Issues on Control Variable

Inclusion of control variables, and in particular multiple control variables, is a safer and more statistically conservative approach than not including them (Carlson & Wu, 2012; Spector & Brannick, 2011). Whereas each of these assumptions relies on rather large inferential leaps, perhaps the biggest and potentially most frequently violated assumption is that control variables hold theoretically meaningful relationships with predictors and criteria (Bono & McNamara, 2011; see results by Carlson & Wu, 2012).

The inclusion of control variables not only reduces available degrees of freedom and statistical power but also has the potential to reduce the amount of explainable variance available in outcomes of interest (Becker, 2005; Breaugh, 2008; Carlson & Wu, 2012). That is, when control variables are related to the predictor or criterion, results may give the appearance that the predictor is not related to the criterion or is related in an unexpected direction (Becker, 2005; Breaugh, 2008; Carlson & Wu, 2012; Meehl, 1971; Spector & Brannick, 2011) - Leading to an incorrect conclusion .

Inclusion of control variables in most cases implicitly assumes that the control variables are somehow either contaminating the measurement of the variables of interest or affecting the underlying constructs, thus distorting observed relationships among them.

However, exclusion of control variables can also lead to an incorrect conclusion that the predictor relates to the criterion when, in fact, there is no such relationship.

Accordingly, the inclusion or exclusion of control variables has important implications for theory and practice as such decisions can change substantive study results (e.g., Rode et al., 2007) as well as limit the ability to replicate, extend, and generalize a study’s findings (Becker, 2005; Breaugh, 2006; 2008; Carlson & Wu, 2012)

What Controls are Included?

Gender, Age, Tenure, Level of Education, Race, Personality, Organization/group size, Work Experience, Workload/hours worked, Family-Related Concerns, Industry, Income-Related Controls, Social Desirability, etc.

Decision-Making of Control Variables

The choices and procedures regarding the handling of statistical controls should be described in detail, regardless of whether or not the control is ultimately included, to ensure transparency and maximize the likelihood of reproducibility of results in the future (Bernerth & Aguinis, 2016).

This is essential to satisfy a skeptical scientific audience that regularly lacks clear understanding as to why a study includes a given control variable or why its absence hinders scientific interpretation and advancement (Aguinis & Vandenberg, 2014)

Also, Spector and Brannick (2011) proposed that researchers should be explicit rather than implicit regarding the role of control variables and match hypotheses precisely to both the choice of variables and the choice of analyses.

Control Variables