This cross-sectional study used primary data collected through a nationwide survey of emerging adult women aged 19 to 25 years in South Korea, referred to as the STOP 2025 survey. This age range reflects the conventional definition of emerging adulthood [19], excluding people aged 18 years, who are minors under Korean law. The study adhered to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for cross-sectional studies.

The survey was conducted via an online panel managed by a professional survey research agency and drawn from the subscriber base of a major South Korean telecommunications provider. This panel included approximately 90,000 registered women aged 19 to 25 years, representing approximately 4.9% of the national female population in this age group. This limited coverage reflects the panel’s restriction to subscribers of a single telecommunications provider who had also enrolled as survey panelists.

Eligible participants were emerging adult women who could read and understand Korean and who provided informed consent prior to participation. Exclusion criteria, assessed via self-report at the online screening stage, included (1) severe cognitive impairment or communication difficulties, (2) current pharmacological treatment for a diagnosed eating disorder, or (3) medically supervised weight loss treatment for a chronic condition.

Stratified sampling with quota allocation was applied based on residential area (metropolitan vs nonmetropolitan) and employment status (employed vs nonemployed) to obtain a demographically diverse sample of women aged 19 to 25 years. Quota targets for these strata were set based on population statistics from Statistics Korea [20], approximating the actual distribution of women aged 19 to 25 years in South Korea (approximately equal proportions for both residential area and employment status).

The minimum required sample size was calculated using the Cochran formula for a single population proportion, assuming a prevalence of 30% [21], a 95% confidence level, and a 5% margin of error, yielding an initial estimate of 323 participants. We assumed that individuals within the same quota stratum might share lifestyle-related characteristics relevant to UWMBs (eg, consumption patterns and environmental exposure), potentially resulting in within-stratum correlation; accordingly, each quota stratum was conservatively treated as a cluster for the sample size calculation. In the absence of Korea-specific intraclass correlation coefficient (ICC) estimates for UWMBs, we applied an ICC of 0.039 [22] and a cluster size (m) of 100 (consistent with the Korea National Health and Nutrition Examination Survey [KNHANES] sampling convention), yielding a design effect of 4.861 and a required analytic sample size of approximately 1569. To allow for potential dropouts and exclusions, 1600 participants were recruited, allocated equally across the 4 resulting strata (400 participants per stratum), of whom 1594 were included in the final analysis. Six participants were excluded due to an outlier BMI value or logically inconsistent responses to the weight management behavior items.

Data were collected via an online self-administered questionnaire distributed by a professional survey agency over a 1-week period in January 2025. The research team provided the agency with the study information sheet, consent form, and questionnaire. Responses within each stratum were automatically closed to further submissions once the predefined quota was reached.

Survey invitations were sent to the provider’s nationwide panel via SMS and pop-up notifications. Participants accessed the questionnaire only after reviewing the online consent form and selecting “Agree.” Those selecting “Disagree” were automatically exited from the survey. Eligible respondents proceeded to a screening page, and only those meeting the inclusion criteria and none of the exclusion criteria completed the survey. Survey completion required approximately 30 minutes. All data were anonymized and deidentified by the survey agency prior to transfer to the research team.

Prior to analysis, the anonymized dataset was screened for normality assumptions, outliers, and missing data. Categorical variables were dummy coded, and negatively worded items were reverse scored. Descriptive statistics (means, SDs, frequencies, and percentages) were computed to summarize participants’ sociodemographic characteristics and the prevalence of UWMBs. No poststratification weighting was applied in the analysis. Group differences between women with and without UWMBs were first examined using independent t tests and chi-square tests, based on variables selected according to the SDOH framework. Subsequently, a multivariable logistic regression analysis was conducted by entering variables that were selected a priori based on the SDOH framework and that showed significant associations in bivariate analyses. Odds ratios (ORs) and 95% CIs were estimated from the multivariable logistic regression model. Multicollinearity among predictors in the multivariable model was assessed using variance inflation factors (VIFs). Model fit was evaluated using the Hosmer-Lemeshow goodness-of-fit test and Nagelkerke R². Statistical significance was set at P<.05 (2-tailed). All analyses were performed using SPSS (version 29.0; IBM Corp).

This study was approved by the Institutional Review Board of Yonsei University Health System (4-2024-1456). A waiver of written consent was granted because the survey was conducted online through a professional research agency. Before participation, all respondents were presented with an online consent form describing the study purpose, procedures, confidentiality, and voluntary participation, including the right to withdraw at any time. Minimal personal information (sex, age, and residential area) was collected to verify eligibility and monitor quota compliance across strata. All responses were anonymized before being provided to the research team, and the survey agency permanently deleted all identifiable data after data delivery. Participants received a mobile gift voucher provided by the survey panel provider upon completion of the questionnaire. All procedures performed in this study were conducted in accordance with the ethical standards of the institutional review board and the Declaration of Helsinki and its later amendments.

The mean age of participants was 22.96 (SD 1.85) years. Most participants were currently enrolled in college (729/1594, 45.7%) or had graduated from college (n=539, 33.8%), and approximately half (n=796, 49.9%) were employed. The average SSS score was 5.26 (SD 1.64). The majority were unmarried (n=1537, 96.4%), and about one-quarter lived alone (n=402, 25.2%; Table 1).

Among all participants (N=1594), 1404 (88%) reported engaging in at least 1 UWMB during the past year. The most frequently reported UWMB methods were eating very little food (n=1277, 80.1%) and skipping meals (n=720, 45.2%), followed by taking diet pills (n=545, 34.2%), using food substitutes (n=518, 32.5%), fasting (n=250, 15.7%), smoking more cigarettes (n=162, 10.2%), self-induced vomiting (n=90, 5.6%), laxative use (n=26, 1.6%), and diuretic use (n=23, 1.4%). Multiple UWMB methods were common: 1002 (71.4%) participants who engaged in UWMBs reported using 2 or more methods, whereas 402 (28.6%) reported using only 1 method.

As shown in Table 1, engagement in UWMBs differed significantly according to educational attainment, employment status, monthly income, and marital status. No significant differences were observed in age, SSS, household composition, or residential or primary activity area.

As shown in Table 2, significant differences between the UWMB and non-UWMB groups were observed in descriptive and injunctive norms and in sociocultural influences related to appearance, including internalization of appearance ideals and perceived appearance-related pressures. The groups also differed significantly in weight self-stigma, media health literacy, social network support, and BMI. Significant group differences were additionally found in perceived body shape and weight disparity, with a higher proportion of participants in the UWMB group reporting desired weights lower than their actual weights and perceiving themselves as obese. In contrast, body weight misperception did not differ significantly between the groups.

Perceiving oneself as having an average or obese body shape, higher educational attainment, stronger descriptive norms, greater internalization of thinness or low body fat ideals, greater social network support, and higher BMI were significantly associated with increased odds of engaging in UWMBs (Table 3). No other factors remained significant after adjustment. VIF for all predictors in the multivariable model ranged from 1.03 to 4.26, indicating no substantial multicollinearity.

Participants who perceived themselves as having an average body shape showed the highest odds of UWMB engagement (OR 2.41, 95% CI 1.52-3.83), followed by those who perceived themselves as obese (OR 2.33, 95% CI 1.13-4.79), compared with those who perceived themselves as thin. Educational attainment showed a positive gradient: college students (OR 1.69, 95% CI 1.09-2.62) and college graduates (OR 1.81, 95% CI 1.13-2.91) had higher odds of engaging in UWMBs than high school graduates. Stronger descriptive norms (OR 1.38, 95% CI 1.19-1.61) and greater internalization of thinness or low body fat ideals (OR 1.32, 95% CI 1.01-1.71) were also significantly associated with UWMBs. Greater social network support (OR 1.29, 95% CI 1.01-1.63) and higher BMI (OR 1.10, 95% CI 1.01-1.19) were also positively associated with UWMB engagement.

This study demonstrated that UWMBs were highly prevalent among emerging adult women in South Korea, underscoring their importance as a public health concern. Within the SDOH framework, this study found that perceived body shape emerged as the factor most strongly associated with UWMB engagement among emerging adult women in South Korea, supporting the existing body of knowledge. Notably, this study further highlights the influence of structural determinants—including descriptive norms and internalization of thin or low body fat ideals—on UWMBs. These findings extend prior research that has primarily emphasized individual psychological and perceptual correlates [10,15,16] by demonstrating that both intermediary and structural determinants remain salient when considered simultaneously. Therefore, UWMBs should be understood not as matters of personal choice but as behaviors embedded within social and structural environments, highlighting the need for structural approaches aimed at shifting sociocultural norms, alongside individual-level efforts.

The prevalence of UWMBs in this study (88%) was substantially higher than the 30% estimate in national statistics [21] and exceeded the international estimates reported in prior research (approximately 50%-60%) [2,23]. In this study, diet pills were operationalized to include weight loss supplements, consistent with a common definition of the term; however, as it is unclear whether diet pills were defined similarly in other studies, comparisons of prevalence estimates across studies should be made with caution. A persistent upward trend in behaviors such as meal skipping and diet pill use has also been observed in Korea over the past decade [8], suggesting a genuine increase in these behaviors over time within this population. Moreover, Korean women frequently attempt weight loss [33] and often engage in unhealthy methods [8], even when they have relatively low BMI or do not perceive themselves as overweight. Comparable patterns have been reported in other Asian countries, where young women show consistently high levels of weight loss attempts across the entire BMI distribution, substantially exceeding those observed in European, Mediterranean, and American populations [34]. Longitudinal studies from the United States have shown a marked increase in extreme weight control behaviors—a subset of the behaviors encompassed by UWMBs—from adolescence to adulthood [2], suggesting that similar upward trends may be occurring elsewhere. Taken together, these findings suggest that UWMBs may be increasingly normalized within the sociocultural context examined in this study, shaped by everyday appearance-based comparison in digital media [4], self-discipline–oriented views of weight control [16], and widely shared thinness norms [15]. Therefore, efforts to reduce or prevent UWMBs require careful consideration of both individual and structural contexts that may encourage emerging adult women to engage in these risky behaviors.

This study reaffirms prior evidence that distorted body image is a salient correlate of UWMBs [10], with perceived body shape emerging as the factor most strongly associated with UWMB engagement. Notably, even women who perceived themselves as having an “average” body shape showed an elevated likelihood of engaging in UWMBs. Previous studies have documented substantial discrepancies between objectively measured BMI and perceived body shape, even among individuals within the normal weight range, with some tending to perceive their body size as larger than it actually is [35]. Such perceptual discrepancies suggest that UWMBs may be more closely related to subjective, visually oriented standards of body evaluation than to objective weight status. This interpretation is consistent with contemporary shifts in body perception, particularly in the context of the expanding use of social media, where body evaluation increasingly relies on visual self-presentation and the perceived gaze of others rather than on measurable indicators such as BMI [36]. Overall, these findings indicate that perceived body shape reflects a socially mediated aspect of body image and call for more multidimensional approaches to its measurement. Unlike previous studies [10], weight disparity and body weight misperception were not significantly associated with UWMB engagement in the present study. Given the heterogeneity of body image indicators currently used in research, future assessment models should incorporate cognitive, emotional, perceptual, and sociocultural dimensions rather than relying on a single index or construct.

The specific role of social media in shaping perceived body shape remains exploratory, as social media use and appearance-related online behaviors were not directly assessed in this study. The perceived gaze of others may nonetheless be interpreted through the lens of objectification theory [37], which posits that women come to adopt an observer’s perspective on their bodies. This phenomenon has been consistently documented in social media–based appearance comparison research [4,36]. In the social media context, this has been conceptualized as appearance-related social media consciousness (ASMC), which shows substantial associations with body surveillance, body shame, and appearance comparison [38] and remains associated with disordered eating among emerging adults even after controlling for general social media use [39]. Other evidence suggests that exposure to appearance-ideal content, rather than social media use itself, is more strongly linked to body dissatisfaction [17]. Future research should clarify which of these mechanisms best accounts for the present findings.

The tripartite influence model [14] helps clarify why descriptive norms and internalization of thin or low body fat ideals emerged as significant structural factors associated with UWMBs in this study. Even though perceived pressures from peers, family, or media were not independently significant, repeated exposure to others’ weight control behaviors may normalize such practices and encourage similar engagement [40]; higher levels of social network support for weight control may likewise reflect embeddedness in appearance-oriented social environments in which dieting and weight control behaviors are more visible and socially reinforced, thereby amplifying descriptive norms. As these norms become more pervasive, appearance-related standards may be internalized as personal expectations even in the absence of explicit pressure. These mechanisms underscore the need for interventions that address both individual self-perception and the broader sociocultural norms that shape and reinforce it.

Unlike the SDOH perspective [18] and prior evidence suggesting that individuals with lower SES are more vulnerable to unhealthy behaviors [41], prior research indicates that women with higher SES tend to monitor their weight more strictly and attempt weight loss more frequently [42], and that even among underweight women, those with higher income levels report more frequent weight loss attempts [33]. In our study, monthly income, employment status, and marital status were associated with UWMB engagement in bivariate analyses, but only higher educational attainment remained significantly associated with greater odds of engaging in UWMBs after multivariable adjustment. These patterns imply that appearance-oriented cultural contexts [16] may create heightened vulnerability among women with higher educational attainment. Although higher educational attainment is associated with greater awareness of weight-related health information [43], which would generally be expected to reduce risky behaviors, media health literacy—the construct measured in this study—was higher among women engaging in UWMBs in the bivariate analysis but lost significance after adjustment. This pattern may reflect more active health information-seeking related to weight control, and may also indicate that eHEALS, as a unidimensional scale [31], does not capture the capacity to critically resist appearance-related media ideals. Future research should therefore operationalize SES in a manner that reflects the developmental characteristics of emerging adult women to further investigate these associations.

UWMBs were assessed using a self-reported behavioral checklist rather than a validated psychometric instrument and were operationalized as a broad binary outcome encompassing 9 behaviors of markedly different severities. This approach may have introduced measurement error and may have contributed in part to the high prevalence observed in this study. Given the resulting imbalance in the outcome distribution, the ORs are more appropriately interpreted as indicators of the direction and relative magnitude of associations rather than as precise estimates of relative risk. Similarly, the measure of social network support for weight control was modified from an existing exercise-related scale, and its validity for this purpose has not been established beyond internal consistency.

Separately, the survey response rate could not be obtained from the survey agency, which further limits the ability to assess nonresponse or self-selection bias in this online panel. Although quota targets were set to approximate the national population distribution [20], the panel was restricted to subscribers of a single telecommunications provider, and no poststratification weighting was applied. The sample should therefore not be assumed to be representative of the general population of South Korean emerging adult women.

A cross-sectional design precludes causal or directional inference, and because associations are situated within specific sociocultural and media environments, the generalizability of the findings to other populations, cultural contexts, or genders remains uncertain. Social media use and appearance-related online exposure were not directly measured, despite their plausible role in shaping UWMB engagement. In this context, although prior research has repeatedly validated pathways linking body image, body dissatisfaction, and disordered eating [4,44], such analyses have rarely been extended to UWMBs. To rigorously examine how these mechanisms operate within sociocultural contexts, future research should use structural modeling approaches, including mediation and moderation analyses.

Beyond theoretical refinement and in line with public health implications, future research should also focus on developing intervention strategies that mitigate the sociocultural norms constructed around the visible body. Social norm–based approaches have shown some promise in shaping health-related behaviors by altering perceived norms and collective expectations [45]. Likewise, media-based public campaigns that shift the focus from appearance to functionality or strength have demonstrated limited yet encouraging effects in improving body image outcomes [46]. Such interventions may gradually contribute to redefining appearance-centered values and promoting more inclusive, health-oriented social norms.

This study demonstrates that UWMBs are highly prevalent among emerging adult women in South Korea and are shaped by both individual and structural determinants. UWMB engagement was associated not only with subjective body evaluation—particularly perceived body shape and BMI—but also with pervasive appearance-oriented social norms and the internalization of thin ideals. These cross-sectional findings suggest that UWMBs may not be effectively addressed through individual-level approaches alone, underscoring the potential need for interventions that target sociocultural norms while supporting healthier self-perceptions and more realistic body evaluations. Future research should clarify the pathways through which structural and individual determinants jointly relate to UWMBs using longitudinal designs and mediation analyses to inform culturally relevant prevention and intervention strategies. In addition, social media–based initiatives promoting diverse and health-oriented body ideals may represent an important strategy for reducing the social normalization of UWMBs.