Independent Variables: The Role of Confounding and Effect Modification

The valid interpretation of observational and experimental research necessitates that the role of confounding and effect modification be taken into account. Confounding occurs when an association (or lack thereof) between a dependent and independent variable is due to a third variable that itself is associated with both the dependent and independent variables. Effect modification, a related phenomenon, is present when an association between a dependent and independent variable varies according to a third variable. This chapter will provide a general overview of the principles of the concepts of confounding and effect modification in observational and experimental research including techniques for detecting and accounting for these issues. The role of confounding is an important consideration in observational research; this is also true, though to a lesser extent, of experimental studies, particularly those with small numbers of participants, such as is the case in driving simulator studies. Confounding and effect modification can exert their effects independently from the scenario being displayed or dependent variable being measured. Confounding and effect modification are important concepts that can undermine the validity of an otherwise well-designed and conducted study. As such they should be prominent in the minds of investigators during the conception and design of a study. Studies that do incorporate these concepts into their design should ensure that they are properly accounted for throughout, particularly during the analytical phase.