Abstract
Cross-relationship comparisons are an integral part of relationship processes, yet little is known about the impact of these comparisons in daily life. The present research employed a dyadic experience-sampling methodology (N = 78 couples) with end-of-day surveys, end-of-week follow-up, and a 6-month follow-up to examine how individuals make cross-relationship comparisons in daily life, the cumulative impact of these comparisons over time, and the dyadic consequences of such comparisons. Participants made more downward than upward comparisons; however, upward comparisons had a more lasting impact, resulting in decreased satisfaction and optimism, and less positive self-perceptions and partner perceptions, at the end of each day and the week. Individuals who made more upward comparisons were also less satisfied 6 months later. Individuals were also affected by their partner’s comparisons: On days when partners made more upward comparisons, they felt less satisfied and optimistic about their relationship and less positive about themselves and their partner.
On the occasion of Michelle Obama’s birthday, husband Barack posted a message on Twitter, noting, “In every scene, you are my star,
Several theories highlight the importance of comparisons in relationship processes. For example, interdependence theory states that such comparisons shape the standards individuals use to evaluate their relationship (Thibault & Kelley, 1959). In turn, these standards influence their satisfaction. Rusbult and colleagues (2000) have proposed that making comparisons with worse-off relationships promotes stability by contributing to the positive illusion that one’s relationship is superior to most others. Despite this theorizing, little is known about how individuals make these comparisons in daily life or their impact on relationships. In the present research, we first examine how individuals make and react to cross-relationship comparisons in naturalistic settings, assessing the frequency, targets, context, and domains of these comparisons, along with individuals’ immediate responses. Second, we assess the cumulative impact of these comparisons over time on perceptions of the self, partner, and relationship. Third, we examine dyadic effects, showing how comparisons may not only impact one’s feelings about oneself and one’s partner but also how one’s partner feels about themselves and the relationship.
Cross-Relationship Comparisons in Naturalistic Settings
A growing literature has examined social comparisons in the context of romantic relationships. For example, individuals who report a greater tendency to make more relationship comparisons tend to be less satisfied (Buckingham et al., 2019; Smith LeBeau & Buckingham, 2008). How individuals interpret upward comparisons, moreover, also plays a role in outcomes: Individuals who view superior couples as illustrating opportunities for improvement experience greater satisfaction than those who focus on their own inferiority (Morry et al., 2019; Murry & Sucharyna, 2016, 2019). On balance, however, past research suggests that, except in specific circumstances, upward comparisons have negative outcomes. For example, individuals tend to be more negatively affected by the example of a superior than inferior couple unless they are highly committed (Thai et al., 2020) or are both high in social comparison orientation and are led to believe that the superior couple had worked hard to become successful (A. P. Buunk, 2006). Thus, individuals exposed to superior couples typically focus on how they are dissimilar from the superior couple (contrast effects), resulting in more negative relationship evaluations (B. P. Buunk & Ybema, 2003) and less satisfaction (Morry & Sucharyna, 2019). Indeed, a meta-analysis suggests that contrast effects are the most likely response to upward comparisons (Gerber et al., 2018). It is unclear, however, whether individuals are especially influenced by upward relative to downward comparisons. Several studies have examined upward and downward comparisons without a control group (B. P. Buunk & Ybema, 2003; Morry et al., 2019), making it impossible to determine whether effects are driven primarily by upward or downward comparisons, or both. Moreover, participants in past studies were presented with (A. P. Buunk, 2006; Thai et al., 2020) or asked to generate (Morry & Sucharyna, 2019) a single relationship example; such examples may not represent the actual relationships that individuals encounter in more naturalistic settings. Finally, it remains unclear whether comparisons have a cumulative impact on individuals over time; even if one comparison offers some hope for improvement, exposure to a series of superior couples may result in more negative outcomes.
We used experience sampling to examine individuals’ responses to cross-relationship comparisons in daily life for 1 week. This enabled us to examine the frequency, direction, and impact of comparisons and to collect descriptive data regarding the domain, context, and targets of comparisons. Although our design did not allow for the inclusion of a no-comparison control group, we were nevertheless able to examine within-person differences in the outcomes that participants experienced in the presence and absence of relationship comparisons, allowing us to more precisely determine the separate impact of upward and downward comparisons.
In assessing comparison impact, we considered individuals’ self-perception, partner perceptions, and relationship perceptions. We were thus able to examine whether these outcomes would operate in parallel or whether a different pattern might emerge. A meta-analysis suggests that upward comparisons are more likely to decrease self-evaluations and downward comparisons are more likely to increase self-evaluations (Gerber et al., 2018); however, this meta-analysis focused on comparisons between the self and another person, in which individuals are at least somewhat responsible for their outcomes. In contrast, although both partners contribute to relationship outcomes, individuals can attribute responsibility in different ways, depending on their motives. Consistent with research on self-serving biases (Malle, 2006), individuals may take personal credit when their relationship is superior, which enhances self-evaluations, but ascribe blame to their partner when their relationship is inferior, leading to more negative partner perceptions and less of an impact on self-perceptions. Given that individuals typically include their partners in their self-concepts to some degree (Aron et al., 1991), and comparison effects on partner and relationship perceptions operate in tandem (Thai et al., 2019), we expected that upward comparisons would negatively affect all three outcomes, relative to downward or no comparisons, and downward comparisons would positively affect these outcomes.
We also included a measure of post-comparison affect. Past research suggests that individuals may feel positively or negatively after making upward or downward relationship comparisons, depending on what aspect of the comparison they focus on, resulting in a variety of affective responses (B. P. Buunk et al., 1990). This study, however, relied on retrospective reports of how relationship comparisons in general made participants feel and thus may not reflect their actual affective responses. It may be that individuals’ feelings about their relationship in the moment influenced their responses. Indeed, this study found that individuals who were more dissatisfied reported more negative responses to both upward and downward comparisons. Consequently, we examined individuals’ mood immediately after they made a specific comparison. We predicted that, consistent with Gerber et al. (2018), individuals would feel worse after upward comparisons relative to downward or no comparisons and better after downward comparisons.
In addition, we examined how these comparisons affected relationship outcomes by assessing relationship satisfaction and optimism; this allows for the possibility that upward comparisons might have a negative impact on one’s satisfaction in the present but simultaneously highlight the opportunity for future improvement (Morrie & Sucharyna, 2019). Because satisfaction reflects the level of positive or negative affect experienced in a relationship (Rusbult et al., 1998), we expected comparisons to influence satisfaction the same way it influences mood: Individuals would feel less satisfied after upward comparisons relative to downward or no comparisons and more satisfied after downward comparisons. We note, however, that such future-oriented outcomes tend to be limited to circumstances when individuals are in a novel situation characterized by uncertainty and can thus benefit from role models who demonstrate a path to future success (Lockwood et al., 2012). Given that individuals making routine relationship comparisons in their daily lives would not necessarily have evidence to suggest that they would become like the superior relationship in the future, nor be in a novel situation, we predicted individuals would be less optimistic after upward than downward comparisons.
Finally, to the extent that upward comparisons make relationship shortcomings more salient, individuals may be more likely to voice their displeasure with the relationship on days when such comparisons occur. Accordingly, we examined whether individuals who make more upward comparisons on a given day would be more likely to experience conflict that day. In contrast, downward comparisons would not be associated with conflict on a given day because such comparisons indicate that the couple is doing well. In sum, we predicted that individual upward comparisons would have a negative impact on individuals’ mood, self-perceptions, and partner perceptions, as well as their relationship satisfaction, optimism, and conflict, relative to downward or no comparisons. In contrast, downward comparisons would have a positive impact on these outcomes.
Cumulative Effects of Cross-Relationship Comparisons
By examining the frequency of comparisons in daily life, we were also able to assess whether these comparisons exert a cumulative impact over time. We predicted that upward comparisons would have significant cumulative effects on perceptions of the self, partner, and relationship, but downward comparisons would not. Past evidence suggests that negative events and outcomes are more impactful than positive ones across a broad range of psychological phenomena (Baumeister et al., 2001). In the case of social comparisons, individuals who make more upward comparisons experience diminished life satisfaction, mood, and self-evaluations (Midgley et al., 2021), regardless of the number of downward comparisons they make, suggesting that upward comparisons have a negative cumulative effect but downward comparisons do not have a positive cumulative effect. Moreover, research examining the multiple means through which individuals assess their relationships (e.g., objective information, personal standards, social comparison) found that although separate subscales assessing individuals’ tendency to make upward and downward comparisons were each negatively associated with satisfaction, this association was stronger for upward comparison (Buckingham et al., 2019). Furthermore, upward comparisons may simply be more memorable, and thus more impactful, because they violate people’s expectations. Individuals may be surprised by an upward comparison because it violates their assumption that their relationship is better than most others (Rusbult et al., 2000) and thus remember it better; downward comparisons confirm this assumption and are thus less memorable. Consequently, whereas responses to upward comparisons may show stability over time, responses to downward comparisons may be more fleeting. Indeed, individuals tend to remember events that violate their expectations better than events that are more consistent with their expectations (Kalbe et al., 2020). In sum, we predicted that frequent upward comparisons would have a negative cumulative impact on individuals’ evaluations of themselves and their partner, as well as their relationship satisfaction and optimism, and that this impact would outweigh any benefits of downward comparisons. We also expected the effects of upward comparisons to be more stable than the effects of downward comparisons.
Our design, which involved participants completing six experience-sampling surveys and a nightly survey each day for a week, enabled us to examine cumulative effects of both within- and between-participant responses to comparisons, at the end of each day. Because comparisons were measured throughout the day and outcomes were measured at the end of the day, we were able to examine the temporal effects of relationship comparisons, in their aggregate form, on a given day, satisfying the temporal precedence criteria for establishing causal relationships from covariation (Mulaik, 2009). Furthermore, by measuring outcomes after the experience-sampling week ended, we could assess the cumulative between-person effects of comparisons. The within versus between distinction is theoretically important because these analyses address different questions relevant to our hypotheses. For example, using between-participant analyses, we can examine whether individuals who make more upward relationship comparisons will have more negative outcomes than individuals who make fewer upward comparisons in the moment, at the end of each day, and the end of the experience-sampling week. Such analyses, however, are open to the alternative interpretation that individuals who are dissatisfied with their relationships are more prone to making upward comparisons. Within-participant analyses, in contrast, enable us to more precisely pinpoint the impact of each individual comparison, as well as their cumulative effects, regardless of individuals’ preexisting relationship satisfaction. Consequently, we can examine whether individuals experience worse outcomes on days when they make more upward comparisons relative to days when they make fewer upward comparisons. Furthermore, these within-person analyses allow us to conduct mediation analyses examining the temporal stability (whether participants continue to feel better or worse) of individuals’ responses to upward and downward comparisons. Past research has typically examined individuals’ responses to comparisons immediately after they occur (A. P. Buunk, 2006; Morry & Sucharyna, 2019; Thai et al., 2020). Thus, it is unclear whether these reactions are short-lived or longer-lasting. Given that individuals spend more time processing negative than positive information (Baumeister et al., 2001), it seems likely that the effects of upward comparisons would last longer than downward comparisons. We examined this directly by testing whether comparisons made in one survey predicted outcomes in the next. Taken together, these within-person analyses provide an in-depth examination of the impact of relationship comparisons, individually and in their aggregate form, as well as the stability of these effects.
These shorter timeframes (end of day and week), however, provide individuals with little time to change their relationships. Thus, we also examined whether comparisons would predict outcomes 6 months later. Only a few studies have examined the longer-term impact of social comparisons, finding that upward comparisons may be motivating in domains with obvious paths for improvement. For example, college students who made more frequent academic upward comparisons experienced decreased self-evaluations in the short term; however, they reported better adjustment four months later (Taylor et al., 1996). Similarly, high school students who compared themselves with a more successful students tended to perform better academically over time (Blanton et al., 1999). When academic upward comparisons are forced rather than chosen, however, individuals can experience more negative outcomes; students who were regularly exposed to higher-performing peers tended to show more negative academic self-concepts over time (Huguet et al., 2009). Thus, although individuals may benefit from selecting a better-functioning relationship as a model, regular and unsought exposure to superior couples encountered in day-to-day life may be demoralizing over time. In the present research, we examined whether we would observe a lagged comparison effect not only immediately after individuals finished tracking their comparisons (one week later) but also after 6 months. This longer timeframe allows us to explore whether upward comparisons would lead to changes in individuals’ relationship perceptions over time. For example, past studies indicate that the individuals’ positive illusions about their partner in one timeframe contribute to their satisfaction 4 to 5 months later (Murray et al., 1996, 2011).
Dyadic Effects of Relationship Comparisons
The present research also provided an opportunity to examine the dyadic consequences of social comparison. Although past research has examined how social comparisons may have interpersonal consequences, by influencing how individuals see close others and their relationships (Pinkus et al., 2008, 2012; Thai & Lockwood, 2015; Thai et al., 2016, 2019), it remains unclear whether one person’s comparisons can influence another person’s outcomes. Past studies indicate that individuals can accurately detect when their partner perceives them to be more discrepant from their ideals (Campbell et al., 2001) and feel less satisfied when they believe their partner evaluates them negatively (Murray et al., 2000). If a comparer makes numerous upward relationship comparisons, the comparer may come to view the partner and relationship more negatively. The partner, in turn, may sense these changes and experience similar negative outcomes. Although such dyadic effects may not be evident following a single comparison, the cumulative effects of a series of comparisons may have an observable impact. The present study is the first to explore the possibility that social comparison is not simply an individual phenomenon but also has implications for dyads. We predicted that the cumulative impact of upward comparisons, which we expected to have negative outcomes for the comparer, would also be associated with more negative perceptions of the self, partner, and relationship for the comparer’s partner.
Present Study
We conducted a dyadic experience-sampling study, including multiple daily reports and an end-of-day diary, with follow-up surveys 1 week and 6 months later. We predicted that individuals would feel better about themselves, their partner, and their relationship immediately after downward comparisons and worse after upward comparisons (contrast effects). Furthermore, we expected the effects of upward comparisons would last longer than those of downward comparisons (stability effect). We also predicted that more frequent upward comparisons would predict more negative outcomes over time; on days when individuals made more upward comparisons than usual, they would feel less satisfied with and optimistic about their relationships, and less positive about themselves and their partner (within-person effect). Moreover, individuals who made more upward comparisons than others would evaluate their relationship less positively 1 week and 6 months later (between-person effect). Finally, we predicted that comparisons would yield partner effects: When individuals made more upward comparisons, their partners would experience more negative relationship outcomes. Thus, this study provides the first comprehensive analysis of the daily cross-relationship comparisons individuals make, the outcomes of these comparisons in the short term and cumulatively over time, and how these comparisons would be associated with partner outcomes.
Method
Participants
We recruited 82 heterosexual couples (164 individuals) from the community using flyers and ads, who met our inclusion criteria (at least a high school education, own an Android or iPhone, currently dating or married). 1 In total, 149 participants had phones that were compatible with the experience-sampling app (ExperienceSampler; Thai & Page-Gould, 2018). Fifteen participants experienced technical difficulties and completed their surveys through Qualtrics. 2
We had daily surveys from 162 participants (50% women; Mage = 27.04, SD = 7.71; Mrelationship length = 59.91 months, SD = 63.46; 42 dating and 39 married couples) and included data from these participants in our analyses looking at the daily impact of comparisons. 3 On average, participants completed 40.36 experience-sampling surveys (SD = 11.71; Mdnresponse rate = 100%) for a total of 6,539 experience-sampling surveys. Participants completed 6.35 nightly surveys (SD = 1.39; Mdnresponse rate = 100%), for a total of 1,028 daily diaries. Of the original 164 participants, 156 (50% women; Mage = 27.29, SD = 7.71; Mrelationship length = 60.66 months, SD = 64.15) returned for the exit session. Our final sample included 40 dating and 38 married couples. Participants received up to CAD$50 for completing the study.
Six months later, 25 individuals (16 dating and 9 married) and 41 couples (18 dating and 23 married) completed the follow-up survey (N = 107; 63 women) for a CAD$5 gift card. Of these participants, seven had broken up (3 couples, 1 individual). Participants who completed the longitudinal survey did not differ from those who did not on our key intake or exit session measures, ts < 1.08, ps > .28.
Sensitivity analyses revealed that we had at least 80% power to detect the following effect sizes for each set of analyses: r = .035 (experience sampling), r = .07 (daily diary), r = .17 (exit session), and r = .15 (6-month follow-up).
Procedure and Materials
During the in-lab intake session, participants completed our pretest measures. Relationship satisfaction was assessed using 5 items (e.g., “I feel satisfied with our relationship”; Rusbult et al., 1998; α = .89) on a 9-point scale (0 = do not agree at all, 8 = agree completely). For relationship optimism, participants rated the likelihood that positive and negative events would happen to them relative to the typical relationship (e.g., “the love my partner and I share continuing to grow”; Murray & Holmes, 1997) on a 9-point scale (1 = much less likely to occur in my relationship than a typical relationship, 9 = much more likely to occur in my relationship than a typical relationship). The dating version includes 11 events (α = .74), and the married version includes 16 events (α = .85). For self and partner perceptions, participants rated themselves (α = .78) and their partner (α = .78) on 20 traits (e.g., “kind and affectionate”) using a 9-point scale (1 = not at all characteristic, 9 = completely characteristic; Murray et al., 1996). We used these baseline assessments as covariates so that our models would reflect changes in these variables after making comparisons over time (1 week and 6 months later). These baseline assessments were not included in the analyses looking at moment-to-moment or daily fluctuations.
The intake session included a comparisons tutorial; a research assistant provided written examples of upward (“your relationship is less happy than a TV character’s relationship”) and downward (“your relationship is healthier than your friend’s relationship”) comparisons, and advised participants that comparisons are not inherently “good” or “bad” for relationships. The research assistant confirmed that participants could correctly identify cross-relationship comparisons and gave them a handout to help them during the experience-sampling period. Finally, the research assistant helped to install the ExperienceSampler app onto participants’ devices and trained participants on how to use it.
The 7-day experience-sampling portion began the following day. Participants were randomly signaled 6 times throughout the day to determine the prevalence of cross-relationship comparisons (Reis et al., 2014); consequently, partners were not signaled at the same time during the day. Signals were spaced 1.5 to 2 hr apart and were customized to each participant’s schedule: Participants received their first signal about 1 hr after their waking time, with schedules varying for weekdays and weekends.
After each signal, participants indicated whether they had made a relationship comparison. If they did, they completed the comparisons questionnaire. If they did not, they completed a questionnaire regarding their most recent social interaction (e.g., who they interacted with, how they felt during the interaction), included to ensure that participants did not simply indicate that they had made no comparison to avoid answering questions. In the comparisons questionnaire (response options and frequencies are provided in Tables 1–3), participants indicated to whom they compared their relationship, in what domain, 4 in what context the comparison was made, and the direction (−3 = the other relationship was very superior/more successful; +3 = my relationship was very superior/more successful). We included an N/A option to allow participants to indicate that the comparison was qualitative or nonevaluative; participants may have noticed that their relationship differed from another relationship (e.g., we like red wine, they prefer white) but was not better or worse-off. At the end of the survey, participants reported their mood (−3 = very negative; +3 = very positive); optimism (−3 = very pessimistic about my relationship; +3 = very optimistic about my relationship); satisfaction (−3 = less satisfied with my relationship than usual; +3 = more satisfied with my relationship than usual); partner perception (−3 = much worse about my partner than usual; +3 = much better about my partner than usual); and self-perception (−3 = much worse about myself than usual; +3 = much better about myself than usual).
Comparison Domains.
Note. Downward refers to comparisons in which the other relationship is inferior to the participant’s relationship. Upward refers to comparisons in which the other relationship is superior to the participant’s relationship. Lateral refers to comparisons in which the other relationship is similar to the participant’s relationship. Qualitative refers to a comparison that is non-evaluative (e.g., they like white wine, we like red wine).
Comparison Targets.
Note. Downward refers to comparisons in which the other relationship is inferior to the participant’s relationship. Upward refers to comparisons in which the other relationship is superior to the participant’s relationship. Lateral refers to comparisons in which the other relationship is similar to the participant’s relationship. Qualitative refers to a comparison that is non-evaluative (e.g., they like white wine, we like red wine).
Comparison Contexts.
Note. Visual refers to seeing a couple interact from afar (e.g., see a couple interacting while walking down the street) and then comparing with that couple. Brief contact refers to brief interactions (i.e., <10 min) with another couple. Imagined comparison refers to when participants think about a hypothetical relationship and compare their relationship with this imagined relationship. Social interaction refers to any interaction lasting 10 min or longer. Listed below social interaction are the various ways the interaction occurred. Downward refers to comparisons in which the other relationship is inferior to the participant’s relationship. Upward refers to comparisons in which the other relationship is superior to the participant’s relationship. Lateral refers to comparisons in which the other relationship is similar to the participant’s relationship. Qualitative refers to a comparison that is nonevaluative (e.g., they like white wine, we like red wine).
Participants also completed an online survey each night. They rated their satisfaction (3 items; Fletcher et al., 2000; 1 = not at all; 7 = extremely; average daily α = .97), optimism (“How optimistic did you feel about your relationship today?”; 1 = not at all optimistic; 7 = very optimistic), and self-perceptions and partner perceptions (1 = terrible; 7 = terrific). Finally, participants indicated whether they experienced conflict with their partner that day (Yes/No).
Participants returned to the lab to complete an exit questionnaire. They completed the same satisfaction (α = .89), optimism (dating α = .84; married α = .88), self-perception (α = .76), and partner perceptions (α = .81) measures used in the intake questionnaire. We then asked participants if we could contact them in 6 months to complete a follow-up questionnaire.
In the follow-up survey, participants completed a satisfaction measure (e.g., “I am perfectly satisfied with my relationship.”; α = .94; Murray et al., 2003) using a 9-point scale (1 = not at all true, 9 = completely true) and the optimism measure (α = .92) used in our previous surveys, adapted to be applicable to both dating and married individuals. Our syntax (https://osf.io/43fcy/) and materials (https://osf.io/w9qyj/) have been posted to OSF. The data can be obtained by emailing the first author.
Results
Overview
We first conducted analyses at the comparison level, reporting descriptive data regarding comparison characteristics (domain, targets, and context) and examining the direction and effect of individual comparisons. We also conducted lagged analyses to examine whether individuals’ outcomes at one time point predicted making a comparison in the next and mediation analyses to examine the temporal stability of comparison responses. Next, we investigated the shorter-term cumulative effects of comparisons on end-of-day outcomes by examining the temporal effects of comparisons made throughout the day on end-of-day outcomes. We then tested the longer-term cumulative effects of comparisons by examining responses to our 1-week and 6-month follow-up questionnaires; we conducted lagged analyses controlling for baseline measures collected in the intake survey to examine changes in these variables over time after making relationship comparisons. We also examined how the actor’s comparisons were associated with their partner’s outcomes using the actor-partner interdependence model (APIM; Kenny et al., 2006). Marital status (dating vs. married) did not consistently moderate our results. 5
We have argued that upward comparisons would be associated with negative outcomes at the moment and cumulatively at the end of each day, week, and after 6 months. We also considered the alternative possibility that individuals experiencing negative outcomes (e.g., low satisfaction) would be more likely to subsequently make an upward comparison and would also make more frequent upward comparisons each day. We tested this alternative explanation in our experience-sampling and daily diary datasets to determine whether this was a within-person (when individuals feel worse, they make more upward comparisons) or between-person (individuals who feel worse than others make more upward comparisons) effect using lagged analyses. In sum, our design enabled us to assess our prediction that social comparisons would temporally precede outcomes and the alternative possibility that outcomes would temporally precede comparisons.
Individual Comparisons
Comparison characteristics
On average, participants made 1.55 comparisons each day (SD = 2.87) and 10.58 comparisons in a week (SD=13.50; see Figure S1). Comparisons occurred across a variety of domains (Table 1). 6 Individuals were most likely to make comparisons regarding happiness; given that individuals can make quick judgments regarding emotions (Tracy & Robins, 2008), they may be able to evaluate other couples quickly on this dimension. Comparisons regarding compatibility, chemistry, and commitment were less common than those in domains such as happiness, communication, and respect, possibly because couples’ standing on the former dimensions is less observable, at least in public contexts. For example, it is likely easier to assess whether another couple communicates well or appears happy than it is to evaluate their sexual chemistry. Similarly, comparisons about the conflict were relatively uncommon, possibly because couples’ overt public displays of animosity may be relatively infrequent.
Individuals primarily compared their relationships with people they knew (close friend, friend, family member); however, participants also reported comparing to strangers’ relationships, even more than to family members (Table 2). Individuals may encounter strangers more often than family in their daily lives, and so have more opportunities for such comparisons. The majority of comparisons were made in person after seeing or interacting with a couple. Notably, however, social media comparisons were also relatively common, consistent with recent research suggesting that social media use is particularly likely to result in social comparisons (Midgley et al., 2021). Participants also compared with hypothetical relationships often, suggesting that they engage in mental exercises in which they imagined how another relationship might be better or worse than their own (Table 3); indeed, past research indicates that imagined comparisons are relatively common (Wheeler & Miyake, 1992). Finally, participants actively sought out only 12% of the comparisons, and simply happened to notice, or did not intend to make, 74% of the comparisons. 7
Comparison direction
Participants made a downward comparison if the target couple was worse (above the scale midpoint), an upward comparison if the target couple was better (below the scale midpoint), and a lateral comparison if the target was similar (at the scale midpoint, indicating neither better nor worse). In the present study, we focused on the impact of directional (upward and downward) rather than lateral comparisons. Although lateral comparisons may be important as a means of understanding how individuals may choose to affiliate or share experiences with other couples (Wills, 1981), they are less relevant to how one evaluates, improves, and enhances oneself, one’s partner, or one’s relationship (Locke & Nekich, 2000). Moreover, we were interested in assessing how comparisons at the level of the relationship would be similar to comparisons as they have been studied at the level of the individual, with a focus on the impact of upward relative to downward comparisons (Gerber et al., 2018). We found that overall, participants reported 1,132 downward comparisons and 182 upward comparisons. 8 Past lab and experience-sampling studies examining comparisons involving the self only have found that upward comparisons are made more or as often as downward comparisons (Gerber et al., 2018); in contrast, we found that individuals are more likely to make downward comparisons about their romantic relationships.
Reactions to individual comparisons
Analytic strategy
The APIM would have been a good fit with our hypotheses, but participants within a dyad were not signaled at the same time (timing of signals was set to each individual’s schedule); thus, the standard over-time APIM was not appropriate for our analyses regarding the impact of individual comparisons because this model requires time to be crossed with person. We analyzed our data using four-level multilevel models because surveys were nested within days, days within individuals, and individuals within couples. We modeled a random intercept for each dyad, participant, and day, allowing the average amounts of each daily outcome to vary between dyads, individuals, and days (see Supplementary Materials for multilevel equations). We also included a centered version of the signal number to control for the effect of time (Bolger & Laurenceau, 2013). There were no random slopes. 9
We first grand-mean centered our predictors. Because our predictors varied both between and within participants, we created between-person (each individual’s mean across all their daily responses) and within-person (centering each individual’s responses on their own mean) versions of the predictors and entered both into our models.
Comparison direction
We examined whether reactions to upward and downward comparisons differed from reactions when participants made no comparison. To test these hypotheses, we created two dummy-coded variables by recoding our direction variable.
Within-person effects
When participants made an upward comparison relative to no comparison, they reported less positive mood, lower satisfaction and optimism, and less positive self- and partner perceptions (Table 4). In contrast, when participants made a downward comparison relative to no comparison, they reported more positive mood, greater satisfaction and optimism, and more positive self- and partner perceptions.
Multilevel Model Analyses Testing the Between-Person and Within-Person Associations Between Type of Comparison and in the Moment Outcomes.
Note. bs are unstandardized multilevel modeling coefficients, with standard errors appearing in parentheses. For all analyses, we also controlled for time to account for any potential differences due to time. In the square brackets, we report bootstrapped 95% confidence intervals with 5,000 resamples using the percentile method. Effect sizes (r2) are estimated using semi-partial R2 (Edwards et al., 2008).
p < .05. **p < .01. ****p < .001.
Between-person effects
Participants who made more upward comparisons reported less positive mood, lower optimism, and less positive self-perceptions than participants who made fewer upward comparisons (Table 4). In contrast, participants who made more downward comparisons than others reported more positive mood, greater satisfaction and optimism, and more positive self- perceptions and partner perceptions.
Summary
Cross-relationship comparisons predict momentary fluctuations in individual experience. Relative to no comparison, individuals experienced less positive outcomes after making an upward comparison and more positive outcomes after making a downward comparison. Furthermore, individuals who made more upward comparisons than others experienced less positive outcomes in the moment, and individuals who made more downward comparisons than others experienced more positive outcomes in the moment.
Lagged analyses
To examine temporal effects (the effect of a predictor on a future outcome), we used reports from one survey to predict outcomes in the next survey while controlling for the immediate response (outcome measured in an earlier survey). For example, to examine whether current mood predicts making a comparison in the next survey, we controlled for whether participants made a comparison in the current survey. Including the covariate allows us to examine whether the unique effect of mood, above and beyond how individuals felt immediately after making a comparison, predicts making a comparison in the next survey. All results are reported in Supplementary Materials.
To examine stability effects (whether comparison responses persist over time), we conducted mediational analyses to assess how stable response variables were from one survey to the next by including reports of the current outcome as a mediator. For example, we tested whether the feelings of decreased satisfaction that individuals experienced immediately after making an upward comparison persisted, such that they also felt less satisfied 1.5 to 2 hr later (when they completed their next survey; see Figure 1).

Mediation model testing stability of comparison effects.
Alternative explanation
We used logistic multilevel models to examine whether how individuals felt in one survey predicted whether they made an upward or downward comparison in the next. We were unable to test this model for each outcome individually because these models either did not converge or were overfitted. Thus, we averaged across outcomes reported in one survey to create a composite capturing individuals’ current state (α = .90). When individuals felt worse in the moment, they were more likely to report an upward comparison in the next survey (within-person effect), b = −0.27, SE = 0.10, z = −2.77, p = .006. In contrast, when individuals felt better in the moment, they were more likely to report a downward comparison in the next survey (within-person effect), b = 0.11, SE = 0.05, z = 2.33, p = .026. These findings are consistent with past research indicating social comparisons are subject to accessibility and mood congruency effects (Wood et al., 2000): When individuals report more positive mood and self-evaluations, they are more likely to make a subsequent downward comparison, but when they report less positive mood and self-perception, they are more likely to make a subsequent upward comparison.
Stability effects
We also tested the stability of comparison outcomes by examining whether a comparison made in one moment predicted how individuals felt in the next because how individuals felt immediately after making the comparison would carry over to the next moment. We tested this mediational hypothesis for both upward and downward comparisons using the method outlined by Bauer et al. (2006) 10 adapted for a multicategorical predictor (Hayes & Preacher, 2014). To test the significance of our indirect effects, we constructed 95% Monte Carlo confidence intervals with 20,000 resamples (Selig & Preacher, 2008). See Table 5 for results and Figure 1 for conceptual model.
Multilevel Mediation Analyses Testing the Within-Person Mediations of How Relationship Comparisons Predict How Momentary Outcomes Measured Concurrently, Which in Turn Predicts Later Outcomes.
Note. All mediation analyses were conducted using the multivariate multilevel method outlined by Bauer et al. (2006) in R using the lme4 package. Values reported in parentheses are the standard errors from the multivariate models. Values reported in square brackets for a, b, and c′ path are 95% confidence intervals. Values reported in the square bracket for the indirect effect are 95% Monte Carlo confidence intervals with 20,000 resamples using the method outlined by Selig and Preacher (2008). Values reported in square brackets for the total effect are 95% confidence intervals calculated using equations (10) and (12) from Bauer et al. (2006). All analyses also control for time. The unique effect of each type of comparison (i.e., upward vs. baseline and downward vs. baseline) was assessed by entering the other type of comparison as a covariate.
Upward comparisons
For all outcomes, upward comparisons in one moment predicted how individuals felt in the next (significant total effects): Relative to making no comparison, making an upward comparison predicted feeling worse in the next survey. Furthermore, our mediation analyses revealed that making an upward comparison in one moment predicted feeling worse in the next moment as a function of feeling worse immediately after making the upward comparison (significant indirect effects). For all outcomes, there was no direct effect of upward comparison on the next moment’s outcome when the indirect path through concurrent outcome was taken into account. Thus, when individuals made an upward comparison, they had outcomes that were 0.11 to 0.20 (ab’s) units less positive than when they made no comparison, and these negative outcomes persisted. For example, when individuals made an upward comparison, they felt less satisfied immediately afterward; these feelings lasted until the next moment, suggesting that this effect was at least somewhat stable over time.
Thus, there may be a bidirectional effect between the current state and relationship comparisons. When individuals make an upward comparison, they feel worse about themselves, their partner, and their relationship, and these feelings tend to last for at least 1.5 to 2 hr (the interval until the next survey). In turn, these negative evaluations make individuals more susceptible to making another upward comparison in the subsequent survey, which further erodes their evaluations of themselves, their partner, and their relationship. This cycle may in part account for the cumulative effect of upward comparisons.
Downward comparisons
Downward comparisons indirectly impacted future moments through their impact on the present moment. For all outcomes, downward comparisons in one moment did not predict how individuals felt in the next (nonsignificant total effects): There was no difference in terms of future outcomes as a function of whether individuals made a downward or no comparison. One possible explanation for a lack of total effects could be that the positive effects of downward comparisons do not persist into the next survey. However, another explanation for a lack of total effects is the presence of indirect effects with opposite signs (Hayes, 2009). In our case, once we included the current outcome as a mediator, our analyses revealed that making a downward comparison in one moment predicted feeling better immediately after making the comparison, which predicted feeling better in the next survey (significant indirect effects). Thus, when individuals made a downward comparison, they experienced outcomes that were 0.04 to 0.07 (ab’s) units more positive than when they had not made a comparison because of the positive effects of making a downward comparison, and these positive outcomes persisted. These indirect effects, however, were smaller than those observed for upward comparisons. In sum, although responses to both upward and downward comparisons show some stability over time, upward comparisons were associated with a larger change from baseline than downward comparisons and were thus more impactful.
Furthermore, the positive indirect effects of downward comparisons were canceled out by negative direct effects for satisfaction, optimism, and partner perceptions 11 (downward comparisons predicted less positive outcomes after accounting for this indirect effect), resulting in an overall nonsignificant total effect (no difference between downward and no comparison). That is, over time, downward comparisons may backfire in some way, resulting in more negative effects that cancel out the positive stable effects of downward comparisons.
Daily Diary: End-of-Day Effects
Analytic strategy
To examine the cumulative effect of comparisons for each day and account for the nonindependence of observations from the same person and the same couple, we used the over-time APIM (Kenny et al., 2006) using multilevel modeling (Garcia & Kenny, 2018). Dyad members completed daily diaries on the same day; thus, time was fully crossed with the person. APIM allows us to examine actor effects while controlling for partner effects, and vice versa. In our models, we included the number of upward comparisons and the number of downward comparisons made by both the actor and the partner for a given day. We also controlled for the total number of comparisons in these and subsequent analyses to investigate the cumulative effect of the number of upward comparisons and the number of downward comparisons above and beyond the cumulative effect of making any type of comparison (upward, downward, lateral, and qualitative). 12 We separated these predictors into their between- and within-person components using the strategy described earlier and entered both in our models. In addition, we included time (grand-mean centered around the middle of the week) as a fixed effect and a random intercept only. We used this analytic strategy for all continuous outcomes. Because our proposed cause (comparisons made throughout the day) was measured before our outcomes (individuals’ end-of-day perceptions of the self, partner, and relationship), these analyses allow us to examine the temporal effects of relationship comparisons in more general relationship perceptions on a given day. In addition, significant within-person effects in these analyses allow us to understand fluctuations in individual experience above and beyond between-person influences (Bolger & Laurenceau, 2013).
We tested whether effects differed for men and women (distinguishable dyads) by comparing models without gender as a moderator to ones with gender as a moderator using likelihood ratio tests (Gistelinck et al., 2018). These global tests allowed us to simultaneously examine whether actor effects are equal, partner effects are equal, and intercepts are equal for all our predictors and test the null hypothesis that dyads are indistinguishable for each outcome. For all outcomes, the tests were significant, and dyads were distinguishable.
Actor effects
On days when individuals made more upward comparisons than usual, they felt less satisfied and optimistic, and less positive about themselves and their partner at the end of the day (within-person effect; Table 6). For satisfaction and optimism, gender moderated these effects: On days when women made more upward comparisons than usual, they were less satisfied and optimistic at the end of the day. This effect was nonsignificant for men. Furthermore, individuals who made more upward comparisons on average felt less satisfied, optimistic, and positive about themselves and their partner (between-person effect).
Actor Partner Interdependence Model Over Time Testing the Within-Person (Top Rows) and Between-Person (Bottom Rows) Associations Between Number of Upward Comparisons and Number of Downward Comparisons Made on Each Day on End of Day Relationship Outcomes.
Note. Bs are unstandardized multilevel modeling coefficients, with standard errors appearing in parentheses. For all analyses, we also controlled for diary day to account for any potential differences due to time. In the square brackets, we report 95% confidence intervals. Effect sizes (r) were estimated using the t to r formula (rD=t/√[df + t2]) and then adjusted using the correlation between actor and partner (rI) and the formula (r= rD√[1 − rI]; Kenny et al., 2006).
p < .05. **p < .01. ***p < .001.
Partner effects
On days when their partner made more upward comparisons than usual, individuals felt less satisfied and optimistic, and less positive about themselves and their partner at the end of the day (within-person effect; Table 6). For partner perceptions, the between-person upward comparison effect was also significant: Individuals with partners who made more upward comparisons than other participants felt worse about their partner.
Conflict
On average, participants reported 1.22 conflicts (SD=1.30; see Table S2). Because conflict is a binomial variable, we used a logistic multilevel model, adapted for APIM, and treated dyads as distinguishable (Loeys et al., 2014). We estimated a cross-classified model, nesting conflict within dyads and day simultaneously because time was crossed with the person. There were no random slopes.
On days when actors made more upward comparisons than usual, they were more likely to experience conflict (within-person actor effect; Table 6). Individuals who made more upward comparisons than others were also more likely to experience conflict (between-person actor effect). On days when partners made more upward comparisons, individuals were more likely to experience conflict (within-person partner effect). 13
Lagged analyses
Using the over-time APIM, we examined whether reports from one day predicted outcomes the next day while controlling for current reports of that outcome. All results are reported in Supplementary Materials.
Alternative explanation
We used Poisson cross-classified multilevel models to test the alternative possibility that how individuals feel about their relationship, their partner, and themselves at the end of a given day would predict the number of comparisons they made the next day, controlling for the number of comparisons they made that day. For brevity, we report results for predictors that were significant for multiple models only.
For upward comparisons, within-person actor effects were inconsistent. Between-person effects of self-evaluation, partner evaluation, and satisfaction were significant, zs > 2.02, ps < .044: Individuals who felt worse about themselves, their partner, and their relationship (averaged over 1 week) than other participants made more upward comparisons the following day.
For downward comparisons, within-person actor effects were inconsistent. No between-person effects were significant, zs < 1.94, ps > .05.
In addition, all but one within- and between-person partner effects were nonsignificant for both the number of upward comparisons and number of downward comparisons made the next day, zs < 1.95, ps > .05. 14 Thus, it is unlikely that having partners who are less happy on a given day or on average leads individuals to make more comparisons the following day.
Temporal analyses
We also examined whether the number of comparisons made on a given day predicted future outcomes (how individuals felt about their relationship, their partner, and themselves the following day) while controlling for how they felt at the end of that given day (immediate outcomes) using person-centered predictors only (within-person effects only). Within-person actor effects of upward or downward comparisons were nonsignificant, ts < 1.43 ps > .152. Thus, the number of comparisons individuals made on a given day did not predict how they felt at the end of the next day.
Summary
Taken together, these results suggest individuals’ daily outcomes are influenced by both their own and their partner’s upward, but not downward, comparisons. On days when actors or their partners made more upward comparisons than they usually did (within-person effect), they felt less satisfied and optimistic, and less positive about themselves and their partner. They were also more likely to experience conflict that day. Furthermore, actors who made more upward comparisons than others did (between-person effect) also felt less satisfied and optimistic, and less positive about themselves and their partner at the end of the day; they were also more likely to experience conflict. Individuals in a relationship with a partner who made more upward comparisons than others felt worse about the partner. Thus, upward comparisons have an impact not only on the comparer but also on the comparer’s partner.
Furthermore, our lagged analyses suggest that participants who were less satisfied with themselves, their partner, and their relationship than other participants were more likely to make upward comparisons the following day; it may be that individuals who are dissatisfied with their relationships view more relationships as superior to their own on subsequent days. However, we found no consistent evidence from our within-participant analyses to support the alternative hypothesis that how individuals feel at the end of a given day predicts the number of upward or downward comparisons they make the following day. We also found no evidence suggesting that the number of comparisons individuals make on a given day predicts how they will feel at the end of the next.
Exit Session: End-of-Week Effects
Analytic strategy
To test the cumulative effects of upward and downward comparisons participants made over 1 week on relationship quality, we used APIM to conduct lagged analyses while controlling for the total number of comparisons, and the actor’s and partner’s baseline assessment of each outcome (Time 1). Dyads were distinguishable for optimism only.
Actor effects
Individuals who made more upward comparisons during the week felt less satisfied and optimistic, and viewed their partner less positively (Table 7). The upward comparison effect for optimism, however, was moderated by gender: For women, the more upward comparisons they made in 1 week, the less optimistic they felt. This effect was nonsignificant for men. Unexpectedly, there was no upward comparison effect on self-perceptions.
Actor Partner Interdependence Model Testing the Associations Between Number of Upward Relationship Comparisons, and Downward Relationship Comparisons Made in 1 Week, While Controlling for Total Number of Relationship Comparisons and Outcome Measured at Time 1, on Various Relationship Outcomes.
Note. Bs are unstandardized multilevel modeling coefficients, with standard errors appearing in parentheses. For all analyses, we also controlled for the outcome measured at intake. In the square brackets, we report 95% confidence intervals. Moderation by sex is only reported for outcomes where dyads are distinguishable. Effect sizes (r) were estimated using the t to r formula (rD = t/√[df + t2]) and then adjusted using the correlation between actor and partner (rI) and the formula (r = rD√[1 − rI]; Kenny et al., 2006).
p < .05. **p < .01. ***p < .001.
For partner perception, the downward comparison effect was significant (Table 7): Unexpectedly, individuals who made more downward comparisons perceived their partner more negatively. No other downward comparison effects were significant.
Partner effects
For all outcomes, no comparison effects were significant (Table 7).
Summary
Participants who made more upward comparisons felt less satisfied, optimistic, and positive about their partner at the end of the week than participants who made fewer upward comparisons. Unexpectedly, the number of upward comparisons did not predict self-perceptions. Unlike end-of-day effects, we did not find end-of-week partner effects. Surprisingly, we found that actors who made more downward comparisons had less positive partner perceptions; no other downward effects were significant. Thus, individuals appear to be sensitive to upward comparisons, with more such comparisons in the course of a week being associated with more negative outcomes; partner’s outcomes, however, do not appear to be associated with the actor’s comparison behaviors.
Six-Month Follow-Up: Longitudinal Effects
We tested whether the comparisons participants made in 1 week were associated with satisfaction and optimism 6 months later using the same analytic strategy used for our 1-week analyses. Dyads were indistinguishable for both outcomes, χ2(10)s < 17.70, ps > .06.
As predicted, individuals who made more upward comparisons during the week we studied them were less satisfied 6 months later (actor effect), b = −0.17, SE = 0.06, t(91) = −2.75, p = .007, r = .14. No other comparison effects were significant, ts < 1.30, ps > .19.
For optimism, there were no significant comparison effects, ts < 1.04, ps > .30.
Summary
Consistent with the end-of-week effects, individuals’ outcomes were associated with their own, but not their partner’s upward comparison activity over 1 week: They felt less satisfied, but not less optimistic, 6 months later if they made more upward comparisons during the experience-sampling period. There were no longitudinal effects of downward comparisons.
Discussion
Individuals will inevitably encounter examples of relationships that seem better or worse than their own. The present study is the first to examine these comparisons through experience-sampling methodology, providing new evidence about the frequency, direction, and effect of cross-relationship comparisons. We demonstrated that individuals make regular comparisons with other couples and feel worse about their relationship, partner, and themselves immediately after an upward comparison relative to no comparison or a downward comparison. Moreover, this research is the first to demonstrate that these comparisons have a cumulative effect; after encountering superior relationships, individuals feel worse about their own relationships not only in the moment, but these upward comparisons, in their aggregate form, are associated with negative outcomes at the end of each day, at the end of each week, and even 6 months later. The present research is also the first to demonstrate that the upward comparisons that one makes throughout the day are negatively associated with one’s partner’s satisfaction with and optimism about the relationship, although this effect appears to be limited to comparisons over the course of 1 day rather than over a longer time span. Thus, cross-relationship comparisons have a far-reaching impact, predicting both immediate and longer-term relationship perceptions and affecting both self and partner.
The present study highlights the importance of comparisons with superior relationships in particular. Although downward comparisons were more common, upward comparisons were more impactful over time. Specifically, when individuals made more upward comparisons on a given day, they were less satisfied and also reported experiencing conflict; furthermore, their partners were also less satisfied and optimistic, suggesting that partners were attuned to the comparer’s negative reactions to superior couples. In addition, the number of upward, but not downward, comparisons during the experience-sampling phase predicted satisfaction 6 months later. It may be that because most individuals assume that their relationship is superior to other couples (Rusbult et al., 2000), downward comparisons are unremarkable on any given day; they simply confirm individuals’ assumptions regarding their perceived superiority but do not provide any lasting benefits to their relationship, similar to downward comparisons made at the individual level (Taylor et al., 1996). Upward comparisons, in contrast, are relatively infrequent and unexpected negative events that do not fit with one’s perceptions of one’s own relationship superiority and so have a more disruptive effect.
It is also possible, however, that upward comparisons are simply more common among couples experiencing distress. Individuals who are less satisfied with and optimistic about their relationships may be more likely to see their relationships as inferior to others and so report more upward comparisons. Indeed, our analyses revealed that individuals experiencing negative outcomes in one survey were more likely to make comparisons with superior couples in the next. This effect, however, appears to be bidirectional, in that upward comparisons also predicted more negative outcomes in the next survey because the immediate negative outcomes were stable over time. We note that in the present research, moreover, we controlled for initial satisfaction and optimism in our analyses examining changes over 1 week and 6 months and focused on within-person effects of upward comparisons in our experience-sampling and daily-diary effects, which suggests that upward comparisons are associated with a negative impact that goes beyond that individual’s or couple’s norms. In addition, our daily diary allowed us to examine the temporal effects of comparisons made on a given day. We found that when individuals made more upward comparisons earlier in the day, they reported worse outcomes at the end of the day (within-person effects), providing further support for our proposed causal pathway. Finally, our mediational analyses provide additional support for our hypothesis that upward comparisons have a cumulative effect. We found that when individuals made an upward comparison in one moment, they continued to feel worse about themselves, their partner, and their relationship up to 2 hr later, suggesting that the effects of upward comparisons tend to linger. Moreover, the present findings are consistent with a large number of studies on comparisons between individuals that have used experimental methods to demonstrate a negative causal impact of upward comparisons on the self (Gerber et al., 2018). Indeed, past experimental research on cross-relationship comparisons more specifically suggests that upward comparisons decrease satisfaction (Morry & Sucharyna, 2019). Thus, it is unlikely that our results can be explained solely by the possibility that relationship dissatisfaction causes upward comparisons; rather, it seems most likely that, regardless of initial satisfaction levels, making more upward comparisons poses a threat to individuals’—and their partners’—relationship perceptions, and that for individuals, this negative effect persists over time.
Past research suggests that individuals will not always be negatively affected by upward relationship comparisons. Indeed, we found that highly committed individuals responded to a highly successful relationship example positively, viewing this couple as an exemplar of what they might become in the future (Thai et al., 2020). When asked to provide an example, many participants described how they were inspired by their parents. In these studies, however, participants were asked to consider one example of a relationship that was more successful in general terms. When individuals are faced with an example of a relationship superior to their own on a specific dimension, they may find it more difficult to construe their relationship as similar. Indeed, past research has found that when couples were aware of their standing in a specific domain (e.g., communication), they had difficulty construing themselves as superior to other couples and were more accurate in determining their standing relative to other couples (Frye & Karney, 2002). Although we asked participants to indicate the domain of comparison and attempted to assess the effect of domain-specific rather than holistic comparisons, we note that we did not ask participants to rate the specificity of the comparison domain. Indeed, it may be that some of the domains we included (e.g., happiness) were construed by participants as representing global rather than specific domains. Future research should assess more directly whether the specificity of the comparison domain limits individuals’ ability to construe their relationship as similar to a superior couple and thus impacts comparison outcomes. It will also be important to examine how different research contexts may influence comparison outcomes: When encountering superior others in the course of their busy daily lives rather than in a controlled lab context, individuals may find it difficult to engage in the cognitive processing needed to draw parallels with their own relationship and so to find the exemplar to be more inspirational. In daily life, individuals make more unintended than motivated comparisons (Wood et al., 2000). Indeed, in the present research, individuals reported making comparisons that were largely unintended rather than chosen. That is not to say that individuals will never be positively affected by a superior couple; rather, the aggregate response to many such comparisons may, over time, be negative. One superior relationship may be inspiring; a series of superior relationships may be more worrisome, indicating potential flaws in one’s own relationship.
Although we found primarily positive or no effects for downward comparisons, one of our findings suggests that downward comparisons may backfire over time. After 1 week, participants who made more downward comparisons perceived their partner less positively. Recent research suggests that downward comparisons are associated with investing less effort into domains in which one is superior (Diel et al., 2021). Moreover, partners tend to make similar types of comparisons (see Supplementary Materials): If Partner A makes more downward comparisons, Partner B is also more likely to make downward comparisons. Consequently, individuals may perceive their partner more negatively after making more downward comparisons because they notice that their partner is putting less effort into their relationship over time, which may also result in declines in relationship quality. This is consistent with other research indicating that incoming college students who made more downward comparisons had a better psychological adjustment in the short term, but were more poorly adjusted after 4 months (Taylor et al., 1996). For these students, the downward comparisons indicated that they were well-adjusted and did not need to put more effort into improving themselves. Future research should explore how relationship comparisons influence relationship maintenance behaviors and the potential adverse longer term effects of downward comparisons.
We did not expect gender differences; however, we allowed for this possibility by recruiting heterosexual couples only. We note that, over the course of the week, women made more upward comparisons than men, and men made more downward comparisons than women. This difference, however, may be due to how equitable individuals perceive their relationship to be: Heterosexual women and some gay men tend to perceive their marriage as less equitable in terms of household duties, whereas lesbian couples perceive their marriage to be more equitable (Shechory & Ziv, 2007). To explore this possibility, future research should recruit a more inclusive sample. It will also be interesting to examine whether—and why—women may perceive their relationships to be less superior than men do. In addition, while women who made more upward comparisons were less optimistic after 1 week, this effect was not significant for men. Given that gender did not moderate the results for satisfaction, self-perception, and partner perceptions, however, it appears that the effect of cross-relationship comparisons is, for the most part, similar for women and men. We found no consistent evidence suggesting that women and men experienced different reactions when encountering relationships superior or inferior to their own.
One of the most significant contributions of this article to the social comparison literature is the finding that comparisons of a dyad to another dyad may have implications not only for the comparer but also for the comparer’s partner. Past research on social comparison has focused on the impact of comparisons on the self, including one’s affect, self-esteem, and motivation. The finding that one can make a comparison—an intrapsychic cognitive process—that may also affect the partner indicates that comparisons can have a greater reach than has previously been recognized in the literature. In future research, it will be important to understand the mechanism through which one’s own reactions to the comparisons are experienced by the partner. It may be that when individuals compare to a superior relationship, they feel worse about themselves; because the partner incorporates the comparer into their own identity (Aron et al., 1991), this threat to the comparer’s self is also experienced by the partner as a threat to the self. Alternatively, it may be that individuals respond to an upward comparison by holding the partner at least somewhat responsible for their shared relationship inferiority; indeed, individuals who made an upward relationship comparison reported more negative partner perceptions in the moment and also at the end of the day. Partners may in turn be attuned to this change, and feel worse about themselves and their relationship as a result. It is also possible that individuals with less happy partners make more upward comparisons. Although our lagged analyses do not provide strong support for this alternative causal pathway, further research will be necessary to establish the causal direction of the association between individuals’ comparison behavior and their partner’s outcomes.
On a daily basis, individuals in romantic relationships must navigate a series of potential threats, including coping with conflict and resisting attractive alternatives. The present research indicates that upward comparisons with other relationships may also pose a threat, chipping away at relationship satisfaction. When other couples seem happier, more loving, or more fun, one may come to question the quality of one’s own relationship. Moreover, numerous downward comparisons indicating they are superior to other relationships do little to counteract this threat. Fortunately, the experience-sampling data suggest that these comparisons with superior couples occur relatively infrequently, and are far outnumbered by comparisons with inferior couples. Nevertheless, given their effects on individuals in the short and longer term, it will be important to understand how people deal with the threat posed by upward relationship comparisons.
Supplemental Material
sj-docx-1-psp-10.1177_01461672211056774 – Supplemental material for The Ups and Downs of Being Us: Cross-Relationship Comparisons in Daily Life
Supplemental material, sj-docx-1-psp-10.1177_01461672211056774 for The Ups and Downs of Being Us: Cross-Relationship Comparisons in Daily Life by Sabrina Thai, Penelope Lockwood and Elizabeth Page-Gould in Personality and Social Psychology Bulletin
Supplemental Material
sj-docx-2-psp-10.1177_01461672211056774 – Supplemental material for The Ups and Downs of Being Us: Cross-Relationship Comparisons in Daily Life
Supplemental material, sj-docx-2-psp-10.1177_01461672211056774 for The Ups and Downs of Being Us: Cross-Relationship Comparisons in Daily Life by Sabrina Thai, Penelope Lockwood and Elizabeth Page-Gould in Personality and Social Psychology Bulletin
Footnotes
Acknowledgements
We are grateful to Jessica Maxwell, Emilie Auger, and CHan for their comments on an earlier draft of this paper. We thank Matthew Ng, Michelle Hu, Aliya Aneja, and Shayne Sanscartier for their assistance with data collection.
Authors’ Note
Partial report of these data was presented at the 2020 Meeting of the Society for Personality and Social Psychology in New Orleans, Louisiana, USA, and the 2021 Virtual Meeting of the Society for Personality and Social Psychology.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by grants from the Social Sciences and Humanities Research Council (SSHRC) of Canada awarded to Penelope Lockwood and Elizabeth Page-Gould. Sabrina Thai was supported by a Joseph Armand Bombardier CGS Doctoral Scholarship and a Postdoctoral Fellowship from SSHRC.
Supplemental Material
Supplemental material is available online with this article.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
