Festinger (1954) argued that humans have a fundamental drive to compare themselves to others for self-evaluation, typically choosing targets that appear similar to themselves (Festinger, 1954; Wood, 1989). However, subsequent research has shown that the targets, motives, and outcomes of social comparison vary considerably, depending on situational factors. For instance, when their self-concept is threatened, people engage in downward comparisons, comparing themselves to less fortunate others, to enhance their subjective well-being (Taylor & Lobel, 1989; Wills, 1981). Conversely, upward comparison often has negative consequences.

Researchers have further highlighted the importance of the social environment in shaping comparison behavior (Wood, 1989). Individuals may engage in unwanted social comparisons because their environment prompts them to do so. SNS platforms represent such an environment. In particular, because people are typically exposed to the idealized self-presentations of their peers on Facebook, which creates an environment saturated with polished and curated images, they tend to engage primarily in upward comparisons (Bareket-Bojmel et al., 2016; Schmuck et al., 2019).

Upward comparison on SNS has been repeatedly linked to negative psychological outcomes. For example, individuals exposed to Facebook photos of physically attractive peers reported lower appearance satisfaction than those exposed to photos of less attractive peers (Kim & Park, 2016). Longitudinal evidence has shown that upward comparisons on Facebook undermine adults’ self-esteem and well-being across diverse age groups (Schmuck et al., 2019). Consistent with this pattern, a review study (Verduyn et al., 2020) noted that because SNS users predominantly engage in upward comparison, most studies report similar detrimental effects.

Social comparisons also occur in romantic relationships (Buunk et al., 1991). Research indicates that couples often evaluate their dating lives relative to others, and when they interpret these comparisons negatively, their relationship satisfaction and commitment tend to decline (Morry & Sucharyna, 2016).

Drawing on this evidence, the present study proposes that couples exposed to images of happy couples on Facebook, where upward comparisons are likely, may experience more negative emotions and reduced relationship satisfaction.

This study was conducted within a specific context. First, the study focused on people in their early 20s because they comprise the largest user base of social media, with over 90% of them utilizing social media and 70% of adults aged 18–29 years using Facebook (Auxier & Anderson, 2021). Specifically, individuals aged 18–24 years, who are in the emerging adulthood phase, exhibit the highest SNS usage (Duggan, 2015). Investigating the effects of couple photos on individuals in emerging adulthood is developmentally meaningful, as romantic relationships play a critical role in identity formation and self-concept during this life stage (Arnett, 2000). Moreover, individuals in this period often encounter greater challenges in maintaining their romantic relationships than those in other life stages (Shulman & Connolly, 2013), making them more susceptible to comparing themselves with other couples as a reference group. Research on social comparisons has demonstrated that individuals experiencing relationship difficulties are more inclined to compare their relationships with others (Buunk et al., 1991). Furthermore, recent findings show that increased SNS use among young adults in emerging adulthood is associated with lower couple satisfaction, which subsequently heightens conflict and ultimately triggers addictive SNS use, suggesting that romantic relationships may be closely intertwined with SNS use patterns (Bouffard et al., 2022). Therefore, an examination of the effects of Facebook couple photos on relationships between individuals in emerging adulthood holds significant relevance from a human development perspective.

This study also focused on women. Compared with men, women spend more time on SNS and upload photos more frequently (Aparicio-Martínez et al., 2020; Kasahara et al., 2019; Twenge & Martin, 2020). Additionally, women place greater importance on viewing photographs as the primary reason for using Facebook (Smith, 2014). They are also more likely to post idealized photos (Haferkamp & Krämer, 2010), and social comparisons with these idealized figures occur more frequently among women (de Felice et al., 2022). Moreover, women tend to use SNS in a relationship-oriented manner by providing and receiving social support (Tifferet, 2020). Given these patterns, the current study specifically examined female users’ responses to photos of their peers’ romantic relationships s on SNS. This focus is grounded in the observation that women’s SNS use, compared with men’s, is more centered on viewing others’ photos and engaging in comparison processes.

This study was conducted in South Korea. South Koreans have grown up in a collectivist culture and more accustomed to thinking about themselves in relation to others/groups (Nisbett, 2003). Within this collectivist culture, individuals exposed to couple photos on Facebook are more likely to engage in social comparison.

South Korea is also one of the countries that pioneered the use of social networking services (SNS) in its early stages (Choi et al., 2013). The national SNS usage rate increased from 55.1% in 2021 to 58.1% in 2023, and the rate among Generation Z showed the largest increase, rising from 72.6% to 87.2% (Kim, 2024). The average daily time spent on SNS was 39 minutes on weekdays and 51 minutes on weekends, while Generation Z reported 55 minutes on weekdays and 1 hour and 16 minutes on weekends. These figures demonstrate that SNS constitute a significant part of everyday life in South Korea.

In Korean SNS research, within a cultural context where SNS have become an essential part of daily life, many studies have focused on identifying predictors of excessive SNS use, often framing it in terms of concerns about SNS addiction tendencies. This body of work mainly emphasizes individual factors such as stress or narcissism (Ha & Kim, 2023; Kim, 2021; Kim & Lim, 2023). However, studies addressing intimate relationships, including romantic ones, remain scarce (Zhao et al., 2022), highlighting the need for research examining the influence of SNS use on couple relationships.

To summarize, in the era of SNS, couple relationships no longer exist in a vacuum. Relationships that were once among the most private aspects of human life have now increasingly become public in the SNS era. Many couples upload their photographs on SNS, which may not only affec t their perception of their own relationships but also influence other couples’ evaluations of their relationships. However, no empirical study has investigated the effects of photos of happy couples on Facebook on other couples’ relationship satisfaction or psychological responses. To fill this research gap, online experiments were conducted.

This study has several important policy, educational, and therapeutic implications. First, it provides foundational data for developing digital policy guidelines to mitigate the negative effects of SNS use. Second, the findings can inform media literacy programs for young adults actively engaged in romantic relationships. Third, as psychological and relational difficulties linked to SNS use are increasingly being reported, this study offers clinicians practical guidance for helping couples use SNS in ways that support and strengthen their relationship health.

The following research question was proposed: What are the effects of viewing photos of happy couples on Facebook on relationship satisfaction and mood in dating relationships among young Korean women?

This study utilized the Solomon four-group design, which included two experimental groups and two control groups (Solomon, 1949), because people do not typically engage in pretests while using Facebook in a natural setting. The Solomon four-group design enhances both internal and external validity by eliminating pretest sensitization and creating an experimental condition that closely resembles real-life scenarios (Jun, 2000; Shepreris et al., 2017). Consequently, this design is considered one of the most rigorous among various experimental designs (Bohecker & Horn, 2016). However, this design incurs additional costs compared to other experimental designs, resulting in a scarcity of studies using this approach. Specifically, this design necessitates twice the number of participants, which leads to significant financial costs (Braver & Braver, 1988). Furthermore, the analysis of the results becomes more complex because of the increased number of comparisons (Oliver & Berger, 1980).

The participants were randomly assigned to four groups: two experimental groups (EG1 with pretest and posttest, EG2 with posttest only) and two comparison groups (CG1 with pretest and posttest, and CG2 with posttest only). In Table 1, the symbol O indicates the outcome measurements (pretest or posttest, as appropriate), whereas X represents the experimental treatment and does not indicate the presence or absence of a pretest.

We believed that because participants were exposed to the pretest, which provides substantial information on the main variables of the study, they might have adjusted their responses as a result of pretest sensitization. Thus, to ensure ecological validity, the experimental procedure should closely resemble participants’ actual activities on SNS, which led to the use of the Solomon four-group design in the present study.

A total of 290 Korean women aged 19–25 years participated in the study. The inclusion criteria were that participants were in a romantic relationship and used Facebook. The participants were recruited by an online research company with over one million panelists nationwide. The company has a survey panel infrastructure that reflects population proportions by age group and region, based on Statistics Korea’s demographic data. The company uses a three-strike-out system. That is, when inconsistent or insincere answers were given three times, the panel was eliminated from the list, guaranteeing the trustworthiness of the panelists’ responses. Participants were paid a few dollars if they completed the experiment and survey without missing data. The online experiment was conducted in October 2017. The mean age of participants was 22.86 years (SD=1.86). Of the participants, 51.7% were college graduates (N=150). Additionally, 36.9% were college students (N=107), 10% had a high school diploma (N=29), and 1.4% had a master’s degree (N=4). Among the participants, 155 (53.4%) provided valid responses regarding the duration of their romantic relationships. Relationship lengths varied as follows: 22.6% had been in a relationship for less than 6 months, 24.5% between 6 months and 1 year, 12.9% between 1 year and 1.5 years, 7.1% between 1.5 and 2 years, 20.6% between 2 and 4 years, and 12.3% for more than 4 years. The remaining 46.6% did not report the duration of their relationship.

To compute the required sample size, a power analysis was conducted using G*Power 3.1. An alpha level of .05 and statistical power of .80 were specified. Since meta-analyses were unavailable in this area, the effect size was estimated by referring to a similar meta-analysis (Kazdin, 2003), which specifically examined the effects of media photos portraying the thin ideal on body image dissatisfaction among viewers, particularly involving young women’s upward comparisons (Groesz et al., 2002). Based on this reference, a medium effect size of .25 in ANCOVA (Cohen, 1988) was expected. Through this process, a minimum of 256 participants were required. Oversampling (N=290) slightly exceeded the minimum number of participants required to ensure a sufficient sample size.

Participants randomly assigned to the experimental groups (EG1 and EG2) were exposed to five Facebook News Feed pages displaying photos of five different couples who looked happy, whereas the other participants in the comparison groups (CG1 and CG2) were exposed to five Facebook News Feed pages featuring nature landscapes. Participants were instructed to view Facebook pages as if they were viewing their own pages and to look at each photo for at least 10 seconds. They were exposed to each Facebook page for 20 seconds, which gave them sufficient time for eliciting psychological responses (Taylor & Lobel, 1989). After viewing all the stimuli pages, they completed a post-survey.

The research team searched for candidate couple photographs by using hashtags such as “#coupleselfie,” “#couplephotos,” etc. on Facebook (Figure 1). The research team consisted of a professor and two undergraduate research assistants (RAs). The professor established initial guidelines for photo selection, after which the undergraduate RAs primarily sourced the photos. The final selection was jointly reviewed by a professor and the RAs. Undergraduate RAs, while possessing less professional expertise, were in their early twenties and highly familiar with their peers’ SNS environment and usage culture. This characteristic was advantageous for ensuring the ecological validity of the study.

The selection criteria were as follows: First, couples should appear to be in their early or mid-20s, as people are more likely to make comparisons with others who are similar in age. Second, couples should look happy because people usually upload idealized images on Facebook. Photographs of couples fighting on Facebook are rare. Such images could potentially provoke upward social comparison among viewers, which is common on Facebook Lastly, the couples needed to appear physically attractive but within a normal and realistic range. This criterion was applied because people are less likely to make comparisons with extraordinarily superior targets. To maintain ecological validity, we aimed to select couples who looked typically attractive in the way ordinary Facebook users present themselves, while excluding images that displayed unusually high levels of attractiveness, such as those resembling celebrities or professionally staged photos. Accordingly, couples that appeared celebrity or model-like were excluded.

Initially, 15 photographs that met the criteria were selected by undergraduate staff. The research team then discussed the appropriateness of the photographs by consulting the above criteria, leaving 10 photographs for further consideration. Next, the team asked 30 women in their early twenties to provide ratings on a ten-point scale regarding how happy the couples appeared and how similar the photos were to those typically uploaded on their Facebook accounts. Finally, the team selected the five photographs with the highest ratings. Participants were recruited through convenience sampling, and an online survey was conducted in August 2017.

Five photographs of natural landscapes were selected for comparison. This choice was made to provide a comparison condition that preserved the ecological context of viewing a Facebook page while removing relational or emotionally salient cues. Neutral images minimize emotional arousal and exclude relationship-relevant content that could otherwise confound the effects of exposure to couple photographs. The research team searched for candidate photos using hashtags such as “#nature,” and so on. Photos that might provoke emotions and thoughts were excluded because they could influence the viewers’ mood and trigger social comparisons. Consequently, photographs that appeared neutral and ordinary were selected.

Finally, we created virtual Facebook News Feed pages and included selected photographs. To this end, we used the actual Facebook News Feed pages of women in their 20s to ensure that the virtual Facebook pages resembled the participants’ own Facebook pages as closely as possible. All procedures are shown in Figure 2.

Based on previous experimental studies (Heinberg & Thompson, 1995; Kim & Park, 2016; Prichard & Tiggemann, 2012), we used a visual analogue scale (VAS) to assess the dependent variables (relationship satisfaction and mood). The question “How satisfied are you with your romantic relationship at this very moment?” was used to measure relationship satisfaction.

Participants’ mood changes were examined using the Positive and Negative Affect Scale (Watson et al., 1988), which asks participants to rate their feelings based on five items: “I feel anxious/happy/depressed/excited/angry at this very moment.” The two positive items (“happy,” “excited”) were reverse-coded, and all five items were summed to produce a composite negative mood score. Higher scores indicated more negative feelings. The internal consistency of the five items was acceptable (Cronbach ’s α=.78 for the pretest; α=.84 for the posttest).

This study follows Braver and Braver’s (1988) metaanalysis approach. The analysis procedures were as follows. The initial step involved a 2 × 2 between-group analysis of variance (ANOVA) to detect pretest sensitization. If the interaction effect of the variables is significant, it can be concluded that there is pretest sensitization, indicating that the pretest affects the treatment effect (Braver & Braver, 1988). In contrast, if the interaction effect is not significant, it means that the pretest does not affect the posttest.

Next, we investigated whether the experiment had a significant effect on the dependent variables. If there is a significant effect in this analysis, it could be evidence of an experimental effect. If there is no significant effect, that should not be considered conclusive evidence of an experimental effect. In this case, an analysis of covariance (ANCOVA) should be performed on the posttest scores, covarying pretest scores (Braver & Braver, 1988). As the research design involved two dependent variables (relationship satisfaction and mood), a Bonferroni correction (i.e., .05 divided by the number of dependent variables) was used to avoid committing a Type I error. Thus, a p-value less than 0.025 was considered statistically significant. The collected data were analyzed using the IBM Statistical Package for Social Sciences (SPSS), version 29.0.

Participants in the experimental groups responded to the statement, “couples in the photos that I’ve just seen look happy,” on a ten-point Likert scale ranging from 1 (strongly disagree) to 10 (strongly agree). Overall, participants rated the couples in the photos as happy (M=8.29, SD=1.39). Additionally, we asked participants in the experimental and comparison groups to respond to the statement, “I can see photos on my own Facebook pages that look similar to the photos I’ve just seen,” on a ten-point Likert scale (1=strongly disagree to 10=strongly agree). Participants in the experimental groups (M=8.48, SD=1.58) and comparison groups (M=7.82, SD=1.94) overall agreed that they could see similar pictures on their Facebook pages. These results indicate that the photos were successfully manipulated as intended.

Descriptive statistics of the dependent variables by groups are provided in Table 2.

The results indicated that there was no significant interaction effect. Relationship satisfaction was not affected by the pretest (F=.45, p=.50, d=.01). There was no significant effect on participants’ relationship satisfaction (F=.90, p=.34, d=.16). Thus, exposure to Facebook couple photos did not affect participants’ relationship satisfaction (see Table 3).

The results showed no significant interaction effect, indicating no evidence of pretest sensitization F(=.00, p=.96, d=.00). There was a significant effect on the participants’ mood (F=6.30, p<.01, d=.43), meaning that participants exposed to Facebook photographs of couples experienced fewer negative and more positive feelings (see Table 3).