureiro & Miranda, 2008).…”