Title

Dealing with Deliberate Distortions: Methods to Reduce Bias in Self-Report Measures of Sensitive Constructs

Date of Award

2012

Document Type

Dissertation

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Psychology/Industrial-Organizational

First Advisor

Milton D. Hakel, PhD

Second Advisor

Scott E. Highhouse, PhD (Committee Member)

Third Advisor

Dale S. Klopfer, PhD (Committee Member)

Fourth Advisor

Jane V. Wheeler, PhD (Committee Member)

Abstract

Psychological constructs are sensitive when they are threatening, have socially acceptable answers, and have different psychological costs associated with disclosure. Among the challenges researchers face when studying sensitive constructs are intentional response distortions by respondents to self-report measures of these constructs. Intentional response distortions are conscious efforts to misreport in a socially acceptable way to be seen more positively. Respondents may overreport their standing on desirable constructs or underreport their standing on undesirable constructs to come across more positively. This tendency towards misreporting makes the validity of the scores of self-report measures of sensitive constructs questionable. Although researchers have proposed many methods to address this challenge of deliberate distortions, a thorough comparison of these methods is missing in the literature. After reviewing seven of these methods, noting their strengths and limitations, a subset of them were compared. Specifically, utilizing a measure of counterproductive workplace behaviors, an experimental investigation of four different methods (i.e., conventional anonymity, counterbiasing, indirect questioning, implicit goal priming) was conducted to determine which method is most effective by addressing two research questions. In the first question, mean reporting of counterproductive behaviors from a 2 (honesty, neutral prime) X 3 (anonymity, counterbiasing, indirect questioning) factorial design was investigated to determine which method resulted in the highest reported mean frequency of counterproductive behaviors. Results showed that the indirect questioning method resulted in the highest mean reporting-the other methods did not differ from each other. To corroborate these results, the triangulation approach to establishing null results was used in question two to determine if the estimates from the self-report conditions corresponded to a point estimate determined from a method currently considered best practice. The proportion of respondents admitting to drinking alcohol or taking drugs while on the job was computed for each cell of the 2 X 3 design and compared to a proportion estimated using the unmatched counts technique. Although the prerequisites for interpreting null results were met, no meaningful null results were found. This study suggests that indirect questioning may be effective at reducing intentional response distortions; however, future research is needed to confirm these results.