While social comparison research has traditionally focussed particularly upon factors affecting individuals' comparison desires and choice of a comparison target, in the past decade the attention has shifted to the consequences of social comparison for mood, well-being and satisfaction. A number of studies have provided evidence for the relation between downward comparison and general well-being, and it has been clearly demonstrated that feeling better off than others on a specific dimension is related to satisfaction with one's standing on that dimension (for reviews, see Buunk, 1994; Taylor & Brown, 1988; Wills, 1981). Although many studies on this issue have been correlational, thus impeding causal conclusions, in an increasing number of laboratory experiments the effects of social comparison information have been examined (e.g., Gibbons & Gerrard, 1991; Klein, 1997; Pelham & Wachsmuth, 1995). The usual paradigm in this area, however, consists of a procedure in which researchers provide participants with social comparison information after which the impact of this information is assessed (for reviews, see Wood, 1989, 1996). Characteristic of this type of paradigm is that participants are passive recipients of social comparison information.

Hence, to examine experimentally whether cognitive downward comparison improves perceived relationship quality, the current research extended the existing literature by applying a new paradigm that induced participants to actively engage in cognitive downward social comparison. Furthermore, since the effects of social comparison have scarcely been studied in the domain of intimate relationships, we sought to extend previous research on social comparison in intimate relationships, and to provide insight into the effects of cognitive downward social comparison on relationship satisfaction.

The experimental paradigm was a thought-generating task which asked participants to generate positive qualities in which they consider themselves, their intimate partners, or their own relationships in noncomparative or comparative terms. This paradigm was designed to test the prediction that relationship quality is not so much positively affected by noncomparative evaluations in terms of good but instead by the belief that one's own situation is better than that of most others (cf. Emmons & Diener, 1985; Klein, 1997). Cognitive downward comparison was expected to be a particularly successful strategy for people who experience problems in their relationship to reevaluate their situation and to make them feel relatively more satisfied with their relationship. Furthermore, we expected that downward comparison would exert a stronger buffering effect for individuals high in social comparison orientation. We examined this model in both laboratory and field studies, testing our hypotheses in samples of undergraduates as well as in a non-student sample of individuals who had been in relationships for a relatively longer period of time.

The current series of studies clearly indicated that cognitive downward comparison can have beneficial effects on the development and maintenance of relationship quality (i.e., global, evaluative judgments of the relationship; cf. Fincham & Bradbury, 1987). Study 1 demonstrated that participants who were induced to engage in self- and partner-enhancing cognitive downward comparison experienced more relationship satisfaction and felt more committed to their relationship than participants who were asked to generate noncomparative self- and partner-evaluations. Consistent with the theoretical model, Study 2 demonstrated that preexperimentally identified levels of relational discontent and individual differences in social comparison orientation moderated the effects of cognitive downward comparison: under high relational discontent, partner-enhancing downward comparisons lead to more satisfaction and commitment, particularly when individuals were dispositionally strongly inclined to compare their own circumstances with those of others. In addition, Study 3 replicated the findings of Study 2 in a sample of undergraduate students. Relational discontent and comparison orientation as measured approximately four weeks prior to the experimental sessions in the laboratory moderated the relationship-enhancing effects of cognitive downward comparison. In both Studies 2 and 3, the negative impact of discontent regarding the relationship with the partner on individuals' relationship quality was buffered by comparison of one's own relationship with other relationships of inferior quality, particularly among those who were high in social comparison orientation.

At least two alternative interpretations of the differential effects of the thought-generating tasks are possible. First, a potential limitation of the current research concerns the self-reporting nature of most of the measures. Therefore, it is conceivable that the beneficial effect of the downward comparison thought-generating task is simply a result of demand characteristics and that the effects do not represent any real improvements in perceived quality of the relationship. That is, participants might have concluded from the thought-generating task that they were supposed to think "good" or "exceptional", and that they should continue in the same manner in answering the subsequent questions (cf. Bless, Strack, & Schwarz, 1993; Schwarz, 1994). Of course, the problem with this interpretation is that it cannot explain why only generating qualities in which one's relationship was superior, and not generating qualities in which one's relationship was good resulted under certain conditions in increased satisfaction. Nevertheless, to rule out the possibility of demand characteristics, in Studies 1 and 3 we measured perceived quality of the relationship by both self-report measures and unobtrusive response latencies measures of which the participants were unaware. In these studies, as we hypothesized, cognitive downward comparison influenced the unobtrusive measures of response latencies (although the response latencies to the commitment index in Study 3 did not reveal acceptable levels of significance, they followed the predicted pattern). That is, the heightened salience of superior relationship qualities seemed to have resulted in increased levels of relationship satisfaction which, consequently, led to increased accessibility of the attitude toward the relationship. This effect of the downward comparison thought-generating task on the response latencies of the dependent variables underlines the strengths of our findings.

A second plausible artifactual explanation is the possibility that the comparison thought-listing task induced participants to list qualities that were more positive than the qualities generated in the no-comparison thought-listing task. That is, the obtained effects might not be due to the fact that participants actively engaged in downward social comparison but to the fact that they generated qualities in the comparison condition that differed in valence from those in the no-comparison condition. However, controlling for positivity of the qualities did not eliminate the effects of the experimental manipulation of cognitive downward comparison. These supplemental analyses speak to the validity of the experimental manipulation and suggest that it is cognitive downward comparison that accounts for the obtained effects in the current series of studies.

Considered as a whole, the present findings indicate that people may enhance their well-being through a downward social comparison process when they are unhappy about the relationship with their partner. The results of these three studies are consistent with Wills' downward comparison theory (1981, 1987) and selective evaluation theory proposed by Taylor et al. (1983) that would predict that when people experience relational discontent, selective evaluation through downward comparison may make individuals feel relatively better about their relationship.

Also, the present research may have methodological implications for the social comparison literature. The use of a thought-generating paradigm to induce active cognitive downward comparison is to the best of our knowledge new to the social comparison literature. That is, participants were not presented with social comparison information from an outside source, but an active, cognitive downward comparison process was induced. As such, the thought-listing task which asked participants to list positive qualities in noncomparative or comparative terms may offer a useful paradigm to examine experimentally the possible beneficial consequences of cognitive downward social comparisons for individuals' psychological well-being.

In addition to these methodological implications, the present research also has a number of potentially important theoretical implications for social comparison research. Not only does the present research clarify how cognitive downward comparisons may affect the way individuals evaluate their situation, the present research is also one of the first to provide direct evidence that there are meaningful individual differences in social comparison orientation. Although various authors have suggested that individuals differ in their tendency to compare themselves with others (e.g., Brickman & Bulman, 1977), there is virtually no research that has addressed this issue, and shown that individual differences in this regard lead to different responses to social comparison. The orientation assessed by the scale developed by Gibbons and Buunk (1999) indeed appears directly relevant with respect to the impact of social comparison. Studies 2 and 3 clearly showed that cognitive downward comparison buffered the effect of discontent with the relationship upon relationship evaluation only for individuals high in social comparison orientation. These findings seem to qualify our knowledge about the beneficial consequences of downward comparison as such consequences are only found concerning individuals who are dispositionally strongly inclined to compare themselves with others. Apparently, the buffering role of downward comparison as proposed by Wills (1981), may not be as general a phenomenon as sometimes supposed.