Exercise identity and physical activity according to socioeconomic status in China: the moderating role of gender

Abstract

Background

Socioeconomic status (SES) has been associated with various health outcomes, yet its relationship with physical activity remains inconclusive.

Aims

Grounded in self-identity theory, this research aimed to examine the effects of both subjective and objective SES on physical activity, and tested whether exercise identity mediates these effects and whether gender moderates them.

Method

This research recruited 339 participants who completed a three-wave survey at two-week intervals. Objective SES was assessed using annual household income; subjective SES was measured with the MacArthur Scale of Subjective Social Status; exercise identity was assessed using the Exercise Identity Scale; and physical activity was measured via a single item assessing activity frequency. Preliminary analyses were conducted using SPSS 26.0, and time-lagged moderated mediation analyses were conducted using Mplus 8.0.

Results

Subjective SES was positively associated with physical activity (β = 0.10, p = 0.049), whereas objective SES showed no direct association (β = 0.01, p = 0.877). Exercise identity significantly mediated the associations of both objective SES (Effect = 0.01, 95% CI [0.002, 0.027]) and subjective SES (Effect = 0.02, 95% CI [0.004, 0.034]) with physical activity. Interestingly, gender moderated the mediated pathway for objective SES (β = 0.14, p = 0.008, 95% CI [0.031, 0.242]). Specifically, the mediation effect via exercise identity was significant for men (β = 0.10, p = 0.003, 95% CI [0.041, 0.172]) but not for women (β = −0.02, p = 0.611, 95% CI [− 0.080, 0.052]).

Limitations

This research relied on self-reported measures and the convenience sample of college students from mainland China, which may introduce bias and constrain the generalizability of the findings.

Conclusion

Objective and subjective SES differentially affect exercise identity and physical activity. Moreover, gendered expectations appear to modify the pathway from objective SES to physical activity via exercise identity, suggesting that interventions to promote exercise might benefit from consideration of both socioeconomic and identity-related factors.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-025-03575-2.

Introduction

Socioeconomic status (SES), broadly defined as an indicator of an individual’s or group’s social standing, is considered an important determinant of various health outcomes [1]. Two complementary indicators of SES are commonly employed in the literature: objective SES (e.g., income, education, occupation) and subjective SES (an individual’s perceived social rank) [2]. Objective SES represents an externally anchored and relatively stable indicator of social status, whereas subjective SES is defined as an individual’s perceived position within the social hierarchy [3]. Subjective SES more directly reflects personal interpretations of social standing and has been shown to exert independent effects on health and behavior, as it allows individuals to integrate past achievements and anticipated future prospects into their self-appraisals [4].

Importantly, both objective and subjective SES are positively associated with physical activity [5]. Objective SES is related to the availability of resources and infrastructure that facilitate engagement in physical activity, whereas subjective SES is associated with the cognitive and psychological processes underlying exercise behavior. Specifically, objective SES affords access to tangible resources and supportive environments. These include facilities for exercise [6], educational materials that encourage active activity [7], and sufficient income to support participation in physical activity [8, 9]. In contrast, subjective SES is considered to impact psychological mechanisms, such as health-related motivation and self-efficacy [10, 11]. For example, individuals with higher subjective SES tend to exhibit stronger exercise intentions, perceive fewer daily obstacles, and feel more confident in managing their schedules to include physical activity [10]. These conceptual distinctions and potential pathways imply that objective and subjective SES may affect physical activity via different mechanisms. Thus, the first research objective was to examine the effects of both objective and subjective SES on physical activity.

Moreover, the college years represent a critical transition from adolescence to adulthood, during which self‑identity formation is especially crucial [12]. Both objective and subjective SES play a role in shaping self-identity and influencing how individuals perceive and engage with physical activity [13]. Notably, exercise identity is widely regarded as a determinant in promoting engagement in physical activity [14]. When individuals view themselves as exercisers, they tend to experience greater enjoyment and confidence in engaging in physical activity, driven by an intrinsic motivation to stay consistent with their self-identity [15]. According to Chinese cultural norms, men are expected to exhibit bravery, ambition, and perseverance, whereas women are encouraged to embody gentleness, modesty, and reserve [16]. These cultural expectations render men’s participation in physical activity more socially accepted, whereas women may encounter greater societal constraints, particularly with respect to vigorous forms of exercise. Given these gender-based expectations around physical activity [17], the present study employed a three-wave longitudinal design, with two-week intervals between waves, to examine how objective and subjective SES impact physical activity. In addition, this study aimed to test test whether exercise identity mediates these effects and whether gender moderates the mediated pathways.

Literature review

SES and physical activity

Researchers commonly assess SES using two approaches. Objective SES is assessed based on individuals’ income level, educational status, and occupational status, whereas subjective SES is measured by evaluating an individual’s perceived rank within a social hierarchy [18]. Objective and subjective SES are related, with objective SES often shaping individuals’ perceptions of their social status through access to material resources and social standing [1, 19]. Beyond this, subjective SES reflects personal perceptions of one’s social status, integrating both past achievements and future prospects into their self-appraisals [3, 4]. Subjective SES is also influenced by various situational and psychological factors, such as mood, neighborhood satisfaction, and perceived health [1].

Both objective and subjective SES are associated with physical activity [5], but they impact behavior through distinct mechanisms and yield different outcomes (See Supplementary Table 1). However, debate persists over which measure more effectively explains social behaviors, as findings linking these two SES indicators to physical activity have been inconsistent (see Supplementary Table 2). Specifically, some studies reported a positive association between objective SES and physical activity [20–25], yet other investigations find no significant relationship [26–29], suggesting that higher objective SES does not uniformly translate into healthier lifestyles. These inconsistencies may stem from heterogeneity in how objective SES is measured: some studies use education and income alone [21, 22, 28], while others also include occupation, living standards, or parental SES [7, 24, 26]. These different indicators capture distinct facets of objective SES and may therefore produce divergent results. Similarly, research on subjective SES and physical activity is mixed. In particular, several adult studies report significant positive associations between higher subjective SES and greater physical activity [24, 26, 30, 31]. Among adolescents, the correlation between school‑referent subjective SES and accelerometer‑measured physical activity is r = 0.23, whereas society‑referent subjective SES shows a non‑significant association [32]. Quon (2015) reports a small negative association with moderate exercise results [33], Lee (2022) finds a positive association in young adults [26], while Zhang (2017) shows no association in elderly Hong Kong adults [27].

These inconsistencies reflect differences in study populations (age, culture, gender composition), measurement heterogeneity (school‑ vs. society‑referent SSS; objective vs. subjective SES; self‑report vs. accelerometry), and unresolved psychological and contextual mechanisms. Specifically, high objective SES can provide material resources and opportunities that facilitate engagement in physical activity. High‑status individuals may also use physical activity as a status signal to distinguish themselves from lower‑status groups [34]. In contrast, low objective and low subjective SES are linked to more frequent adverse life experiences and a reduced sense of control [35–38]. These individuals are more inclined to adopt unhealthy coping behaviors, such as reduced physical activity, in response to psychological distress [39]. Thus, the present research aimed to clarify the associations between both objective and subjective SES and physical activity, and to examine the psychological mechanisms that may account for those associations.

Exercise identity as the mediator

College students undergo a significant transition period from adolescence to adulthood, during which the development of self-identity, according to self-identity theory, is of paramount importance [12]. Exercise identity is defined as the degree to which individuals integrate physical activity into their self-identity and consider themselves exercisers [40, 41]. Research has shown that exercise identity positively influences both exercise intention and the long-term maintenance of physical activity [14].

Social status is a critical factor in self-identity formation and development. A well-defined sense of identity is largely influenced by social factors, with SES playing an important role in this process [13]. Self-identity theory posits that individuals construct their self-identity through the interplay of various factors, including sociodemographic characteristics, personal traits, social roles, and group affiliation [42]. Family economic status is thought to impact children’s identity-related information processing and decision-making because their parents are more capable of investing time and money in their growth [13]. Conversely, parents with low economic status often face economic hardship and poor health, leading them to focus on their financial needs rather than on their children’s growth [43].

Although studies have focused on the effect of SES on self-identity [13, 42], the relationships between the two SES measures and self-identity within the exercise domain remain underexplored. According to past literature, individuals with low objective SES are often exposed to disadvantaged living conditions and limited social resources, which heighten feelings of uncertainty and unpredictability (e.g., unemployment) [39]. These adverse life events and chronic stressors have been shown to impair cognitive functioning and negatively affect health-related behaviors [44, 45]. As a result, individuals with lower objective SES are more likely to focus on immediate financial needs rather than invest in long-term health behaviors [43]. In contrast, individuals with higher objective SES have greater access to material and social resources, enabling them to pursue personal goals and interests (e.g., physical activity) [39]. Actually, access to these resources not only promotes participation in physical activity but also fosters the development of exercise identity, which is considered a key psychological mechanism for long-term exercise engagement [6, 8, 9, 14]. According to self-determination theory [46], having adequate resources enhances individuals’ autonomy and competence in exercise, strengthens intrinsic motivation, and supports the integration of physical activity into the self-concept.

Subjective SES is also linked to exercise identity through the processes of group identification and social comparison. According to social identity theory, individuals who perceive themselves as having higher SES are more motivated to align with the social norms and cultural values of high-status groups [47]. Accordingly, they engage in physical activity as a means of signaling status, reinforcing their identity as exercisers. This internalization process helps incorporate physical activity into their broader self-concept [47]. Conversely, individuals with lower perceived SES often experience a sense of deprivation, which negatively impacts their self-identity [48, 49]. Notably, exercise identity is widely considered a factor that promotes physical activity [14]. When individuals perceive themselves as exercisers, they are more likely to derive enjoyment from physical activity and feel capable of sustaining it, because they are motivated to maintain their identity [15]. Based on this reasoning, exercise identity may serve as a mediator in the relationship between the two SES measures and physical activity.

Gender as the boundary condition

Research has consistently demonstrated gender differences in physical activity engagement [31]. Gender is considered a critical factor in shaping physical activity, as it can signal one’s social identity, including perceptions of masculinity and femininity [17]. According to exercise intensity, men tend to participate in more moderate-to-vigorous physical activity and leisure-time physical activity compared to women [5, 50]. This gender gap in physical activity may be attributed to the “hegemonic notion of athleticism as a masculine trait.” Masculinity is often described as the image of a self-reliant, strong, hypersexual, and dominant man, whereas femininity is defined as the image of a compliant, nurturing, empathetic, and sexually passive woman [51, 52].

Given the gendered nature of physical activity and the socially constructed meanings associated with masculinity and femininity [53, 54], the relationship between SES and physical activity appears to differ between men and women [5]. For men, social status is often attributed to the demonstration of masculine traits, such as physical strength and athleticism [54]. They are thus more likely to engage in physical activity and cultivate an exercise identity, as these behaviors align with socially valued masculine ideals [55]. Conversely, strong femininity is strongly associated with lower perceived sports competence, which is considered a barrier to physical activity engagement for women [56]. The significance of maintaining a masculine, athletic self-image positively influences men’s participation in physical activities, whereas women’s motivations for physical activity are more often related to health and well-being, perhaps reflecting the gendered nature of femininity [54]. This underscores the critical role of gender in shaping the complex interplay between social status, exercise identity, and physical activity behaviors. Specifically, the effects of SES may therefore lead men to express their masculine identities through engagement in physical activity. In contrast, women may be more reluctant to view themselves as exercisers, as this could conflict with societal expectations and perceptions of femininity. Moreover, women in adulthood appear to be disproportionately affected by low childhood socioeconomic status (SES) [57]. Possible explanations include greater physiological vulnerability to early disadvantage and stronger social‑status marginalization for females, which can amplify the lifelong impacts of low childhood SES [58]. Family SES also interacts with gender to shape educational opportunities. In lower‑SES families, girls are often more likely than boys to encounter constrained access to higher education, contributing to gendered inequalities in attainment [59].

Traditional gender norms encourage men to be achievement‑oriented, such as positive, brave, ambitious, and persevering, while women are expected to be gentle, modest, and reserved in Chinese culture [16]. These cultural prescriptions make men’s participation in physical activity more socially acceptable and even valued, whereas women may face greater social constraints on visible, vigorous exercise. Consistent with this, parental physical‑activity orientation appears to shape sons’ attraction to activity more strongly than daughters’ in China. Thus, boys were significantly influenced by both parents’ orientations, whereas girls showed no such effect [60]. Therefore, gender is likely to moderate the SES–physical activity relationship, with SES differences in physical activity expected to be larger for men than for women.

The present study

Although prior research has documented links between socioeconomic status (SES) and physical activity, the mechanisms through which SES influences activity remain underexplored. College students undergo a transition from adolescence to adulthood. This population also faces distinct stressors, such as academic pressure, changing lifestyles, and evolving social comparisons, that are considered to shape both perceived socioeconomic standing and health behaviours [61–63]. According to self-identity theory, the development of self-identity during this period is of critical importance [12, 64]. Importantly, exercise identity has been shown to promote exercise intentions and long‑term maintenance of physical activity [14, 40, 41], and is plausibly shaped by socioeconomic resources and perceptions. Objective SES confers material resources and opportunities that facilitate involvement in exercise, whereas subjective SES reflects perceived relative standing and social comparison processes that may independently influence identity formation. In Chinese culture, the emphasis on male achievement and the salience of material markers of status suggest that gender may condition these processes, with objective SES likely having a stronger influence on men’s exercise identity and behaviour. This research proposed these hypotheses:

• H1: Objective SES (H1a) and subjective SES (H1b) are positively correlated with physical activity.

• H2: Exercise identity mediates the relationships between objective SES (H2a) and subjective SES (H2b) and physical activity.

• H3: Gender moderates the mediation of exercise identity in the relationship between SES and physical activity. Specifically, the mediation effect of objective SES on physical activity via exercise identity (H3a) and the mediation effect of subjective SES on physical activity via exercise identity (H3b) are expected to be stronger for men than for women.

Methods

Participants and research design

The study was conducted between November and December 2022. We recruited 352 college students from several universities in northern China using the convenience sampling method. Inclusion criteria were: (1) college student; (2) aged 18 years or older; (3) no history of psychiatric disorders; (4) no physical disabilities; (5) voluntary participation in the study; and (6) completion of all three waves of the survey. Exclusion criteria included: (1) non-college student; (2) under 18 years of age; (3) history of psychiatric disorders; (4) presence of physical disabilities; or (5) absence from any of the three survey waves. Thirteen participants were excluded from further analysis due to incomplete data (12 completed the first wave, and one completed the first two waves), resulting in a final sample of 339 participants (Age = 22.63 ± 2.22 years; 100 women). All participants were informed of the study procedures and provided informed consent. The study protocol was approved by the Ethics Committee of Macau University of Science and Technology (Ethics Approval Number: MUST-20240412001). Participants completed a three-wave online survey at two-week intervals, following the approach employed in previous longitudinal research designs [65–67]. All survey administrations were conducted on weekends. At Time 1 (T1), participants reported age and gender (Women = 0, Men = 1) and completed measures of objective and subjective socioeconomic status, exercise identity, and baseline physical activity. The exercise identity scale was administered again at Time 2 (T2), and the physical activity scale was administered again at Time 3 (T3).

Measures

Objective socioeconomic status

Annual household income was used to evaluate objective SES based on Ng & Diener (2014) [68]. Annual household income was assessed using a single item: “What is your annual household income?”. Respondents answered based on an eight-point scale (1 = less than 10k, 2 = 10k-20k, 3 = 20k-40k, 4 = 40k-80k, 5 = 80k-160k, 6 = 160k-320k, 7 = 320k-640k, 8 = more than 640k).

Subjective socioeconomic status

Subjective SES was assessed using the MacArthur Scale of Subjective Social Status developed by Adler et al. (2000) [69]. This scale was presented in a visual ladder with 10 rungs, representing the social hierarchy in society. Participants were asked: “What class do you think you are currently in?”. The score of 1 indicates the perception of being at the lowest level of social status, while a score of 10 indicates the highest. The MacArthur scale has demonstrated strong reliability and validity within Chinese populations [70, 71].

Exercise identity

Exercise identity was measured using the nine-item exercise identity scale developed by Anderson & Cychosz (1994) [12], the Chinese version of which has demonstrated good reliability and validity in previous studies [72]. Participants responded to items on a 7-point Likert scale ranging from 1 (completely disagree) to 7 (completely agree). Cronbach’s alpha for the scale ranged from 0.93 to 0.95 in this study.

Physical activity

Physical activity was measured using an item adapted from previous research [73–76], which asked respondents to indicate their frequency of physical activity on a 4-point scale (1 = never; 2 = rarely; 3 = occasionally; 4 = regularly). Although single-item measures have been employed in previous studies, they remain limited in capturing the multiple characteristics of physical activity (See details in Supplementary Table 3).

Data analysis

We first conducted descriptive and correlational analyses to examine the study variables using SPSS (version 26.0). Attrition analysis and Harman’s single-factor test were performed to assess potential systematic bias in SPSS (version 26.0). Second, we used Mplus (version 8.3) to conduct linear regression analyses to examine the direct effects of objective SES (T1) and subjective SES (T1) on physical activity (T3). Third, we tested a mediation model to examine the mediating effect of exercise identity (T2) on the relationships between objective SES (T1) and physical activity (T3), and between subjective SES (T1) and physical activity (T3). The mediation effects were examined using the bootstrapping approach with 5,000 resamples. Fourth, we tested a moderation model to examine the moderating role of gender on the mediated relationship between SES (both objective and subjective) and physical activity through exercise identity. Simple slope analyses were conducted to further show the interaction between gender and both objective and subjective SES (T1) in predicting exercise identity (T2). Statistical significance was determined using a two-tailed threshold of p < 0.05.

Results

The descriptive statistics

Table 1 presented the descriptive statistics for all study variables, including means, standard deviations, and correlation coefficients. There were 339 participants (100 women, 239 men; M_age = 22.63, SD = 2.22), with women averaging 22.82 years (SD = 1.49) and men 22.56 years (SD = 2.47). Mean scores were 4.61 (SD = 1.64) for objective SES and 4.75 (SD = 1.62) for subjective SES. Exercise identity averaged 5.13 (SD = 1.19) at T1 and 5.14 (SD = 1.20) at T2. Physical activity scores were 3.28 (SD = 0.86) at T1 and 2.89 (SD = 1.00) at T3.

Table 1.

Means, standard deviations, and pearson correlation coefficients of the variables (N = 339)

	1	2	3	4	5	6	7	8
1.Age (T1)	1
2.Gender (T1)	−0.05 [−0.150, 0.030]
3.Objective Socioeconomic Status (T1)	−0.05 [−0.152, 0.056]	−0.05 [−0.152, 0.054]
4.Subjective Socioeconomic Status (T1)	−0.09 [−0.209, 0.021]	−0.07 [−0.175, 0.034]	0.19*** [0.044, 0.318]
5.Exercise Identity (T1)	−0.13* [−0.231, −0.025]	0.28*** [0180, 0.374]	0.05 [−0.058, 0.161]	0.12* [−0.002, 0.232]
6.Exercise Identity (T2)	−0.16** [−0.261, −0.064]	0.28*** [0.171, 0.375]	0.13* [0.017, 0.241]	0.21*** [0.098, 0.330]	0.75*** [0.686, 0.805]
7.Physical Activity (T1)	−0.18*** [−0.274, −0.082]	0.20*** [0.088, 0.309]	0.04 [−0.076, 0.140]	0.10 [−0.001, 0.212]	0.44*** [0.328, 0.542]	0.42*** [0.315, 0.521]
8.Physical Activity (T3)	−0.15** [−0.252, −0.060]	0.24*** [0.134, 0.344]	0.04 [−0.063, 0.138]	0.14* [0.022, 0.247]	0.38*** [0.274, 0.475]	0.43*** [0.334, 0.518]	0.48*** [0.380, 0.570]	1
Mean	22.63	–	4.61	4.75	5.13	5.14	3.28	2.89
SD	2.23	–	1.64	1.62	1.19	1.20	0.86	1.00

Note. * p < 0.05, ** p < 0.01, *** p < 0.001; Gender: Women = 0, Men = 1; T1, T2 and T3 indicate that the variables with T1 were measured at time 1, the variables with T2 were measured at time 2, and the dependent variable T3 measured at lagged time 3; 95% confidence intervals are shown in brackets below the coefficients

Objective SES at T1 was positively correlated with subjective SES at T1 (r = 0.19, p < 0.001) and exercise identity at T2 (r = 0.13, p = 0.014). Subjective SES at T1 was significantly associated with exercise identity at both T1 (r = 0.12, p = 0.026) and T2 (r = 0.21, p < 0.001), as well as with physical activity at T3 (r = 0.14, p < 0.012). Exercise identity at T1 was positively correlated with exercise identity at T2 (r = 0.75, p < 0.001), physical activity at T1 (r = 0.44, p < 0.001), and physical activity at T3 (r = 0.38, p < 0.001). Similarly, exercise identity at T2 was positively associated with physical activity at both T1 (r = 0.42, p < 0.001) and T3 (r = 0.43, p < 0.001). In addition, physical activity at T1 showed a significant positive correlation with physical activity at T3 (r = 0.48, p < 0.001).

Attrition analysis and common method variance

Attrition analyses revealed no significant differences between participants who completed all waves of data collection and those who dropped out in terms of gender (χ² = 0.25, p = 0.617), age (t = −0.76, p = 0.446), objective SES (t = 0.17, p = 0.862), subjective SES (t = 1.04, p = 0.298), exercise identity (t = 0.45, p = 0.650), and physical activity (t = 1.01, p = 0.314; see Table 2).

Table 2.

Comparison of baseline characteristics between completers and dropouts

Variables	Completers (n = 339)	Dropouts (n = 13)	t/χ2	P
Age, Mean [96]	22.63 (2.23)	22.15 (2.23)	−0.76	0.446
Gender, n (%)			0.25	0.617
Male	239 (70.50)	10 (76.92)
Female	100 (29.50)	3 (23.08)
Objective Socioeconomic Status, Mean [96]	4.61 (1.64)	4.69 (2.02)	0.17	0.862
Subjective Socioeconomic Status, Mean [96]	4.75 (1.63)	5.23 (1.79)	1.04	0.298
Exercise Identity, Mean [96]	5.14 (1.19)	5.29 (1.25)	0.45	0.650
Physical Activity, Mean [96]	3.16 (0.92)	3.42 (0.76)	1.01	0.314

To assess for common method variance, we conducted Harman’s single-factor test (see Table 3). The first factor accounted for 49.03% of the total variance across the four common factors, which is below the recommended threshold of 50%, suggesting that common method variance was not a significant concern in this study [77].

Table 3.

Harman’s single-factor test for common method bias

Factor	Eigenvalue	Variance explained (%)	Cumulative variance explained (%)
1	11.77	49.03	49.03
2	1.69	7.05	56.08
3	1.27	5.30	61.38
4	1.18	4.90	66.28

Note: Only factors with eigenvalues greater than 1 are presented

The effects of social status on physical activity

We conducted a linear regression analysis with objective SES and subjective SES (T1) as independent variables and physical activity as the dependent variable (T3). The results indicated that subjective SES was positively associated with physical activity (β = 0.13, s.e. = 0.03, p = 0.016; see Table 4), whereas the association between objective SES and physical activity was not statistically significant (β = 0.01, s.e. = 0.03, p = 0.843). These results remained consistent after controlling for covariables (objective SES: β = 0.01, s.e. = 0.03, p = 0.877; subjective SES: β = 0.10, s.e. = 0.03, p = 0.049). Thus, our results supported H1b but not H1a.

Table 4.

Linear regression predicting physical activity at time 3 (N = 339)

Predictor	Model 1				Model 2
	β	s.e.	p	95% CI	β	s.e.	p	95% CI
Objective socioeconomic status (T1)	0.01	0.03	0.843	[−0.097, 0.119]	0.01	0.03	0.877	[−0.087, 0.102]
Subjective socioeconomic status (T1)	0.13	0.03	0.016	[0.026, 0.242]	0.10	0.03	0.049	[0.001, 0.191]
Physical activity (T1)					0.42	0.06	< 0.001	[0.329, 0.521]
Age					−0.06	0.02	0.212	[−0.154, 0.034]
Gender					0.16	0.11	< 0.001	[0.066, 0.255]
R2	0.019				0.264
F	3.177*				23.941***

Note: * p < 0.05, *** p < 0.001; Gender: Women = 0, Men = 1; T1 indicates that the variables with T1 were measured at time 1; 95% CI represents the 95% confidence interval

The mediation effect of exercise identity

We conducted a mediation model to examine whether exercise identity (T2) mediated the relationships between objective and subjective SES (T1) and physical activity (T3). The results showed a significant positive association between subjective SES and exercise identity (β = 0.20, s.e. = 0.06, p = 0.001), but not between objective SES and exercise identity (β = 0.10, s.e. = 0.06, p = 0.080). After controlling for exercise identity (T1), physical activity (T1), and demographic variables, the associations became statistically significant (objective SES: β = 0.08, s.e. = 0.04, p = 0.036; subjective SES: β = 0.11, s.e. = 0.04, p = 0.002; Fig. 1; Table 5).

Fig. 1.

The mediating effect of exercise identity on the relationship between socioeconomic status and physical activity (N = 339). Note.* p < .05, ** p < .01; T1, T2, and T3 indicate that the independent variables were measured at time T1, the mediators were measured at time T2, and the dependent variable measured at time T3; 95% confidence intervals are presented in brackets following the coefficients; Black arrows represent significant paths, whereas gray arrows denote non-significant paths; The “-” sign indicates a negative effect, while the absence of a sign indicates a positive effect

Table 5.

Path coefficients for mediation effects (N = 339)

Path	β	s.e.	95% CI	p
Objective socioeconomic status (T1) → Exercise identity (T2)	0.08	0.04	[0.009, 0.152]	0.036
Objective socioeconomic status (T1) → Physical activity (T3)	−0.01	0.05	[−0.111, 0.080]	0.763
Subjective socioeconomic status (T1) → Exercise identity (T2)	0.11	0.04	[0.046, 0.192]	0.002
Subjective socioeconomic status (T1) → Physical activity (T3)	0.06	0.05	[−0.030, 0.173]	0.221
Exercise identity (T2) → Physical activity (T3)	0.22	0.08	[0.130, 0.442]	0.004

Note: 95% CI represents the 95% confidence interval

Considering predictors of physical activity, there was a significant effect of exercise identity (β = 0.42, s.e. = 0.05, p < 0.001). However, neither objective SES (β = −0.03, s.e. = 0.05, p = 0.552) nor subjective SES (β = 0.05, s.e. = 0.05, p = 0.330) was directly associated with physical activity. The results remained consistent after controlling for exercise identity (T1), physical activity (T1), and demographic variables (Exercise identity: β = 0.22, s.e. = 0.08, p = 0.004; objective SES: β = −0.01, s.e. = 0.05, p = 0.763; subjective SES: β = 0.06, s.e. = 0.05, p = 0.221). Importantly, consistent with Hypotheses 2a and 2b, there were significant mediating effects of exercise identity on the relationship between objective SES and physical activity (Mediation effect = 0.01, s.e. = 0.01, 95% CI [0.002, 0.027]), and between subjective SES and physical activity (Mediation effect = 0.02, s.e. = 0.01, 95% CI [0.004, 0.034]; Table 6). To further examine the independent effects of objective SES and subjective SES on physical activity via exercise identity, we conducted separate mediation analyses. The results were consistent with the main findings, supporting the mediating role of exercise identity in both models (see Supplementary Tables 4–5).

Table 6.

The results of the mediation effects test and moderating effects (N = 339)

Path	Mediation Effect	s.e.	p	95% CI
Objective socioeconomic status (T1) → Exercise identity (T2) → Physical activity (T3)	0.01	0.01	0.081	[0.002, 0.027]
Subjective socioeconomic status (T1) → Exercise identity(T2) → Physical activity (T3)	0.02	0.01	0.044	[0.004, 0.034]
The moderating effect of gender: Men	0.03	0.01	0.013	[0.013, 0.068]
The moderating effect of gender: Women	−0.01	0.01	0.623	[−0.029,0.018]

Note: 95% CI represents the 95% confidence interval

The moderating effects of gender

To examine the moderating role of gender in the relationships between SES (T1) and exercise identity (T2), we conducted a moderated mediation model. The results indicated that gender significantly moderated the association between objective SES and exercise identity (β = 0.14, s.e. = 0.05, p = 0.008, 95% CI [0.031, 0.242]). However, gender did not moderate the relationship between subjective SES and exercise identity (β = 0.01, s.e. = 0.07, p = 0.999, 95% CI [−0.138, 0.146]; Fig. 2). Simple slope analyses revealed that the positive association between objective SES and exercise identity was significant for men (β = 0.10, s.e. = 0.03, p = 0.003, 95% CI [0.041, 0.172]), but not for women (β = −0.02, s.e. = 0.03, p = 0.611, 95% CI [−0.080, 0.052]; Fig. 3).

Fig. 2.

The moderating effect of gender on the mediating effect of exercise identity (N = 339). Note. ** p < .01, *** p < .001; Gender: Women = 0, Men = 1; T1, T2, and T3 indicate that the independent and moderator variables were measured at time T1, the mediators were measured at time T2, and the dependent variable was measured at time T3; 95% confidence intervals are presented in brackets following the coefficients; Black arrows represent significant paths, whereas gray arrows denote non-significant paths; The “-” sign indicates a negative effect, while the absence of a sign indicates a positive effect

Fig. 3.

The interaction effects of gender and objective socioeconomic status on exercise identity (N = 339). Note: Simple slope analyses revealed that the positive association between objective SES and exercise identity was significant for men (β = 0.10, s.e. = 0.03, p = 0.003, 95% CI [0.041, 0.172]), but not for women (β = -0.02, s.e. = 0.03, p = 0.611, 95% CI [-0.080, 0.052]

Next, we examined whether gender moderated the mediating effect of exercise identity (T2) in the relationship between objective SES (T1) and physical activity (T3). Consistent with H3a, the mediating effect of objective SES on physical activity through exercise identity was significant for men (Mediation effect = 0.03, s.e. = 0.01, p = 0.013, 95% CI [0.013, 0.068]; Table 6). However, the mediating effect was not significant for women (Mediation effect = −0.01, s.e. = 0.01, p = 0.623, 95% CI [−0.029, 0.018]), failing to support Hypothesis 3b. To evaluate the potential impact of missing data, we reanalyzed the full sample (N = 352) using full information maximum likelihood estimation. The results from these analyses were consistent with those obtained in the primary analyses (see Supplementary Tables 6–7).

Discussion

Our research revealed a direct relationship between subjective SES and physical activity, whereas no such relationship was observed for objective SES. Furthermore, exercise identity served as a time-lagged mediator in the associations between both SES measures and physical activity. Notably, the study identified a moderating effect of gender on the relationship between objective SES and exercise identity. Specifically, men with higher levels of objective SES were more likely to engage in physical activity than those with lower objective SES. However, this association was not observed among women, suggesting that the influence of objective SES on physical activity through identity-related mechanisms may be more pronounced in males.

Although our findings were partially inconsistent with previous studies [10, 21, 25, 31, 33], our research employed a longitudinal design and demonstrated that subjective SES and objective SES play distinct roles in promoting physical activity. Specifically, individuals with higher subjective SES were more likely to engage in physical activity, whereas objective SES showed no direct effect. Importantly, subjective SES is not equivalent to objective SES; rather, it represents an individual’s perceived social standing within a hierarchy. Subjective SES not only reflects objective indicators such as income or education but also incorporates past achievements and future expectations into one’s self-identity [78]. Thus, subjective SES captures how individuals evaluate their social position and identity within a broader social context, which in turn may influence health functioning, including health motivation and health behaviors [3].

The present study found no significant association between objective SES and physical activity, which may be explained by two primary factors. First, cultural context likely plays a critical role. In China, the collectivist cultural orientation emphasizes the cultivation of harmonious interpersonal relationships and adherence to social norms [79]. Engagement in physical activity may be influenced not only by objective socioeconomic resources but also by social support and normative expectations [80]. These collective values may attenuate the direct impact of objective SES on physical activity, a relationship more commonly observed in individualistic societies. Second, variations in the measurement of objective SES may also account for the divergent findings. In this research, objective SES was assessed solely through income, whereas other studies have utilized broader composite indicators that incorporate educational attainment, occupational status, and additional socioeconomic markers [22, 24].

Although objective SES reflects an individual’s economic resources and social position relative to others, its influence on physical activity may depend on moderating factors such as gender, age, and habitual exercise patterns. Prior research has emphasized the critical role of self-perception in the adoption of health behaviors, as such behaviors are often perceived as markers that distinguish high-SES individuals from lower-SES groups [34]. Specifically, individuals with low subjective SES are more likely to encounter negative life events and perceived lack of control, which may lead them to adopt unhealthy coping behaviors in response to psychological distress [39, 81]. Notably, we found no direct effect of objective SES on physical activity, exercise identity served as a significant mediator in this relationship. This aligns with prior literature suggesting that identity plays a mediating role in the association between objective SES and outcomes such as altruistic behavior [82]. However, those studies often identified a direct effect of objective SES on prosocial behaviors, even when identity was included as a mediator. One possible reason is that individuals often expect positive feedback after prosocial behaviors, such as reputational benefits [83]. In contrast, the benefits of physical activity tend to be long-term and less immediately tangible, which may explain the absence of a direct SES effect in our findings. It is therefore plausible that objective SES influences physical activity indirectly, by shaping self-perceptions, such as exercise identity. This interpretation is consistent with the Reserve Capacity Model, which posits that individuals with greater socioeconomic resources are more likely to develop positive self-concepts, which in turn promote adaptive health behaviors.

The relationship between objective SES and exercise identity is moderated by gender. Existing literature highlights the influence of societal gender role expectations on physical activity, particularly in Asian contexts [84]. For instance, Asian women may be discouraged from outdoor physical activity due to cultural preferences for fair complexion [85], and prevailing masculine norms that limit women’s participation in such activities [56]. Conversely, regular physical activity is often perceived as a marker of affluence and education among men, particularly in Asian societies [86]. This societal expectation may lead wealthier men to internalize an active identity, resulting in greater engagement in physical activities. Furthermore, traditional Chinese culture emphasizes male achievement, social responsibility, and upward mobility, making objective SES indicators (e.g., parental education, household income) particularly salient for men. Consequently, men may be more responsive to socioeconomic resources when it comes to engaging in physical activity. It’s worth noting that gender stereotypes continue to influence the sports domain among Chinese university students [87]. Female students often encounter varying degrees of gender-based discrimination when participating in physical activities [88].

Subjective SES, by contrast, captures perceived relative standing and relies more on self‑evaluation and social comparison. Among college students, gender differences in perceived status are typically attenuated, so gender is less likely to moderate subjective‑SES effects on physical activity. In line with that, our findings suggest that gender significantly moderates the relationship between objective SES and physical activity, whereas it does not alter the association between subjective SES and physical activity. Globalization may be contributing to shifts in younger generations’ values and perceptions of gender roles. Future research should further examine the influence of cultural factors in shaping gender-based exercise identity.

Implications

Previous research suggests the important roles of objective SES and subjective SES in physical activity [10, 89, 90]. This research provides further evidence for the distinct roles of objective SES and subjective SES in shaping physical activity, suggesting that the two constructs may impact physical activity through different pathways. Specifically, our findings extend self-identity theory by showing that exercise identity is not only shaped by individual cognition but also by socioeconomic resources and gender differences, thereby highlighting the broader sociocultural context in which identity formation occurs. This research contributes to the literature on socioeconomic disparities by revealing that subjective SES is positively and consistently associated with both exercise identity and physical activity, whereas the mechanisms linking objective SES to physical activity vary by gender. Moreover, by distinguishing the roles of subjective and objective SES, this study highlights the multidimensional nature of socioeconomic influences on health behavior and underscores the need for theory to account for both material resources and subjective perceptions when explaining identity‑based health outcomes.

Prior research has suggested that social comparison nudges, interventions that provide individuals with information about others’ performance, effectively encourage physical activity [91–93]. Building on this, social comparison nudges could be implemented through digital platforms (e.g., TikTok, Instagram, or Facebook) by sharing peers’ exercise achievements or progress. Such strategies may help motivate students to engage in and sustain regular physical activity. Moreover, interventions aimed at promoting physical activity among college students should account for both SES and gender differences. For example, students from lower objective SES backgrounds may benefit from financial support measures, such as subsidized gym memberships or free access to fitness facilities. Female students, who often face additional barriers to participation, could be supported through gender-specific programs that address psychological and social constraints. In addition, to foster the development of exercise identity, universities could implement campus-wide health promotion campaigns that utilize social media platforms to celebrate students’ physical activity achievements. By increasing the visibility of active lifestyles, these programs may help promote the internalization of exercise identity.

Limitations

The present study has some limitations. First, variables were assessed through self-report measures, causal inferences cannot be established. Specifically, physical activity was measured using a single-item indicator, as in prior research [73–76], which may not fully capture its multidimensional nature. Similarly, objective SES was assessed solely based on income. The use of single-item measures may be influenced by psychological biases, such as mood, social desirability, and individual interpretation. This may help explain the relatively low correlations across waves, especially between SES and psychological outcomes. Second, our research recruited only college students from mainland China, and excluded 13 for incomplete data. Although the attrition analysis revealed no significant differences in study variables between participants with complete and incomplete data, 13 participants were excluded due to missing responses. Given that physical activity patterns vary across age groups [94], caution is needed when generalizing these results to other populations or cultural contexts. Third, there are possible cultural differences in the social concepts of masculinity and femininity. For instance, in traditional Confucian Chinese culture, femininity is often associated with gentleness rather than strength [95], which may shape gendered experiences of physical activity and related psychosocial factors. Such cultural considerations may limit the cross-cultural applicability of the findings. Finally, this study focused solely on daily exercise and examined the mediating role of exercise identity. However, other forms (e.g., occupational physical activity) may be linked to different psychological and behavioral outcomes. Future research should replicate these findings across other types of physical activity and explore additional social and psychological factors that may serve as mediators.

Future directions

First, future research could employ longitudinal designs to examine the causal relationships among objective and subjective SES, exercise identity, and physical activity. Second, multiple approaches to measuring physical activity and SES could be adopted. For instance, subjective physical activity could be assessed using the International Physical Activity Questionnaire, whereas objective physical activity could be captured via accelerometers. Objective SES indicators could also be expanded to include multiple dimensions, such as parental education and neighborhood-level SES. Third, future studies should be conducted across diverse populations and cultural contexts to assess the generalizability of the findings. Fourth, although gender differences in physical activity shaped by regional cultures remain insufficiently understood, future research could replicate or extend the present findings across various cultural settings to further elucidate the cultural influences on physical activity among men and women. Additionally, future research should consider implementing strategies such as participant incentives, regular reminders, and flexible methods of data collection to minimize dropout rates. Finally, future research should include diverse forms of physical activity and integrate motivational variables (e.g., intrinsic motivation) within a self-determination theory [96, 97].

Conclusion

Subjective socioeconomic status (SES) was found to be significantly associated with physical activity, whereas objective SES did not demonstrate a significant direct effect. However, exercise identity emerged as a significant mediator in the relationships between both objective and subjective SES and physical activity, indicating that individuals with higher SES, whether perceived or actual, are more likely to engage in physical activity through stronger identification with the exerciser role. Importantly, gender moderated the mediation effect of objective SES on physical activity via exercise identity. This mediation pathway was significant among men, but not among women, suggesting that the effects of objective SES on physical activity for males.

Abbreviations

SES

Socioeconomic status

T1

Time 1

T2

Time 2

T3

Time 3

Authors' contributions

CY: Conceptualization, Supervision, Project administration, Writing – review & editing; LW: Validation, Data curation; YY: Validation, Data curation; LZ: Conceptualization, Supervision, Funding acquisition, Project administration, Writing – review & editing; WZ: Formal analysis, Methodology, Visualization, Writing – original draft.

Funding

We acknowledge financial support from The Science and Technology Development Fund (FDCT: 0062/2024/RIB2).

Data availability

The datasets and code used in the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Macau University of Science and Technology (Approval Number: MUST-20240412001). Informed consent was obtained from all participants involved in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.