Warming Up and Cooling Down: Self-Esteem and Behavioral Responses to Social Threat During Relationship Initiation

Abstract

Social threats during relationship initiation often cause people to engage in cold behaviors that bring about rejection. However, interpersonal risk-regulation theory suggests that such processes will be moderated by global self-esteem. In two experiments that manipulated the threat of rejection, single participants communicated via video camera with an opposite-sex interaction partner (actually a confederate). As expected, social threat caused higher self-esteem individuals (HSEs) to exhibit a warming-up behavioral response but caused lower self-esteem individuals (LSEs) to exhibit a cooling-down behavioral response, according to both observer-reports and self-reports, which in turn led observers to like HSEs more than LSEs. Furthermore, these effects were independent of similar, previously documented, interpersonal risk-regulation effects on participants' perceptions of acceptance from the confederate.

Keywords::

• Self-esteem
• Risk-regulation
• Self-fulfilling prophecy
• Threat
• Acceptance
• Warmth

Notes

^* E-mail: [email protected]

^** E-mail: [email protected]

^† E-mail: [email protected]

^1. Cameron et al. (2010) included crossing legs in their acceptance cues composite. Although lay theories suggest that crossing one's legs toward one's partner can indicate liking (e.g., Potter, 2014), empirical evidence suggests that crossed legs is a closed posture that conveys a lack of romantic desire (especially when enacted by men; e.g., Grammer, 1990). Cameron and colleagues choice to include crossed legs in their acceptance cues variable was not particularly consequential, because the confederates did not cross their legs in the response videos that participants rated. However, when assessing participants' own behavior in the present research, we have opted to conceptualize leg crossing as a cold behavior indicative of romantic disinterest.

^2. Cold behavior still predicted liking when warm behavior was not included in the model, β = − 0.34, 95% CI [ − 0.22, − 0.05], t(71) = − 3.07, p = .003, ΔR² = .091, p = .003.

^3. Coders also rated participants' frequency of crossing their arms across their chests, but one coder perceived that no participants engaged in this behavior. This behavior was omitted from the composite variable described in text.

^4. Although the reliability of this composite is lower than ideal, it is common for researchers to aggregate and analyze reports of behavior from different sources to gain a complete picture of the behavior in question, even when sources do not agree at conventionally accepted levels (e.g., mother, father, and teacher reports of a child's behavior; Koss et al., 2013).

^5. We conducted two additional sets of mediation analyses, one using participant-reported mediators and the other using observer-reported mediators. The results of those supplemental analyses were remarkably consistent with the analyses we report in the main body text, and can be found in the Supplementary Materials for this manuscript, available at http://home.cc.umanitoba.ca/∼cameron2/.