Introduction

Social comparison describes the process of evaluating oneself in relation to others, in domains such as abilities, status, and performance [1]. This process can provide information about one’s current status and guidance for how to achieve specific goals [2]. Yet, it is perhaps most known for its potential negative consequences; because it can also highlight one’s perceived inferiority relative to others and inspire negative emotions such as jealousy, it is often characterized in idioms such as “comparisons are odious” and “comparisons are the thief of joy.” One contributor to this negative reputation is the ubiquity of social media and other digital tools that facilitate social networking. These platforms present many opportunities for social comparison: users tend to post about only the positive aspects of their lives and thereby, present idealized views of their experiences [3].

Indeed, some studies on this topic show that social comparisons are associated with negative mental health outcomes. For example, McCarthy and Morina’s [4•] review found a positive association between self-reported or experimentally induced social comparison and depression and anxiety in clinical populations. Social comparison orientation, which describes an individual’s reported perception of their own tendency to compare themselves with others, has often been identified as a cross-sectional correlate of such outcomes [5] and as a risk factor for the development of clinically significant depressive symptoms [6] and symptoms of disordered eating [7] among women ages 12–24. Reports of social comparisons made on social media have also been correlated with negative mental health outcomes such as poor subjective well-being [8], and much has been made of the observation that increased rates of depression and self-harm have corresponded with increases in technology use [9]—particularly among young women [10].

Such evidence has led to public health recommendations to limit social media and other socially networked screen time (as a way to reduce comparison) [11] and to unhelpful lay and expert advice to “stop making comparisons.” The latter is particularly problematic. As comparisons occur automatically in response to information about others’ status or performance [12], advice to not make comparisons is as useful as any other advice to avoid certain thoughts: they only make these thoughts more likely to occur [13] and breed frustration with the resulting inability to avoid them. A more useful approach involves cognitive restructuring (cf. cognitive therapy), whereby the comparer is taught to evaluate comparative thoughts by weighing the evidence for and against its accuracy [14]. This typically results in recognition that comparative self-evaluations are inaccurate—or at least, not the absolute or only truth—and leads to reductions in negative affect.

Evidence consistent with this idea shows that the association between reported comparisons and negative affect is lower among individuals with more effective cognitive appraisal skills [15]. In intervention work, techniques such as written emotional disclosure that encourage cognitive processing or reappraisal of thoughts and feelings about a topic can show incidental effects on reported comparisons, which mediates the effect of the intervention on mental health outcomes (e.g., sleep quality, disordered eating behavior) [16]. Yet, social comparison is rarely the particular focus of treatment research; it is typically one of several types of maladaptive thoughts that can be addressed with cognitive intervention, either directly or indirectly [17]. Attempts to directly change negative comparisons by replacing them with positive comparisons have been unsuccessful [18], though meta-analysis shows that cognitive techniques can reduce reported comparison activity in the context of interventions to reduce body satisfaction [19].

Does Social Comparison Warrant Treatment?

Along with evidence for social comparison’s associations with negative mental health outcomes, there is also evidence that social comparison is unrelated to mental health, or has positive associations with mental health outcomes (in general or specific to social media). Results from a recent umbrella review showed that although there is some evidence of a negative association between social media use (and associated comparison opportunities) and adolescent mental health, the literature predominantly suggests weak or inconsistent associations [20]. Interestingly, recent evidence suggests that social media use (and associated comparison opportunities) is one of the least powerful influences on adolescent mental health, as opposed to experiences such as bullying or insufficient family support [21]. Further, in a study conducted during the COVID-19 pandemic, social comparison specific to social media was positively associated with concurrent depressive symptoms, but predicted lower levels of loneliness and anxiety two to three weeks later [22•]. Similarly, time spent on Instagram (and associated comparison opportunities) was not related to psychological well-being, though using Instagram for social interaction was related to an individual’s general positive emotions (e.g., happiness, joy) and flourishing (e.g., “I lead a purposeful and meaningful life”), and using Instagram for browsing was related to positive emotions [23].

Critically, not all comparisons are created equal, and differences between comparison experiences could help to explain equivocal findings. For example, comparisons of abilities (or behaviors) on social media are linked to negative cognitive experiences such as rumination, whereas comparisons of opinions are linked to positive cognitive experiences such as reflection [24]. In addition, we can compare ourselves to others we perceive as doing better than we are in a given domain (upward comparison), as doing worse than we are (downward comparison), or doing similarly (lateral comparison) [25•]. Upward comparisons have been identified as particularly problematic: comparisons to those perceived to be better off are the ones that can highlight the comparer’s relative inferiority and inspire negative emotions such as envy as well as low self-esteem [26] and rumination [24]. Even when we have a range of options to choose from, many people approach comparison targets that are associated with negative outcomes (e.g., upward targets) [27,28,29].

Yet, even upward comparisons can be associated with benefits such as improved well-being, and correlations between social comparison and mental health outcomes show considerable heterogeneity between people [30]. Multiple theoretical models have proposed that people either identify with or contrast themselves against comparison targets, focusing on perceived similarities versus differences (respectively) [31, 32]. Evidence supporting these models shows that contrasting against upward targets and identifying with downward targets result in unfavorable self-evaluations, which are associated with negative experiences such as anxiety, discouragement, and stress [15, 33]. Conversely, identifying with upward targets and contrasting against downward targets result in favorable self-evaluations, which are associated with positive experiences such as hope, inspiration, and confidence in one’s ability to achieve desired goals [34]. As a result, concluding that comparisons are generally bad (or good) for mental health is problematic, particularly in the context of mental health treatment.

Weaknesses in Existing Work

Across comparison features and mental health outcomes of interest, much of our existing evidence relies on a narrow set of research designs that have considerable methodological limitations (see Table 1). First, directional (and often, causal) conclusions about the association are routinely drawn from cross-sectional designs that involve only retrospective self-report [5]. Available measures ask respondents to report on their comparisons aggregated over long periods—or worse, over unspecified periods [35]—to determine the extent to which they believe they are a person who makes comparisons. As noted, social comparison occurs automatically and often without conscious processing, and it carries stigma. Consequently, global, retrospective assessments are subject to a range of recognition problems, recall biases, and efforts to maximize social desirability. Further, similar comparison opportunities could have distinct effects on the same person, at different times and/or in different contexts—possibly due to within-person variability in identification and/or contrast processes [36••, 37]. These nuances are essential to our understanding of the role of comparisons in mental health outcomes, but cannot be detected with global, retrospective self-report.

Table 1 Summary of strengths and weaknesses of traditional study designs used to link social comparison to mental health outcomes
Full size table

In addition, that global assessments of social comparison are associated with negative mental health outcomes such as anxiety and depressive symptoms indicates only that people who report “more” social comparison (or who more strongly identify as a person who makes comparisons) have worse outcomes. It is likely that experiencing these mental health symptoms also increases sensitivity to and/or the likelihood of making comparisons that have negative consequences. Although the experience of negative-outcome comparisons may reciprocally contribute to negative mental health outcomes, this does not indicate that comparisons cause these outcomes. Longitudinal studies conducted over long periods represent a step toward establishing directionality. However, existing studies suffer from similar measurement issues (e.g., using retrospection over long periods) and have rarely controlled for mental health symptoms concurrent with social comparison at baseline.

Experimental designs do allow for drawing directional and causal conclusions, as well as for mapping mechanistic pathways that have direct implications for treatment. With respect to studying the effects of social comparison, however, experimental designs carry their own limitations: each participant typically experiences only one of a limited range of comparisons and the outcomes assessed are extremely short term (e.g., affect or behavioral intentions immediately after comparison exposure) [29]. These designs have high internal validity but limited external validity, and outcomes represent transient experiences rather than “mental health” as typically conceptualized. Such limitations dampen the immediate utility of traditional experiments for informing mental health treatment, though innovation in this area holds promise, as outlined below.

In sum, social comparison is a complex phenomenon that includes significant measurement challenges. Consequently, unraveling the link(s) between social comparison and mental health outcomes—and identifying options for treatment, if appropriate—requires innovative and sophisticated research designs.

Innovative Work Linking Social Comparison to Mental Health Outcome Improvements in Observation

As an initial step toward mapping the complexity of social comparison, intensive ambulatory assessment designs use technologies such as text messages and smartphone applications to access experiences of comparison and its correlates as they unfold in daily life. This involves people reporting on comparison and other experiences of interest one or more times per day, focusing on what happened today, in the past few hours, or what is happening at the time of the report [36••]. As a result, this approach substantially reduces (if not eliminates) retrospective recall bias and allows for understanding what happens when we make comparisons, rather than merely who makes them. Consequently, strengths of this approach are the ability to determine within-person variability in comparison (and its contextual predictors) as well as the temporal sequencing of comparisons and experiences such as positive and negative affect as they occur (to determine the direction of potential effects; see Table 2) [38, 39•].

Table 2 Summary of recommendations for future work to understand associations between social comparison and mental health outcomes
Full size table

For example, event-contingent experience sampling asks participants to complete a self-report each time they recognize that they have made a comparison. Reports indicate that comparisons occur on a wide range of dimensions, in various directions, approximately 24 times over 13 days [40]. Ecological momentary assessment (EMA) using signal-contingent reporting (i.e., participants report on comparisons multiple times per day, in response to signals or prompts to do so) is an increasingly popular approach, particularly in the domain of body satisfaction and disordered eating behavior. These studies, which have typically focused on the comparison experiences of young women, show that upward social comparisons of appearance are associated with negative cognitive reactions (specifically body dissatisfaction) and behavioral responses such as binge eating, body checking, and restriction of food intake [41]. Unfortunately, existing findings of this nature are highly confounded by pre-existing body concerns (as associations are strongest among those with more severe body dissatisfaction prior to intensive assessment [42•]) and often draw incorrect conclusions about “increases” or “decreases” in comparisons (in the absence of a true baseline as the referent [43]). However, such studies illustrate the potential for EMA and other intensive ambulatory assessment methods to elucidate the nature of associations between comparisons and mental health outcomes as they occur in people’s natural environments.

Of note, many of the mental health outcomes of interest represent constellations of symptoms that occur over weeks or months (e.g., depression). Daily or momentary reports of a subset of these experiences are not typically intended to operationalize the syndrome or diagnosis, but can reveal details about how people experience these clinical outcomes in daily life. For instance, EMA of suicidal ideation and its predictors (e.g., hopelessness) shows that the intensity of these experiences varies considerably within person over 28 days [44], even among those who are at high risk for suicide [45]. EMA allows for both tracking the (in)stability of affective states and the specific emotions associated with high instability (e.g., hostility, fear, sadness) [46, 47•]. EMA of depressive symptoms has also shown associations with depression scores obtained from in-person assessments and has validity for tracking daily symptomatic changes in depression [48]. EMA shows good adherence, even in populations with affective disorders, indicating that the associated participant burden is manageable [49, 50]. Evidence also shows that single-item measures of symptoms are valid, which could be particularly useful with populations that have mental health concerns that make it difficult to respond to multi-item surveys [51].

With respect to social comparisons, EMA can also reveal the affective states that accompany comparisons [36••], to elucidate the temporal sequencing and timing of comparisons and relevant emotional or mental health experiences. For example, in a study of individuals with eating disorders, EMA showed that negative cognitive and emotional experiences were reported concurrently with social comparisons based on appearance [42•]. Further, when EMA has been examined in contrast with traditional pencil and paper responses, EMA has captured significant post-test changes that methods without the same time sensitivity overlook [52••]. As a result, EMA may be particularly useful for detecting changes in affect associated with depressive, affective, and personality disorders. This could improve the evaluation of interventions in these populations (relative to single assessments over longer periods), by detecting whether interventions change patterns of cognitive/emotional experiences (e.g., comparisons, affective states) or their within-person associations in daily life [52••].

This approach can also help to clarify causal and mechanistic models to be tested with other methods. For example, EMA allows us to move beyond questions of what happens in daily life, to determine how we get from naturally occurring opportunities (e.g., for social comparisons) to mental health outcomes (e.g., depression). As noted, researchers have proposed that it is not the occurrence of particular comparisons that result in positive or negative affect; it is how we interpret the information gleaned from a given comparison that determines how we respond emotionally (i.e., identification vs. contrast) [31, 32]. This is a promising model for several reasons. First, evidence consistently supports its validity with multi-item, between-person reports (global and retrospective) [34]. Although the full model has not yet been validated using temporally sensitive within-person methods, formative work shows that using single items in intensive ambulatory studies that use patient mobile phones for assessment is feasible and shows concurrent validity among adults with schizophrenia [53•].

Second, this approach provides logical, testable, and potentially modifiable pathways between comparisons and immediate experiences that can contribute to and/or maintain mental health conditions. For example, identification and contrast offer specific targets for cognitive restructuring, helping patients to target the precise link in the cognitive chain that can be addressed with alternative interpretations. This may enhance the effectiveness of cognitive approaches to treatment for anxiety, depression, and other mental health conditions where problematic comparisons are prevalent, particularly if EMA or similar methods can be used to help patients reflect on their comparisons in the moment. Importantly, however, we first need to validate identification and contrast processes within person, to verify these processes as determinants of response when comparisons occur (not just for whom).

Novel and Creative Approaches to Experimental Manipulation

More broadly, there is a critical lack of evidence indicating a causal link between comparisons and relevant outcomes. Recent work shows advances in this area using novel and/or creative approaches to experimental design, particularly with respect to social media and associated opportunities for comparisons (see Table 2). For example, a straightforward first step would be to establish a causal or dose–response association, via random assignment to social media exposure (vs. none) or to specific doses of exposure (such as hours per day, days per week, or particular platforms). Yet, prior or typical exposures are challenging confounds to eliminate, and ensuring adherence to a randomly assigned condition is difficult and costly, if it is possible at all.

To address these problems, researchers have used reduction methods: participants are asked to reduce their exposure by specified amounts from what is typical for them, and assessments of mental health outcomes are conducted before and after the reduction period. These studies showed a decrease in disordered eating symptoms among those who did not use social media for one week (vs. continued use as usual) [54] and decreased loneliness and depression among those who reduced social media use to 10 min per platform per day (vs. continued use as usual) [55••]. Although this method does not definitively identify social media or related opportunities for comparison as the root cause(s) of mental health problems, it provides compelling experimental evidence that reductions in or removal of exposure has benefits. As social media exposure can lead to poor mental health in more ways than via social comparison (e.g., bullying, lack of desired response from others), an important next step would be to isolate comparisons in such work, to determine its unique contributions.

Other researchers have used experimental work to establish social comparison as a mediator of the cross-sectional association between social media use and mental health outcomes. For example, Samra et al. [56] randomly assigned groups of college students to view different comparison targets and assessed their reactions and affect, as well as self-reported social media use and social comparison behaviors, showing that fixating on upward comparisons and negatively comparing oneself to others on social media mediates the cross-sectional relation between problematic social media use and depressive symptoms. Examining social comparison response experimentally is a strength and an advancement in this area. As discussed, however, cross-sectional assessment of predictors and outcomes of interest is highly limited; depressive symptoms could be the driver of social media use and still show that social comparison response mediates this association.

In our view, research to investigate the potential causal pathway(s) between social comparison and mental health would be most compelling if it capitalized on the combined strengths of ambulatory, within-person assessment and experimental methods. One step in this direction asked college students to report on their comparisons via social media once per day for 14 days; midway through this period, those randomized to a “social savoring” intervention viewed a video about savoring as an alternative to comparison [57•]. Relative to those who did not watch the video, those who did reported better self-esteem and reported fewer comparisons on post-intervention days when they engaged in more social savoring. This offers compelling preliminary evidence that brief interventions may influence the impact of negative social comparisons on mental health. As there was no active control for the intervention in this study, and as both groups decreased their comparisons in the last 7 days of assessment, it is possible that merely raising awareness of comparisons via intensive assessment changed the process or its consequences. Further research is needed to fully identify the potential of such substitution approaches and the duration of their effects.

In addition, there is great promise in the use of multiple randomizations for the same person, across distinct assessments, to determine how and under what circumstances different types of comparisons lead to positive versus negative outcomes [58, 59]. This approach would allow for drawing strong causal conclusions about what happens when we make comparisons in daily life, improving our ability to identify and test (1) mechanistic pathways in the real world, and (2) effective intervention approaches. It could also help us to determine whether there are in fact person-level characteristics that differentiate those who use comparisons in problematic ways (suggesting that personalization or person-level tailoring of interventions might be useful), versus context-specific determinants that vary within person (suggesting that contextual adaptation would be more effective).

Personalization, Adaptation, and Tailored, Technology-Supported Intervention

It is in the areas of personalization and adaptation that advances in technology have afforded the most meaningful insights into social comparison’s link to mental (and other) health outcomes. Initial steps in this area have assumed that there is an optimal type of comparison for each person and have focused on identifying the one that is most beneficial (i.e., personalization). A straightforward approach is to ask people what kind of comparison target they prefer to see (e.g., upward vs. downward), and facilitate exposure to either this type of target or a non-preferred type. In one such study, participants who were matched to their preferred type marginally increased their daily step count over two weeks, whereas those who were matched to a non-preferred type demonstrated significant decreases in their step count over two weeks [60]. As social media companies are well aware, giving people what they prefer is an excellent way to promote engagement with technology. There are also benefits for treatment, as engagement with digital mental health tools is positively associated with improvement in outcomes of interest (e.g., depressive symptoms) [61].

However, here we raise several potential issues: people make comparisons for different reasons and may achieve one goal at the expense of another (e.g., feeling good in the moment vs. longer term) [62]. As a result, people may not have optimal insight into what works best for them, particularly if they are asked to aggregate over multiple, potentially distinct contexts. An alternative is to infer the optimal type of comparison for someone, which has been done in several creative ways. Jablonska and Zadel [63] used global, retrospective self-reports of women’s social comparisons on Instagram (including identification and contrast processes) and mental health symptoms such as depression, anxiety, and well-being, to construct comparer types. These types were women who reported predominantly positive responses (i.e., using upward identification and downward contrast and experiencing associated positive affect), women who reported predominantly negative responses (i.e., using upward contrast and downward identification, with associated negative affect), and women who reported equal use of positive- and negative-outcome processes. Less depression and anxiety and greater well-being were reported by positive types than negative types.

This is an interesting step that confirmed the between-person associations between identification/contrast comparison processes and mental health. Yet, it relied on cross-sectional, self-report methods that do little to help identify the optimal comparison for a given person or context. An alternative approach is to assess an individual’s comparison response across exposure to multiple types of comparison targets. Across multiple studies, Zhu et al. [64••] randomly assigned adults to a set of four upward, downward, or mixed targets (two upward, two downward) based on physical activity, once per day for nine days (phase 1). Participants could choose one of their assigned targets to view a profile, and motivation to be physically active in the moment was assessed before and after this process. Based on selection, change in motivation, and behaviorally assessed steps per day, a machine learning algorithm identified the optimal set of comparison targets for each participant to view and choose from, and either kept this consistent for the following 21 days or continued to randomize the choice set (phase 2). Those who received the personalized comparison target choice set showed greater increases in motivation to be physically active during phase 2 than those who continued to receive randomized sets.

Although motivation for physical activity is not a traditional mental health outcome of interest, there is considerable evidence that physical activity is strongly (and causally) associated with mental health [65,66,67]. Zhu et al.’s [64••] work also serves as an example of an approach that could help to identify the optimal comparison target for a given person—assuming that “optimal” is a person-level characteristic, rather than one that changes within person. As people use social media and related technology in multiple contexts that may affect their response to comparison opportunities (e.g., distinct cognitive or affective states), this is an open question that warrants considerable attention. In a follow-up to Zhu et al., Arigo and colleagues [68] used the baseline, daily randomization period across three studies to demonstrate extensive within-person variability in selection and response to physical activity-based comparison targets. In other words, the same person selected the same type of target on multiple days and showed a range of responses, from moderate increases to decreases in motivation to be active. Understanding whether factors such as pre-comparison affective state and/or identification or contrast processes contribute to the consequences of a given comparison are critical to advancing both basic and intervention science in this area.

Such work would enable effective adaptation of comparison opportunities, to ensure that each opportunity is matched to the comparer’s context as it changes over short periods. Mobile just-in-time adaptive interventions (JITAI) [69] test the effects of distinct treatment components on short-term outcomes of interest over multiple exposures, often using micro-randomization that embeds randomization into EMA or other intensive ambulatory assessment designs [59]. It then adjusts content, timing, frequency, and other aspects of the intervention to promote the most positive response. In this sense, it learns the contextual determinants of a user’s variability and adapts to offer the optimal treatment for the user in their immediate context. Meta-analytic work in this area shows that JITAIs work better than non-adaptive interventions for improving mental health outcomes and that adaptation to context is more efficacious than personalization [70•].

To our knowledge, social comparison is rarely included as a treatment component in JITAIs for promoting physical health [71] and has not yet been tested in JITAIs for mental health outcomes. This work will be particularly useful if it elucidates the immediate experience of comparison and its short-term effects with longer-term outcomes. As noted, and given the current availability of necessary technology, this is a critical area of opportunity for advancing our basic understanding of social comparison, its associations with mental health outcomes, and our available treatments.

Discussion

In contrast to common assumptions, social comparison is a complex process: comparison opportunities are everywhere, though people show meaningful heterogeneity in the extent to which they make and respond to comparisons (via social media and elsewhere). Importantly, heterogeneity exists both between and within person; the latter is usually given little attention, and ignoring either is problematic with respect to treatment recommendations. Specifically, our current understanding of associations between comparisons and mental health is based on weak evidence, including cross-sectional or longitudinal associations that cannot speak to causality or traditional experimental designs that reveal only the immediate effect of a single comparison event. Recent advances in research designs and mobile technology afford opportunities to overcome these limitations, by elucidating the comparison process as it unfolds in daily life and revealing the comparison contexts that might prompt optimal responses (see Table 2).

For instance, intensive ambulatory assessment designs such as EMA offer high ecological validity and help to clarify the timing, immediate context, and temporal sequencing of comparisons and their responses under naturalistic conditions. As noted, evidence generated with these methods could identify promising mechanistic pathways for intervention and point to opportunities to intervene in the moment, as needed. This approach could also be used to support patient progress in therapy, by providing daily or momentary reports for discussion during sessions or as assessments of progress (e.g., with respect to comparison responses). Micro-randomization (within person), JITAI, and related approaches also hold promise for identifying the people for whom and conditions under which distinct types of comparisons are beneficial (vs. harmful) for mental health outcomes. Specific next steps involve determining the extent to which variability in identification and contrast processes contribute to the effects of comparisons in real time and the extent to which they are modifiable. Intervention components such as cognitive restructuring may improve response; alternatives may include acceptance-based approaches such as cognitive de-fusion [72] and technological support using priming in the moment to prompt a particular response [73].

The ubiquity of social media and other socially networked technologies is only increasing, though their effects via social comparison remain unclear. As such, addressing open questions about comparison is critical for realizing the potential of personalized medicine. This line of work could reveal that comparisons do not directly affect mental health outcomes, freeing researchers and treatment providers to focus on the correct culprits. As it is likely that comparisons have a complex set of effects on mental health outcomes, however, using the combined strengths of intensive ambulatory assessment and experimental methods will be optimal for identifying and addressing the negative effects of social comparisons, for whom and under what circumstances they are most likely.