The relationship between exercise identity and exercise behavior among Chinese college students: the moderating role of weight bias.

doi:10.21203/rs.3.rs-2947661/v1

Download PDF

Article

The relationship between exercise identity and exercise behavior among Chinese college students: the moderating role of weight bias.

https://doi.org/10.21203/rs.3.rs-2947661/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective: Based on the deviance regulation theory, this paper aimed to explore the relationship between exercise identity, and adaptive and maladaptive exercise behavior, as well as its underlying mechanism among Chinese college students.

Method: A total of 1300 college students (mean aged: 18.75±3.2, 50% male) were investigated with Exercise Identity Scale, Physical Activity Rating Scale, Exercise Dependence Scale, and Weight Bias Questionnaire. SPSS27.0 and Mplus8.3 were used to conduct the statistical analysis.

Results: There were significant differences in adaptive and maladaptive exercise behaviors among college students of different genders (t=14.66, p =0.00; t=12.55, p=0.00).

Results: There were significant differences in adaptive and maladaptive exercise behaviors among college students of different genders (F=12.11, p=0.00; F=3.80, p =0.02; F=22.53, p =0.00; F=11.04, p=0.00). Among college students, ' exercise identity, weight bias and adaptive exercise behavior (r=0.50, p＜0.01; r=0.33, p＜0.01) and its dimensions, and maladaptive exercise behavior (r=0.51, p＜0.01; r=0.36, p＜0.01)and its dimensions were significantly positively correlated. Weight bias contributed to a positive moderating role in college students' exercise identity and adaptive exercise behaviors, as well as in maladaptive exercise behaviors.

Conclusion: Exercise identity can positively predict adaptive and maladaptive exercise behaviors among Chinese college students, in which weight bias contributes to a positive moderating role.

Biological sciences/Psychology

Biological sciences/Psychology/Human behaviour

exercise identity

weight bias

adaptive and maladaptive exercise behavior

Chinese college students

Physical exercise as an important section of a healthy lifestyle has become a public consensus (Li & Ma 2011). Previous studies have found that participation in regular physical exercise can promote the development of personal physical and mental health (Taghian et al. 2014), and bring quite a few health benefits to life, study, and work (Kruk J., 2009). Therefore, as a group of college students who promote the future development of society, it is of great significance to form good adaptive exercise behavior habits for their future development. Meanwhile, in recent years, there have also been some excessive exercise behaviors that are not conducive to physical and mental health among college students, sometimes called exercise dependence behaviors (Li et al., 2015). Veale first proposed the concept of exercise dependence in 1987, arguing that people with exercise dependence behavior often take physical exercise as one of the important contents of their daily life. Once regular physical exercise is stopped, adverse physical and mental symptoms such as insomnia, anxiety, loss of appetite, and irritability, similar to those caused by withdrawal from drugs, can occur (Veale, 1987). Accordingly, it is of practical significance to explore the factors of Chinese college students' adaptive and maladaptive exercise behavior and to intervene in these factors.

Deviance regulation theory (Blanton & Christie, 2003; Blanton, Stuart, & Van den Eijnden, 2001) suggests that an individual’s identity and relevant social norms (weight bias) influence behavioral decisions. Individuals are motivated to be unique when doing so is associated with positive evaluations, but also are motivated to not stand out when standing out may be perceived negatively. Based on the deviance regulation theory, adaptive and maladaptive health behavior intentions such as hand washing (Hall & Blanton, 2009), marijuana use (Dvorak et al., 2018), alcohol use (Ferrer et al., 2012), and condom use (Blanton, VandenEijnden, et al., 2001) can be predicted. Several researchers have developed some alcohol dependence intervention programs based on social norms, and achieved positive intervention effects (Dvorak et al., 2016, 2018). So far, deviance regulation theory has not been applied to the field of study of physical exercise behavior (adaptive or maladaptive exercise behavior). However, the theory suggests that individuals with an exercise identity will behave in a manner consistent with their identity, that is, actively participating in physical activity, and may gain approval from others. Research evidence of exercise psychology shows that factors affecting college students' physical exercise behavior mainly include exercise identity (Murray et al., 2013), weight bias (Palermo et al., 2021), exercise motivation (Li, 2018), psychological needs (Li, 2018), and social fitness anxiety (Cook et al., 2015). Therefore, based on the theoretical framework of deviance regulation theory, this study attempts to explore the relationship between college students' exercise identity, weight cognitive bias and physical exercise behavior.

Exercise identity is defined as the extent to which a person observes exercise as part of who they are. Exercise identity is associated with exercise behaviors broadly, including greater frequency of exercise (Strachan et al., 2009; Verkooijen & de Bruijn, 2013), exercise effort (Anderson et al., 1998), and commitment to exercising (Lu et al., 2012). However, these studies mainly reflect adaptive exercise behaviors from exercise results, that is to say, these studies are based on the premise that exercise promotes physical and mental health. Only a few studies have investigated the association between exercise identity and maladaptive exercise behavior (exercise dependence behavior), and exercise identification is a risk factor for exercise dependence (Allegre et al., 2006). Individuals with a strong exercise identity are more likely to be motivated to engage in exercise behaviors and may be at particular risk for maladaptive exercise (Murray et al., 2013). Exercise identity is central to exercise-dependent behavior. Maladaptive exercise behaviors can both reinforce exercise identity, as well provide the opportunity to present as unique in a way that generates positive evaluations from others (Murray et al., 2013). For example, regular exercisers are considered to be more physically attractive, fitter, stronger, and healthier than non-exercisers (Ginis et al., 2003). Furthermore, some studies suggest that exercisers are perceived as more independent, intelligent, friendlier, and braver than non-exercisers (Mack, 2003). These studies suggest that exercisers have benefits in self-presentation that may be heightened for individuals who hold exercise as central to their identity.

One norm-based attitude that may be relevant to whether exercise identity leads to adaptive or maladaptive exercise behaviors is weight bias. Weight bias is defined as negative attitudes directed toward individuals who are perceived to have excess body weight (Forbush et al., 2013). Although the relationship between exercise identity and weight bias is ambiguous, individuals who hold exercise as an important aspect of their self-identity (e.g., personal trainers) are documented to demonstrate bias towards individuals with excess weight. For example, personal trainers may discriminate against overweight clients and be reluctant to coach overweight exercisers (Flint & Reale, 2018; Ntoumanis et al., 2018). Notably, fitness professionals and personal trainers also hold weight biases for the average exercisers (Flint et al., 2015; Robertson & Vohora, 2008), and regular exercisers have a high sense of exercise identity (Anderson & Cychosz, 1995). Taken together, it can be inferred that people with high exercise identity have weight bias, which can enhance the association between exercise identity and adaptive or maladaptive exercise behavior.

In summary, this study based on the deviance regulation theory to investigates the relationship between college students’ exercise identity, weight bias and adaptive and maladaptive exercise behaviors, focus on exploring the moderating role of Chinese college students’ weight bias between exercise identity and adaptive and maladaptive exercise behavior.

2.1 Data Collection and Participants

In this study, a questionnaire survey was used to collect data, and a convenient sampling method was applied to investigate college students in a university in Jiangsu Province. This cross-sectional study issued 1300 questionnaires, with the recovery of 1200, and excluded 37 invalid questionnaires (effective recovery rate = 98.6%). A total of 1183 respondents were included, including 592 males (50.0%) and 591 females (50.0%). There were 322 freshmen (27.2%), 316 sophomores (26.7%), 301 juniors (25.4%) and 244 seniors (20.6%). Informed consent was obtained from all subjects participating in the study, as well as from their legal guardians where applicable. Prior to their involvement, participants and/or their legal guardians were provided with a detailed explanation of the study objectives, procedures, potential risks, and benefits. They voluntarily agreed to participate and were assured of their right to withdraw from the study at any time without consequence.

2.2 Measures

2.2.1 Demographic information

College student were required to self-report demographic information. The demographic questionnaire compiled in this study included the following four parts: gender, grade, as well as height and weight which were used to calculate Body mass index (BMI).

2.2.2 Physical Activity Rating Scale (PARS-3)

The measurement of adaptive exercise behavior in this study was based on the Physical Exercise Rating Scale (PARS-3) revised by Deqing et al (1994). From the intensity of physical exercise, exercise time (a duration) and exercise frequency (several times a week) three aspects to examine the level of exercise participation was measured by the amount of adaptive exercise behavior. Three dimensions from 1 to 5 levels were recorded 1 to 5 points, physical exercise scale score = exercise intensity * (exercise duration − 1) *exercise frequency. The highest score is 100 points, and the lowest score is 0 points. The criteria to divide the amount of physical exercise are low exercise level ≤ 19 points, medium exercise level 20 ~ 42 points, and high exercise level ≥ 43 points. In this study, the Cronbach coefficient of PARS-3 is 0.669, which demonstrated its high reliability.

2.2.3 Exercise Dependence Scale-Revised Chinese version (EDS-C)

This study used the exercise dependence scale-revised Chinese version (EDS-C) revised by Peiying Yang et al (Yang, Yu, Montag, Becker, & Zou, 2021) to measure maladaptive exercise behavior. EDS-C consists of the following seven dimensions, each consisting of three items. Each dimension are (i) tolerance (e.g., 'I will continue to increase exercise intensity in order to achieve the desired effect / benefit'), (ii) withdrawal effect (e.g., 'I avoid anxiety through exercise'), (iii) intention effect (e.g., 'I will hurt exercise'), (iv) lack of control (e.g., 'I cannot lower my exercise frequency'), (v) reduction of other activities (e.g., 'I would rather exercise than with my family / friends'), (vi) time (e.g., 'I spend most of my free time on exercise') and (vii) continuance (e.g., 'I actually exercise longer than I expected'). Using the Likert 6 level measure, from ' never ' to ' always' 1–6 level score, each standard can get a maximum of 18 points. The higher the total score on the scale and the average score of the subscale, the higher the degree of participation in the dependence on exercise. In this study, the Cronbach 's alpha coefficient of EDS-C was 0.961, and the reliability index was good, which could be used as an effective measurement tool for college students' exercise dependence behavior.

2.2.4 Exercise identity

Exercise Identity Scale (EIS) was developed by Anderson and Cychosz (1994), and tested by Li Menglong et al. (Li Menglong, Yang Jiao & Ren Yujia, 2019). The Chinese version of EIS is feasible in Chinese college students and provides a reliable measurement tool for the localization of exercise identity. The Exercise Identity Scale (EIS) consisted of 9 items. The factor analysis showed a single-dimensional structure. The Likter 7-level scale was used for scoring, from 'completely disagree' to 'completely agree' 1–7 scores. Each fit index of EIS is good (χ²/df = 3.71, RMESA = 0.078, CFI = 0.907, IFI = 0.981), and Cronbach 's alpha coefficient is 0.930. To evaluate and measure whether the individual is a physical exerciser, the higher the score, the higher the degree to which the individual is an exerciser.

2.2.5 Weight bias questionnaire

This study used a 5-item subscale of the Eating Pathology Inventory (ESPI) for negative attitudes toward obesity (α = 0.89). The ESPI is a 45-item self-report that comprehensively assesses clinically relevant symptoms of eating disorder psychology (Forbush et al., 2013). Items are rated on a scale from 1 (never) to 5 (very often), reflecting the importance of each statement over the past 4 weeks. Among them, the negative attitude towards obesity has only 5 questions but can show good psychometric properties and effectiveness, and is suitable for adults of any weight range. The results of the validation factor analysis showed good fit (χ²/df = 3.67, RMESA = 0.069, CFI = 0.994, IFI = 0.994) with a Cronbach's coefficient of 0.898. Higher scores represent higher levels of weight bias among college students. Therefore, the subscale of negative attitudes toward obesity from the Eating Pathology Scale was used in this study to measure attitudes toward weight perception bias in the college student population.

2.3 Data analysis

The data obtained from all scales were reviewed, and SPSS27.0 was used to describe and analyze the relevant data. Then Mplus8.0 is used to fit the structural equation model, which can test the adjustment effect of latent variables and analyze the relationship between multiple latent variables. The LMS method was used to explore the correlation analysis and interaction effect analysis of each research variable, and a simple slope diagram of the latent variable interaction model was made. The slope was divided into three levels: high level (+ 1SD), medium level (= 1SD) and low level (-1SD).

2.4 Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files. We confirm that all methods described in our study were performed in accordance with these guidelines. The experimental protocols were approved by the Yangzhou University.

3.1 Common method bias test

In view of the application of self-reported exercise identity, weight bias, exercise behavior and exercise dependence scales in this study, in order to avoid the common method bias that may occur when measuring the same group of subjects at the same time, it is necessary to test the data obtained by Harman single factor method. The Bartlett's test of spherical (χ² = 1258.90, df = 613, p＜0.001) and KMO measure (KMO = 0.964) showed that this study was suitable for factor analysis. The Principal Component Analysis revealed 5 distinct factors with eigenvalues greater than 1.0. The variance interpretation rate of the first common factor is 31.22%, which is less than 40% of the critical standard. Therefore, there was no serious common method bias problem in this study.

3.2 Descriptive statistics

The results of Table 1 using an independent sample t test showed that show that there was no significant difference in exercise identity and weight bias between college students of different genders, but there was a significant difference in adaptive and maladaptive exercise behavior (t = 14.66, p = 0.00; t = 12.55, p = 0.00). Male college students scored higher than female college students on adaptive and maladaptive exercise behaviors. The moderate exercise behavior and exercise dependence behavior of male college students were at the middle level, and the female college students were at the lower middle level. This indicates that male college students had better moderate exercise behaviors than female college students, and the risk of developing maladaptive exercise behaviors (exercise dependence behavior) was also higher than female college students.

Table 1

Gender differences in college students’ exercise identity, weight bias, adaptive and maladaptive exercise behaviors
Variable	Male (N = 592)	Female(N = 591)	t	Cohen d	p
Variable	M ± SD	M ± SD	t	Cohen d	p
Exercise identity	47.71 ± 10.56	37.17 ± 10.49	12.32^**	0.72	0.37
Weight bias	16.06 ± 5.48	14.04 ± 5.64	6.27^**	0.37	0.16
Adaptive exercise	52.79 ± 27.50	31.43 ± 22.35	14.66^**	0.85	0.00
Maladaptive exercise	67.20 ± 25.67	50.24 ± 20.54	12.55^**	0.73	0.00
Note: ^p＜0.05;^*p＜0.01

BMI is an important index to reflect the individual’s physique and body shape. Its calculation formula is BMI = weight (kg)/height (m)². Refer to the recommended standards of ‘Chinese Adult Overweight and Obesity Prevention and Control Guidelines’, underweight was identified as BMI < 18.5 kg/m², normal was defined as 18.5 ≤ BMl＜24kg/m², overweight was established as 24 ≤ BMI＜28 kg/m², and obese was determined as BMI ≥ 28 kg/m². The results of Table 2 using one-way analysis of variance showed that college students with different BMI had significant differences in exercise identity, weight bias and adaptive and maladaptive exercise behavior (F = 12.11, p = 0.00;F = 3.80, p = 0.00; F = 22.53, p = 0.00; F = 11.04, p = 0.00). College students with normal (41.77 ± 11.10; 47.03 ± 27.68; 54.82 ± 24.58) and overweight (42.10 ± 9.79; 48.20 ± 27.75; 50.22 ± 22.99) BMI had higher levels of exercise identity, moderate exercise behavior, and exercise dependent behavior than those college students with underweight BMI (36.37 ± 11.51; 33.85 ± 20.80; 50.22 ± 22.99) and obese (37.68 ± 11.93; 40.57 ± 18.88; 50.22 ± 22.99). In terms of weight bias, college students with normal (15.36 ± 5.83) and overweight (14.96 ± 4.92) BMI were higher than those with underweight (13.65 ± 5.56) BMI.

Table 2

Differences in BMI among college students’ exercise identity, weight bias, adaptive and maladaptive exercise behavior
Variable	underweight(N = 133)	normal(N = 790)	overweight(N = 181)	obese(N = 79)	F	LSD
Variable	M ± SD	M ± SD	M ± SD	M ± SD	F	LSD
Exercise identity	36.37 ± 11.51	41.77 ± 11.10	42.10 ± 9.79	37.68 ± 11.93	12.11^**	2, 3＞1, 4
Weight bias	13.65 ± 5.56	15.36 ± 5.83	14.96 ± 4.92	14.53 ± 5.26	3.80^**	2, 3＞1
Adaptive exercise	33.85 ± 20.80	47.03 ± 27.68	48.20 ± 27.75	40.57 ± 18.88	22.53^**	2, 3＞1, 4;
Maladaptive exercise	44.32 ± 18.94	54.82 ± 24.58	52.84 ± 22.94	50.22 ± 22.99	11.04^**	2, 3＞1, 4
Note: ^p＜0.05;^*p＜0.01

3.3 Correlation analysis

The results of Table 3 using Pearson correlation analysis showed that college students’ exercise identity, weight bias and adaptive exercise behavior (r = 0.50, p＜0.01; r = 0.33, p＜0.01) and each dimension (exercise intensity, exercise duration, exercise frequency) had significant positive correlation. Similarly, college students’ exercise identity, weight bias and maladaptive exercise behavior (r = 0.51, p＜0.01; r = 0.36, p＜0.01) and its seven dimensions (withdrawal effect, persistence, tolerance, lack of control, reduction of other activities, time, and intention effect) also had significant positive correlation.

3.4 Test the moderating effect

This study uses the LMS method of analysis of latent variable adjustment effect to test (J. Fang, et al., 2022). Under the control of gender, grade, major and BMI demographic variables, the moderating effects of weight bias on exercise identity and adaptive and maladaptive exercise behavior were tested respectively.

Table 3

The correlation analysis results of college students’ exercise identity, weight bias, adaptive and maladaptive exercise behavior and its various dimensions
Variable		1	2	3	4	5	6	7	8	9	10	11	12	13	14
1 Exercise identity		1
2 Weight bias		0.23^**	1
3 Adaptive exercise		0.50^**	0.33^**	1
4 Maladaptive exercise		0.51^**	0.36^**	0.57^**	1
Adaptive exercise	5 exercise intensity	0.45^**	0.27^**	0.80^**	0.48^**	1
	6 exercise duration	0.39^**	0.18^**	0.71^**	0.39^**	0.48^**	1
	7 exercise frequency	0.30^**	0.24^**	0.61^**	0.33^**	0.26^**	0.19^**	1
Maladaptive exercise	8 withdrawal effect	0.45^**	0.24^**	0.44^**	0.83^**	0.39^**	0.31^**	0.24^**	1
	9 persistence	0.42^**	0.38^**	0.54^**	0.90^**	0.45^**	0.35^**	0.33^**	0.66^**	1
	10 tolerance	0.49^**	0.33^**	0.52^**	0.91^**	0.45^**	0.35^**	0.30^**	0.77^**	0.78^**	1
	11 lack of control	0.46^**	0.35^**	0.49^**	0.92^**	0.41^**	0.33^**	0.29^**	0.69^**	0.83^**	0.82^**	1
	12 reductions in other acticities	0.42^**	0.36^**	0.50^**	0.91^**	0.42^**	0.33^**	0.29^**	0.67^**	0.84^**	0.76^**	0.85^**	1
	13 time	0.50^**	0.35^**	0.56^**	0.93^**	0.46^**	0.39^**	0.32^**	0.70^**	0.83^**	0.81^**	0.85^**	0.86^**	1
	14 intention effect	0.49^**	0.27^**	0.53^**	0.89^**	0.44^**	0.38^**	0.31^**	0.72^**	0.74^**	0.80^**	0.77^**	0.75^**	0.82^**	1
M		41.00	14.99	40.37	58.68	3.31	3.93	3.80	9.41	7.50	9.11	7.98	7.61	8.06	9.01
SD		11.11	5.62	27.33	24.46	1.13	0.96	1.06	3.96	3.89	3.98	3.73	3.82	3.75	4.08
Note: Correlation is significant at the 0.05 level (two-tailed); *Correlation is significant at the 0.01 level (two‐tailed).

As shown in Table 4, comparison between the Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) of the models showed that the model with the interaction term had lower AIC and BIC (model with interaction: AIC = 135574.888, BIC = 135864.847; model without interaction: AIC = 135609.83, BIC = 136006.305), thus providing the best fit to the data (Akaike, 1974). At the same time, Loglikelihood =-67737.915 for the SEM model with interaction terms, which is 4.529 more than Loglikelihood (H0) for the benchmark SEM model (greater than standard 4.41), 0.03 reached the significance of the chi-square test. The results show that the SEM model with interaction terms has better fitting than the SEM model without interaction.

Table 4

Comparison results between the baseline model and the interaction item model (adaptive exercise behavior)
Fit index	Model without interaction	Model with interaction	Analysis result	Does it pass
Akaike (AIC)	135609.83	135574.89	34.942	pass
Bayesian (BIC)	136006.31	135864.85	141.46	pass
Loglikelihood(H0)	-67742.44	-67737.92	0.003	pass
Free parameters	60	61

The path coefficient of the product term of exercise identity and weight perception bias (XW→Y)was 0.060. The level of statistical significance is p = 0.026, which meets the criterion of significance (p＜0.05). The results showed that weight perception bias had a positive moderating effect on the relationship between exercise identity and adaptive exercise behavior (see Fig. 1).

The weight bias (moderating variable w) was fixed at 1 to standardize it. After constructing the latent variable interaction term (xw), the coefficients of the independent variable (x), moderator variable (w) and interaction term (xw) are defined as b1, b2, b3, respectively, where the regression coefficient b3 of the interaction term (xw) is significant or not is used to determine whether the moderating effect is significant or not. As shown in Table 5, when the regulating variables are specified as low (LOW_w), medium (MED_w), and high (HIGH_w), the simple slope values are 0.163, 0.173, and 0.184 respectively.

Table 5

Variation of the simple slope for different values of the moderating variable w (adaptive exercise behavior)
Group	Simple slope	S.E.	t
SIMP_LO(Low Group)	0.163	0.015	10.727
SIMP_MED(Medium Group)	0.173	0.011	15.375
SIMP_HI(High Group)	0.184	0.015	12.073

Taking exercise identity as independent variable, adaptive exercise behavior as dependent variable and weight bias as moderating variable. As shown in Fig. 2, weight bias has a positive moderating effect on exercise identity and adaptive exercise behavior. Simple slope plot analysis showed that adaptive exercise behavior increased as exercise identity increased, and the trend of high weight bias was the same as low weight bias. When the level of weight bias is high (W = + 1SD), exercise identity has a strong positive effect on adaptive exercise behavior (B = 0.184, p＜0.01). When the level of weight bias was low (W=-1SD), exercise identity had a positive effect on adaptive exercise behavior, but the effect was small (B = 0.163, p＜0.01).

As shown in Table 6, the comparison between the Akaike Information Criteria (AIC) of the models and Bayesian Information Criteria (BIC) showed that the model with the interaction term had lower AIC and BIC (model with interaction: AIC = 86128.10, BIC = 86392.04; model without interaction: AIC = 86150.29, BIC = 86525.90), thus providing the best fit to the data (Akaike, 1974). At the same time, Loglikelihood=-43001.14 for the SEM model with interaction terms, which is more than Loglikelihood (H0) for the benchmark SEM model (greater than standard 4.41), 0.00 reached the significance of the chi-square test. The results show that the SEM model with interaction terms has better fitting than the SEM model without interaction.

Table 6

Comparison results between the baseline model and the interaction item model (maladaptive exercise behavior)
Fit index	Model without interaction	Model with interaction	Analysis result	Does it pass
Akaike (AIC)	86150.29	86128.10	22.19	pass
Bayesian (BIC)	86525.90	86392.04	133.85	pass
Loglikelihood(H0)	-43012.05	-43001.14	0.000	pass
Free parameters	72	74

The path coefficient of the product term of exercise identity and weight perception bias (XW→Y)was 0.160. The level of statistical significance is p = 0.000, which meets the criterion of significance (p < 0.001). The results showed that weight bias had a positive moderating effect on the relationship between exercise identity and maladaptive exercise behavior (exercise dependence) (see Fig. 3).

Similar to the simple slope to test adaptive exercise behavior, whether the coefficient b3 of the interaction term (xw) is significant is significant to judge whether the moderating effect is significant. The values of the specified moderating variables are low (LOW_w=-1SD), medium (MED_w = 1SD), and high (HIGH_w = + 1SD), and the simple slopes are 0.832, 1.230, and 1.628 when the moderating variables are valued as low, medium, and high, respectively, as shown in Table 7.

Table 7

Variation of the simple slope for different values of the moderating variable w (maladaptive exercise behavior)
Group	Simple slope	S.E.	t
SIMP_LO(Low Group)	0.832	0.095	8.800
SIMP_MED(Medium Group)	1.230	0.075	16.355
SIMP_HI(High Group)	1.628	0.106	15.372

Exercise identity was the independent variable, exercise dependence behavior was the dependent variable, and weight bias was the moderating variable. As shown in Fig. 4, weight bias has a positive moderating effect on exercise identity and exercise dependence behavior. Simple slope plot analysis showed that exercise dependence behavior increased as exercise identity increased, and the trend of high weight bias was the same as low weight bias. When the level of weight bias is high (W = + 1SD), exercise identity has a strong positive effect on exercise dependence behavior (B = 1.628, p＜0.01). When the level of weight bias was low (W=-1SD), exercise identity had a positive effect on exercise dependence behavior, but the effect was small (B = 0.832, p＜0.01).

4.1 Difference in demographic variables

In this study, Chinese college students of different genders had significant differences in adaptive and maladaptive exercise behaviors. The adaptive exercise behavior of male students (52.79 ± 27.50) was better than that of female students (31.43 ± 22.35) and the maladaptive exercise behavior of male students was more serious (67.20 ± 25.67) than that of female students (50.24 ± 20.54). This is consistent with the research results of Dong et al. (2020) and Qu et al. (2017). There were significant differences in the body between men and women, leading to significant differences in the athletic ability of college students of different genders. Male and female college students had obvious gender characteristics in sports project selection. In terms of content selection, male students tended to choose more intense and competitive events, while female students preferred non-direct confrontation, bright rhythm, relaxed and happy, games and entertainment strong and high aesthetic value of sports. Influenced by the campus sports atmosphere, schools generally carry out sports such as soccer, basketball, volleyball and athletics, which mainly focus on physical exercise. These sports are more in line with the needs of male exercise behavior, while the girls like aerobics, Latin dance and other programs to carry out less, resulting in the exercise behavior of male students to generally better than female students. According to the research of Qu et al. (2015), a large amount of physical exercise is one of the characteristics of college students' maladaptive exercise behavior, that is, the more college students exercise, the easier to form exercise dependence behavior. Similarly, Zhang's study (2012) proposed that the exercise frequency and exercise duration of exercise dependence among Chinese college students in the high group were significantly higher than those in the low group. The results of this study show that male college students generally had higher exercise intensity, duration, and frequency than female college students in exercise behavior, so male students were more likely to form exercise dependence behavior compared to female students.

There were significant differences in exercise identity and weight bias among Chinese college students across body indices. This is slightly different from the study by Karr et al. (2014), which showed that a higher BMI was associated with a lower exercise identity. Given the extent to which individuals with higher BMI engage in exercise behaviours (Tucker et al., 2011), it is normal for this population to have lower levels of exercise identity. However, it is important to consider that many people with a strong exercise identity (such as athletes or bodybuilders) also have a higher BMI due to increased muscle mass (Jonnalagadda et al., 2004), resulting in a high BMI associated with a high exercise identity. Gumble and Carels’ study (2012) echoed the findings of this study and concluded that weight bias can be observed in individuals of all weight status (i.e., normal, overweight, and obese). Wen Yancheng et al. (2018) showed the same results as this study, suggesting that overweight and obese individuals have a higher level of weight bias than lean individuals. According to the present findings, weight bias of individuals with various BMIs can be reduced by promoting college students' exposure to body diversity and normalizing college students' cognition of various body shapes (Stewart & Ogden, 2020). In addition, providing education about the multiple determinants of individual weight and reducing negative attitudes toward obesity or overweight among college students (Diedrichs & Barlow, 2011) could reduce weight bias among college students of all BMIs.

There were also significant differences in the adaptability of Chinese college students’ maladaptive exercise behavior, especially college students whose BMI was in the lean range. They believe that their body shape conforms to the body shape promoted by modern mainstream media, and they are less active in exercising. The exercise intensity, duration and frequency are less than those of college students in other BMI ranges. Phelan et al. (2011) concluded that overweight and obese individuals scored significantly higher on exercise-dependent behaviors and its subscales than individuals with BMI normal. The reason is that BMI overweight and obese people want to rely on exercise behavior to lose weight and maintain normal weight, while BMI is the "unhealthy" motivation of normal weight individuals to rely on exercise behavior to achieve "pleasure" or "stimulation" through greater exercise. Therefore, the main reason for college students in different BMI ranges to distinguish whether they have maladaptive exercise behaviour is to generate a higher level of exercise motivation to control weight or to demand "unhealthy" aspects of exercise-dependent behavior.

4.2 Research on the Correlation Between Various Variables

This study analyzed the correlation between exercise identity, weight bias and adaptive and maladaptive exercise behavior of college students. The results showed that exercise identity and weight bias were positively correlated with three dimensions of exercise intensity, duration and frequency in moderate exercise behavior, and also positively related to the seven dimensions of exercise dependence behavior. It is fully verified that the identity and relevant social norms (weight bias) in deviance regulation theory could better predict adaptive and maladaptive health behaviors. This result is consistent with previous research results (Palermo et al, 2021). For example, the research of Dong Baolin and Mao Lijuan (2020) concluded that the higher the degree of college students' exercise identity, the more reasonable the exercise behavior, which proves that the exercise identity has a positive promoting effect on college students' exercise behavior. Zaroubi et al. (2021) showed that college students' emphasis on exercise behavior in education program will enhance their own weight bias level. Popular societal messages of blame and personal responsibility for weight may explain the prevalence and rise among weight bias of college students. In addition, the emphasis on the controllability of weight and the rigid concept of "eat less, exercise more" may lead to higher levels of weight bias among college students (Teachman et al., 2003). College students with a strong exercise identity and higher levels of weight bias self-reported that they engaged in behaviors that were consistent with their sense of identity and were considered socially acceptable deviant behaviors. This finding is consistent with previous findings that people with higher levels of identity in exercise and weight cognitive biases were at higher risk of maladaptive exercise behaviors. Therefore, it is important to include assessment of exercise identity and weight bias in general exercise interventions as well as prevention interventions for maladaptive exercise behaviors, so individuals at higher risk for maladaptive exercise behavior can be identified.

4.3 Test the moderating effect

Through the analysis of the moderating effect of latent variables, the study results showed that weight bias positively moderates the relationship between exercise identity and moderate exercise behavior, and the higher the level of weight bias, the greater impact of exercise identity on adaptive exercise behavior, which is consistent with deviance regulation theory. Weight bias is a negative attitude held by overweight people, and its essence is an internal stereotype. For college students with a high weight bias about themselves, they are afraid that they will be discriminated against by the same weight bias, which will deepen the degree of self-exercising identity, thereby promoting their own exercise behavior. In the university campus environment, one common intervention strategy for promoting adaptive health behaviors is based on social norms theory (Ajzen, 1991). Some studies believe that manipulating peer norms to lead people to believe that their peers were engaging in physical activity or exercise behavior was effective for increasing their own exercise behavior (Crozier & Spink, 2017; Wally & Cameron, 2017). Consistent with deviance regulation theory, when positive social recognition is achieved, individuals may deviate from low descriptive norms (i.e., a few people are physically active) and were more likely to conform to high descriptive norms related to exercise behavior in order to avoid any negative social identity consequences (Crozier & Taylor, 2019). This indicates that when college students' exercise identity brings positive significance to themselves, they will abide by the high descriptive norms (weight bias) related to exercise behavior and promote their own exercise behavior.

The results of the study also showed that weight bias positively moderates the relationship between exercise identity and exercise dependence behavior, and the higher the level of weight bias, the greater impact of exercise identity on maladaptive exercise behavior. Although this finding is interpreted as reflecting socially acceptable deviant behavior, this interpretation may only be related to specific social environment or social norms. Indeed, deviance regulation theory suggests that an individual's social reference group is critical to whether a behavior will be perceived as positively deviant and subsequently promotes specific behavioral intentions (Blanton & Christie, 2003). In the current study, maladaptive exercise behaviors may only be perceived as “positively” deviant in the presence of negative social norms associated with overweight or obese status. There is ample evidence to support this this interpretation. For example, data were collected in the USA where there is an increased emphasis on fitness culture (McKenzie, 2013) and the exercise dependence behavior was positively regarded as an expected deviant behavior. Nevertheless, more detailed research is needed on the application of deviation regulation theory in maladaptive behavior, including the measurement of social norms and social reference groups.

Most domestic research on college students is adaptive exercise behaviors that promote the development of college students' physical and mental health, while often ignores the maladaptive exercise behaviors (exercise dependence behavior) formed by excessive exercise. Maladaptive exercise behaviors have a detrimental effect on college students’ physical and mental health. People with a strong exercise identity and high levels of weight bias may easily develop maladaptive exercise behavior. Therefore, the exercise intervention biased toward social norms and the exercise prevention intervention with maladaptability should focus on the impact of exercise identity and weight bias. This can provide a research basis for schools and other relevant education departments to reduce the weight bias of college students, clarify their own exercise identity and exercise motivation, promote college students to form healthy and adaptive exercise behavior, and reduce maladaptive exercise behavior.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211.
Allegre, B., Souville, M., Therme, P., & Griffiths, M. (2006). Definitions and measures of exercise dependence. Addiction Research, 14(6), 631-646.
Anderson, D. F., & Cychosz, C. M. (1994). Development of an exercise identity scale. Perceptual & Motor Skills, 78, 747–751.
Anderson, D. F., & Cychosz, C. M. (1995). Exploration of the relationship between exercise behavior and exercise identity. Journal of Sport Behavior, 18(3), 159.
Anderson, D. F., Cychosz, C. M., & Franke, W. D. (1998). Association of exercise identity with measures of exercise commitment and physiological indictors of fitness in a law enforcement cohort. Journal of Sport Behavior, 21(3), 233.
Blanton, H., & Christie, C. (2003). Deviance regulation: A theory of action and identity. Review of General Psychology, 7(2), 115–149.
Blanton, H., Stuart, A. E., & Van den Eijnden, R. J. (2001). An introduction to deviance-regulation theory: The effect of behavioral norms on message framing. Personality and Social Psychology Bulletin, 27(7), 848–858.
Blanton, H., VandenEijnden, R. J., Buunk, B. P., Gibbons, F. X., Gerrard, M., & Bakker, A. (2001). Accentuate the negative: Social images in the prediction and promotion of condom use 1. Journal of Applied Social Psychology, 31(2), 274–295.
Cheng, M. Y., Wang, S. M., Lam, Y. Y., Luk, H. T., Man, Y. C., & Lin, C. Y. (2018). The Relationships Between Weight Bias, Perceived Weight Stigma, Eating Behavior, and Psychological Distress Among Undergraduate Students in Hong Kong. The Journal of nervous and mental disease, 206(9), 705–710.
Cook, B., Karr, T. M., Zunker, C., Mitchell, J. E., Thompson, R., Sherman, R., Erickson, A., Cao, L., & Crosby, R. D. (2015). The influence of exercise identity and social physique anxiety on exercise dependence. Journal of behavioral addictions, 4(3), 195–199.
Crozier, A. J., & Spink, K. S. (2017). Effect of manipulating descriptive norms and positive outcome expectations on physical activity of university students during exams. Health Communication, 32(6), 784–790.
Crozier, A. J., & Taylor, K. L. (2019). An exploratory study examining the interactive effect of descriptive norm and image appeal messages on adults’ physical activity intentions: A test of deviation regulation theory. Journal of Health Communication, 24 (2), 195–202.
Diedrichs, P. C., & Barlow, F. K. (2011). How to lose weight bias fast! Evaluating a brief anti-weight bias intervention. British Journal of Health Psychology, 16(4), 846–861.
Dong & Mao (2020). Exercise identity, internal motivation, exercise commitment and college students ' exercise behavior: chain mediation model. Journal of Tianjin University of Physical Education (04), 415-422.
Dvorak, R. D., Pearson, M. R., Neighbors, C., Martens, M. P., Stevenson, B. L., & Kuvaas, N. J. (2016). A road paved with safe intentions: Increasing intentions to use alcohol protective behavioral strategies via Deviance Regulation Theory. Health Psychology, 35(6), 604.
Dvorak, R. D., Raeder, C. A., Kramer, M. P., Sargent, E., Stevenson, B. L., & Helmy, M. (2018). Using deviance regulation theory to target marijuana use intentions among college students. Experimental and Clinical Psychopharmacology, 26(1), 29.
Dvorak, R. D., Raeder, C. A., Kramer, M. P., Sargent, E., Stevenson, B. L., & Helmy, M. (2018). Using deviance regulation theory to target marijuana use intentions among college students. Experimental and Clinical Psychopharmacology, 26(1), 29.
Ferrer, R. A., Dillard, A. J., & Klein, W. M. (2012). Projection, conformity and deviance regulation: A prospective study of alcohol use. Psychology & Health, 27(6), 688–703.
Flint, S. W., & Reale, S. (2018). Weight stigma in frequent exercisers: Overt, demeaning and condescending. Journal of Health Psychology, 23(5), 710–719.
Flint, S. W., Hudson, J., & Lavallee, D. (2015). UK adults’ implicit and explicit attitudes towards obesity: A cross-sectional study. BMC Obesity, 2(1), 31.
Forbush, K. T., Wildes, J. E., Pollack, L. O., Dunbar, D., Luo, J., Patterson, K., … Stone, A. (2013). Development and validation of the Eating Pathology Symptoms Inventory (EPSI). Psychological Assessment, 25(3), 859.
Ginis, K. A. M., Latimer, A. E., & Jung, M. E. (2003). No pain no gain? Examining the generalizability of the exerciser stereotype to moderately active and excessively active targets. Social Behavior and Personality: An International Journal, 31(3), 283–290.
Gumble, A., & Carels, R. (2012). The harmful and beneficial impacts of weight bias on well-being: The moderating influence of weight status. Body Image, 9(1), 101–107.
Hall, D. L., & Blanton, H. (2009). Knowing when to assume: Normative expertise as a moderator of social influence. Social Influence, 4(2), 81–95.
Jonnalagadda, S. S., Skinner, R., & Moore, L. (2004). Overweight athlete: Fact or fiction? Current Sports Medicine Reports, 3(4), 198–205.
Karr, T. M., Bauer, K. W., Graham, D. J., Larson, N., & Neumark-Sztainer, D. (2014). Exercise identity: Healthy and unhealthy outcomes in a population-based study of young adults. Journal of Sport Behavior, 37(2).
Kruk J. (2009). Physical activity and health. Asian Pacific journal of cancer prevention: APJCP, 10(5), 721–728.
Li M. (2018). The Influence of Psychological Needs and Exercise Motivation on Exercise Dependence among Chinese College Students. The Psychiatric quarterly, 89(4), 983–990.
Li Menglong, Yang Jiao & Ren Yujia. (2019). The reliability and validity of the Chinese version of Exercise Identity Scale for College Students. Chinese Journal of Clinical Psychology (01), 63-66.
Li ML & Weiping, M. (2011). Analysis on Factors Influencing Exercise Dependence of Regular College Student Exercisers. Journal of Tianjin University.
Li, M., Nie, J., & Ren, Y. (2015). Effects of exercise dependence on psychological health of Chinese college students. Psychiatria Danubina, 27(4), 413–419.
Liang Deqing. (1994). Stress level of college students and its relationship with physical exercise. Chinese Journal of Mental Health (01).
Lu, F. J., Hsu, E. Y., Wang, J. M., Huang, M. Y., Chang, J. N., & Wang, C. H. (2012). Exercisers' identities and exercise dependence: the mediating effect of exercise commitment. Perceptual and motor skills, 115(2), 618–631.
Mack, M. G. (2003). Does exercise status influence the impressions formed by college students? College Student Journal, 37(4), 483–490.
McKenzie, S. (2013). Getting physical: The rise of fitness culture in America. Lawrence：University Press of Kansas.
Murray, A. L., McKenzie, K., Newman, E. & Brown, E. (2013). Exercise identity as a risk factor for exercise dependence. British Journal of Health Psychology, 18(2), 369–382.
Murray, A. L., McKenzie, K., Newman, E., & Brown, E. (2013). Exercise identity as a risk factor for exercise dependence. British journal of health psychology, 18(2), 369–382.
Ntoumanis, N., Guerrero, M. D., Gadeke, C., & Thøgersen-Ntoumani, C. (2018). Do exerciser weight status and perceived motivation predict instructors’ motivation and beliefs about the exerciser? A test of motivation contagion effects. Body Image, 26, 10–18.
Palermo, M., Staples, C., & Rancourt, D. (2021). Examining the impact of weight bias on the association between exercise identity and maladaptive exercise behaviors. Eating behaviors, 41, 101503.
Phelan, S., Bond, D. S., Lang, W., Jordan, D., & Wing, R. R. (2011). "Exercise dependence"--a problem or natural result of high activity. American journal of health behavior, 35(6), 726–733.
Qu Hui, Yao Jiaxin & Shi Jianguo. (2017). Research on the relationship between physical exercise persistence, exercise addiction and trait fluency. Journal of Shenyang Sport University (04), 77-83.
Quhui, Zhou Qianyun, Cao Zhenxing & Liu Liwei. (2015). Unconditional Logistic regression analysis of the influencing factors of sports addiction. Journal of Tianjin Institute of Physical Education (05), 425-431.
Robertson, N., & Vohora, R. (2008). Fitness vs. fatness: Implicit bias towards obesity among fitness professionals and regular exercisers. Psychology of Sport and Exercise, 9 (4), 547–557.
Strachan, S. M., Brawley, L. R., Spink, K. S., & Jung, M. E. (2009). Strength of exercise identity and identity-exercise consistency: Affective and social cognitive relationships. Journal of Health Psychology, 14(8), 1196–1206.
Taghian, F., Zolfaghari, M., & Hedayati, M. (2014). Effects of Aerobic Exercise on Serum Retinol Binding Protein4, Insulin Resistance and Blood Lipids in Obese Women. Iranian Journal of Public Health, 43, 658-665.
Teachman, B. A., Gapinski, K. D., Brownell, K. D., Rawlins, M., & Jeyaram, S. (2003). Demonstrations of implicit anti-fat bias: the impact of providing causal information and evoking empathy. Health psychology: official journal of the Division of Health Psychology, American Psychological Association, 22(1), 68–78.
Tucker, J. M., Welk, G. J., & Beyler, N. K. (2011). Physical activity in US adults: Compliance with the physical activity guidelines for Americans. American Journal of Preventive Medicine, 40(4), 454–461.
Veale DMW. (1987). Exercise dependence. Addiction.;82(7):735–40.
Verkooijen, K. T., & de Bruijn, G.-J. (2013). Exercise self-identity: Interactions with social comparison and exercise behavior. Psychology, Health & Medicine, 18(4), 490–499.
Wally, C. M., & Cameron, L. D. (2017). A randomized-controlled trial of social norm interventions to increase physical activity. Annals of Behavioral Medicine, 51(5), 642–651.
Yang, P., Yu, Q., Montag, C., Becker, B., & Zou, L. (2021). Validation of chinese version of exercise dependence scale-revised (eds-r). International Journal of Mental Health and Addiction, 1-20.
Zaroubi, L., Samaan, T., & Alberga, A. S. (2021). Predictors of Weight Bias in Exercise Science Students and Fitness Professionals: A Scoping Review. Journal of obesity, 2021, 5597452.
Zhang Yongfeng. (2012). Research on college students ' sports addiction (master 's degree thesis, Hunan Normal University).

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

The relationship between exercise identity and exercise behavior among Chinese college students: the moderating role of weight bias.

Status:

Version 1

Abstract

Figures

1. Introduction

2. Object and method

2.1 Data Collection and Participants

2.2 Measures

2.2.1 Demographic information

2.2.2 Physical Activity Rating Scale (PARS-3)

2.2.3 Exercise Dependence Scale-Revised Chinese version (EDS-C)

2.2.4 Exercise identity

2.2.5 Weight bias questionnaire

2.3 Data analysis

2.4 Data availability

3. Result

3.1 Common method bias test

3.2 Descriptive statistics

3.3 Correlation analysis

3.4 Test the moderating effect

4. Discussion

4.1 Difference in demographic variables

4.2 Research on the Correlation Between Various Variables

4.3 Test the moderating effect

5. Conclusion

References

Additional Declarations

Supplementary Files

Status:

Version 1