Spontaneous Correction in the Behavioral Confirmation Process: The Role of Naturally-Occurring Variations in Self-Regulatory Resources

Abstract

Perceivers' tendencies to correct for expectancy-related biases can be affected by experimental manipulations of goals and cognitive resources. In the current research, we examined the role of naturally-occurring, environmentally-produced variations in self-regulatory resources, represented by the time of semester in which college students participated. A pilot study established the association between time of semester and self-regulatory resources. Using a simulated job interview paradigm, interviewers were induced with extreme expectancies regarding their applicants. The effects of expectancy valence depended on time of semester. Interviewers with depleted self-regulatory resources (i.e., late-semester participants) asked expectancy-biased questions, elicited expectancy-confirming behavior from applicants, and formed expectancy-consistent impressions, while their less-depleted, early-semester counterparts did not. The findings suggest that interviewers engaged in spontaneous behavioral and perceptual correction for the biasing influence of extreme expectancies, but only when sufficient self-regulatory resources were available. The discussion focuses on theoretical implications for models of bias correction and applied significance.

ACKNOWLEDGEMENTS

We would like to thank Leigh Ann Vaughn, Robert Mather, Brandon Randolph-Seng, and Monica Munoz for their helpful comments on earlier drafts of this manuscript, as well as Jennifer Gnepper and Kelli Parsons for their hard work in coding the interviews. Partial reports of these data were presented at the 2004 Society for Personality and Social Psychology conference.

Notes

¹Participants were evenly spread across the weeks of the semester, resulting in a slightly platykurtic distribution (skewness = −.09, kurtosis = −1.32). However, kurtosis values were within 2 standard errors and were within an acceptable range for treating this variable as normal in subsequent analyses (Tabachnick & Fidell, 2007).

²A reviewer of a previous draft noted that self-regulatory resources may not drop off linearly as the semester progresses. Therefore, for the four primary dependent variables, quadratic and cubic effects of time of semester, as well as their interactions with interviewer expectancies were also examined. Only one of these higher-order terms across analyses (out of 16) attained significance. Since the finding did not replicate across measures and may have been due to chance, it will not be discussed further.

³A reviewer noted that self-regulatory resources may also fluctuate within the course of a single day. When the same analysis on question quality was conducted replacing time of semester with hour-of-the-day, a conceptually consistent pattern of results emerged. Specifically, an interaction between expectancy valence and centered hour-of-day was found, β = .50, sr ² = .13, t(36) = 2.53, p = .016. For those participating early in the day (+1 SD, about 1 pm), the simple slope for interviewer expectancies predicting question quality was not significant, β = −.06, sr ² = .001, t(36) = −.26, p = .797. For those participating at the mean hour of the day (about 3 pm; β = .33, sr ² = .10, t(36) = 2.21, p = .034) and those participating later in the day (about 5 pm; β = .72, sr ² = .23, t(36) = 3.31, p = .002), the simple slopes were significant, such that interviewers with positive expectancies asked more positively-focused questions than those with negative expectancies. This finding was consistent with the notion that temporary situational fluctuations that reduce self-regulatory resources can enhance aspects of the behavioral confirmation process; as the day progressed (and self-regulatory resources were presumably drained by daily events), interviewers' behavior toward the applicants was influenced more by their expectations. However, this pattern of results did not emerge on the applicant performance and interviewer impression measures, and therefore will not be discussed further.