How Can We Tell When a Heuristic Has Been Used? Design and Analysis Strategies for Capturing the Operation of Heuristics

Abstract

Communication research based on social cognition draws heavily from dual process models such as the heuristic-systematic model. Both heuristic and systematic processing are said to be aided by quick rules of thumb or mental shortcuts called heuristics. However, the operationalization of heuristics is quite problematic because their use in decision-making is not directly measured. Scholars claim the operation of specific heuristics in specific situations, based often on clever experimental evidence. We propose a methodological framework that calls for both manipulation and measurement of heuristics in order to offer more direct evidence of their operation. We first review different existing approaches in the literature for operationalizing heuristics. We then discuss our approach and describe the application of moderated mediation to analyze the resulting data. We conclude with a study idea and simulated data that illustrate how our proposed design and analysis framework could be applied in communication research.

Notes

1. ¹Participants were asked to indicate the extent to which they typically agreed (disagreed) with a person they liked (disliked).

2. ²X = independent variable, W = moderator, M = mediator and Y = dependent variable.

3. ³Before moving to steps (ii) and (iii), participants will be given the impression that they are participating in a different, unrelated study.

4. ⁴PROCESS is one option for researchers using SPSS and SAS; there are other statistical programs such as MPlus that can run similar analyses.

5. ⁵Hayes (2013b) observes that a significant interaction effect is neither necessary nor sufficient for probing a moderated mediation. On a related note, Preacher et al (2007) have argued that evidence of an indirect effect is not required in order to probe whether that indirect effect is moderated or not.

6. ⁶It is important to note that the numbers in the PROCESS output are not based on real data. They are used to illustrate one potential pattern of findings that could emerge from employing the design and analysis strategy described in this article.