Reactions to Physician-Inspired Shame and Guilt

Abstract

Participants reported attributions and responses to shame (Study 1 and Study 2) and guilt (Study 2) experienced during an interaction with a physician. These emotional interactions elicited both negative and positive reactions. Regardless of whether the person felt guilt or shame, self-condemnation and the perception that the physician intentionally induced the emotion were associated exclusively with negative outcomes (e.g., ceasing physician visits), whereas negative attributions about one's behavior (rather than the self as a whole) were associated with primarily beneficial outcomes (e.g., positive impact). We discuss how these data bear on theories of shame and guilt.

Notes

¹Dispositional shame predicted having a shaming experience, r(483) = .13, p = .003. However neither dispositional shame nor dispositional guilt significantly predicted any of the other dependent measures. (Here, as elsewhere in the study, Holm-Bonferroni corrections were used.) We therefore do not report further on the Test of Self-Conscious Affect.

^{2 2}One item was inadvertently excluded from our measurement of self-condemnation.

^{3 3}As an exploratory question for our next study, we also asked, in a separate question, whether doctor was purposely trying to induce guilt. The results were nearly identical to those produced by the question about purposefully inducing shame, and the correlation between these items was extremely high, r(111) = .79, p < .001. This suggests that most patients felt that their doctor was trying to induce both shame and guilt.

^{4 4}Although this alpha level is low, the two items show similar results when analyzed individually. The construct of “possibility of change” is assessed more extensively in Study 2 with six similar items (α = .87).

⁵Due to the relatively small number of men that reported a shaming encounter, we did not initially examine gender differences. In response to an anonymous reviewer's inquiry, we did so. Women tended to have more negative reactions and men more positive reactions on outcome measures, although none of these were statistically significant after Holm-Bonferroni alpha correction.

^{6 6}To assess whether outcomes varied by the topic of the shame incident, we attempted to examine the responses of participants who reported that the event focused exclusively on one topic. However, even in the three most common types of incidences, samples sizes were too low to perform these analyses (weight n = 10, sex n = 17, teeth n = 15).

Note. Months since incident uses the natural log of the time since the shame incident occurred.

*p < .05.

**p < .01.

***p < .001, uncorrected.

^{7 7}This correction caused some effects that would otherwise be significant to no longer reach statistical significance. However, Study 2 shows a strikingly similar pattern of results.

Note. Outcomes are designated as positive (+) or negative (−) responses. The r values describe the zero order correlation between each outcome and the degree to which the doctor was perceived as intentionally inflicting shame.

*p < .05.

***p < .001, after a Holm-Bonferroni stepwise correction.

Note. For type of incident, participants could choose more than one option. Categories were not mutually exclusive, hence entries sum to more than 100%. The “Other” category includes both people who chose the option “other” and people who chose a type of incident or medical professional reported by less than 5% of participants in both conditions.

^a n = 81.

^b n = 109.

Note. No alpha correction applied. Months since incident uses the natural log of the time since the shame or guilt incident occurred.

*p < .05.

**p < .01.

***p < .001.

^{8 8}These effects remained significant when self-condemnation was added to the model as a covariate.

^{9 9}We found the same pattern of results that were seen in Study 1 when we tested the possibility that condemnation mediated the effects associated with intentionality.

^{10 10}In Study 1, before correcting for alpha inflation, perception of control also mediated the relationship between behavior condemnation and positive health behaviors. However, after we employed Holm-Bonferonni alpha protection, behavior condemnation was not significantly related to specific positive health behaviors; thus, mediational analysis was not performed. We used this as an exploratory analysis to motivate the more expanded measure of possibility of change developed for Study 2.