Angel or demon? Self-monitoring differences in the mental representations of current versus former romantic partners

Abstract

We investigated whether high and low self-monitors had (a) segregated/integrated and (b) uniformly favorable (angelic)/uniformly unfavorable (demonic) mental representations of actual romantic partners. Self-monitoring was assessed using the 25-item Self-Monitoring Scale and the structure of mental representations was calculated using a modified segregation/integration measure. In Study 1, low self-monitors were more likely than high self-monitors to have (a) segregated and (b) disproportionally favorable mental representations of their current romantic partners. In Study 2, low self-monitors were more likely than high self-monitors to have (a) segregated and (b) generally unfavorable mental representations of their former romantic partners. Self-monitoring differences were not moderated in either study by relationship longevity or conflict.

Keywords:

• Self-monitoring
• mental representations
• partner knowledge structures
• romantic relationships
• relational illusions

Acknowledgment

We are grateful for the assistance of Kaye Sommer in collecting data for the studies reported in this paper and for the editorial comments of Ashley Allen, LouAnne Hawkins, and several anonymous reviewers.

Notes

1. In order to avoid having cell sizes with expected frequencies of five or less in our analysis, we collapsed the relationship status variable into three categories: casually/seriously dating a romantic partner, in a committed romantic relationship, and cohabitating/married.

2. When we performed regression analyses with self-monitoring as a categorical predictor variable and centered scores on relationship length as a continuous predictor variable, we again found a non-reliable main effect of relationship length, F(1,162) = 2.55, p = .112, and no interaction of self-monitoring and relationship length, F < 1.00.

3. As was the case with relationship length, there was no substantive difference between the results of our 2 × 2 ANOVA and an analogous regression analysis with centered scores for relationship conflict as a continuous predictor: a main effect of relationship conflict, F(1,162) = 11.28, p = .001, but no self-monitoring × relationship conflict effect, F < 1.00.

4. There was skewness and/or kurtosis in the distributions of responses concerning time elapsed since former romantic relationships ended (coefficient of skewness = 2.41, coefficient of kurtosis = 7.18), length of former romantic relationships (coefficient of skewness = 2.46, coefficient of kurtosis = 7.84), and length of current romantic relationships if any (coefficient of skewness = 3.88, coefficient of kurtosis = 21.83). For responses concerning current relationship status, there were several cells with expected frequencies of five or less. We therefore collapsed response options for our χ² analyses. Additionally, our χ² analysis of participants’ responses concerning the length of their current relationship only included participants who were in a new romantic relationship (N = 117).

5. As one reviewer also suggested, it is also plausible that as the relationship of low self-monitors began to deteriorate, the valence of their mental representations of their soon to be (but not quite yet) former partner may shift from favorable to unfavorable. This shift, in turn, may have been the precipitating event in the decisions of low self-monitors to end that relationship rather than low self-monitors changing those mental representations after the fact to cope with their loss.