Physical Activity, Social Media, & Self-esteem

Abstract

With the rise of obesity and the push for healthier lifestyles along with the increased usage of social media, the relationship between self-esteem, physical activity, social media use, and the interaction effect of physical activity and social media use comes into question. For this study, 65 participants self-reported data following the online anonymous survey, which asked for age, gender identity, rating of self-esteem based on the Rosenberg Self-Esteem scale, social media usage, and physical activity frequency based on the scale developed for this study (Appendix). The survey results showed that though there was a significant relationship between social media use and self-esteem, the results were not significant for the relationship between physical activity and self-esteem or the interaction effect. The study’s results were inconclusive due to the limited sample size and potential biases that resulted during data collection. Nonetheless, this study followed the same trends as seen in previous studies with the relationship between physical activity and self-esteem, physical activity and self-esteem, and their interaction effect.

Introduction

Standards of beauty have changed and fluctuated throughout history, and with them, our perception of worth has often been contested. What is considered most desirable is widely known to impact the lives of many, especially in terms of how one views themselves. Self-esteem is crucial to a person’s identity and confidence as it can influence a myriad of factors, from success to sociability to happiness. Given the recent push for healthier lifestyles along with the formation of various diet fads, many of which are advocated by celebrities who symbolize ideal images (Rousseau, 2015), the connection between one’s physical health and self-esteem is relevant.

Additionally, obesity is on the rise, possibly due to increased food intake, much of which is highly processed and maximized in oil, sugar, and fat to increase taste, along with decreased exercise (Blair & Nichaman, 2002). Increasing rates of obesity have sparked debate of where the line should be drawn between embracing all body types and encouraging healthier lifestyles (thereby encouraging a standard of health/body type). Past research has found that there is relationship between self-esteem and physical activity as well as physical activity and body image. In a study conducted in Iran, researchers concluded that those with low self-esteem should be encouraged to exercise (Zamani, et. al, 2016). Though establishing the link between physical activity and self-esteem, past research neglects to include other external factors.

With more and more people engaging with social media platforms and circulating ideal images, the variable of social media comes into the picture. Social media has been studied in the past with regard to self-esteem by examining the relationship between social media addiction and self-esteem (Hawi & Samaha, 2017), which indicates that social media addiction leads to lower self-esteem. Also, images of idealized celebrity body types have been found to impact body satisfaction and dieting practices, especially in young adolescent girls (Mooney, Farley, & Strugnell, 2009). With results from past research, the factor of social media may be relevant in the analysis of self-esteem, especially when coupled with the variable of physical activity since exercise has a direct impact on external characteristics.

This study assessed the relationship between self-esteem and physical activity, with the additional independent variable of social media use. Much of previous research considers other internal factors, such as body image and satisfaction with life, to better understand the link between self-esteem and physical activity. Differing from past research, the relationship between self-esteem and physical activity for those living in the Western world was assessed and the use of popular social media platforms known for circulating “perfect” or desirable images, was an independent variable included to see if there was an impact of social media usage instead of social media addiction on self-esteem.

For this study, based on previous research and expected trends, the following hypotheses were developed.

Hypothesis 1:

H_o: Physical activity will have no effect on self-esteem.

H_a: Greater physical activity will cause self-esteem to increase.

Hypothesis 2:

H_o: The relationship between self-esteem and physical activity will not be impacted by use of social media.

H_a: The relationship between self-esteem and physical activity will be impacted by use of social media.

Hypothesis 3:

H_o: The degree to which physical activity affects self-esteem will not depend on social media use.

H_a: The degree to which physical activity affects self-esteem will depend on social media use.

Hypothesis 1 predicts the main effect of whether physical activity is related to self-esteem. Hypothesis 2 controls for the possible confounding variable of social media use. Hypothesis 3 predicts for a possible interaction effect between physical activity and social media usage for the dependent variable, self-esteem.

Methods

There were 65 participants in his study, ranging from 18 to 74 years old and averaging 35.22 years old. 60% of participants (n = 39) identified as female, 40% of participants (n = 26) identified as male, and 0% identified as other. To recruit participants, I posted the survey on the Berkeley-affiliated Facebook groups Free & For Sale and UC Berkeley Class of 2020, as well as on my personal account. After receiving few responses, I messaged friends directly through Facebook and texted them with the survey link asking them to take the survey. Additionally, I asked my parents to circulate the survey (and take it themselves) to family and family friends, which also allowed me to expand the age range of participants. Since my study was not based on just UC Berkeley students, I encouraged everyone who took the survey to circulate it to their friends/roommates/family.

The survey was created through Qualtrics, a subscription software used to collect and analyze data, which I was able to access through my affiliation with UC Berkeley (email account). The survey asked for participants’ age, gender identity, self-esteem rating using the Rosenberg Self-Esteem Scale (Rosenberg, 1965), social media usage, and physical activity frequency using a Likert scale I developed. To complete the study, participants clicked on an anonymous link that took them to the survey and then self-reported the data.

The dependent variable, self-esteem, was measured using the Rosenberg Self-Esteem Scale (RSES) (Rosenberg, 1965). The scale consisted of 10 items and participants chose their level of agreement with each item on a 5-point scale instead of the original 4-point scale because the option of “Neutral” was given in case participants felt they could not identify or did not want to choose a more extreme option. Additionally, data analysis was made easier by having both continuous variables utilize the same 5-point scale. Participants chose from the 5 options of Strongly Disagree (1), Disagree (2), Neutral (3), Agree (4), and Strongly Agree (5). Examples of statements presented to participants include “I feel that I’m a person of worth, at least on an equal plane with others.” and “All in all, I am inclined to feel that I am a failure.” The primary independent variable, physical activity, was measured through a Likert scale I developed. The scale consisted of 8 items and participants reported their frequency of performing each item on a 5-point scale. Participants chose from the 5 options of Never (1), Rarely (2), Occasionally (3), Frequently (4), and Always (5). Examples of the statements presented to included “I am more active than I am sedentary in any given day.” and “I am not involved in vigorous physical activities that make me sweat or puff and pant (i.e. aerobics, heavy lifting, jogging, etc.).” The second independent variable, usage of social media, was measure as a categorical variable. Participants were asked if they used any of the following popular social media platforms: Facebook, Instagram, Twitter, Snapchat, Google+, or YouTube, and participants answered “Yes” or “No”.

Data Cleaning

Participants’ survey data was exported from Qualtrics into a CSV file with each of the Likert scales being converted to the aforementioned corresponding numerical values. 3 participants who did not answer any questions were removed in Qualtrics itself, giving the final number of participants as 65. Unnecessary words and columns of data (i.e. IP address) were removed from the data in Excel and the questions were relabeled for easier analysis in R. Then, the data was imported into R. The two continuous variables, self-esteem and physical activity, were combined and reverse-scored as necessary. Higher values of self-esteem and physical activity corresponded to higher self-esteem and more frequent physical activity. The categorical variable, social media usage, was converted from a continuous variable to a categorical one. No outliers were found and thus no further data was removed or cleaned.

Results

The variables measured in this study were gender (39 female, 26 male, 0 other), age (mean = 35.22, SD = 17.39, range = [18, 74]), self-esteem (mean = 3.60, SD = 0.88, range = [1.1, 5], alpha = 0.92), usage of social media (92.3% Yes, 7.7% No), and frequency of physical activity (mean = 3.15, SD = 0.84, range = [1, 4.625], alpha = 0.85). The spread of data for each variable is shown in the graphs below:

Graphs of Key Variables

Linear regressions were used to predict self-esteem from physical activity, use of social media, and their interaction effect. Although self-esteem and frequency of physical activity were on the same scale, I z-scored each variable in the models to ensure I could compare all the variables (for model 1, I got the same results before and after standardizing). Beta coefficients are reported for estimates as they indicate that slopes are in units of standard deviations, and the statistics are summarized in Table 1.

I ran a bivariate linear regression between self-esteem and physical activity to test Hypothesis 1. Looking at the first model of Table 1, the bivariate model predicting self-esteem from physical activity suggests that self-esteem is not related to frequency of physical activity (β = 0.36, 95% CI = [-0.04, 0.45], t(63) = 1.68, p = 0.098). Although there was a positive effect of greater frequency of physical activity on self-esteem, the effect was insignificant (R² = 0.043, F(1, 63) = 2.81, p > 0.05). Thus, the hypothesis was not supported and the results caused the failure to reject the null hypothesis. Power calculated, which indicates the probability that any test correctly rejects the null hypothesis, was low (P = 37.47%).

Additionally, I ran a bivariate linear regression between self-esteem and social media use. As seen in the second model of Table 1, the bivariate model predicting self-esteem from social media use suggests that self-esteem is related to social media use (β = 0.80, 95% CI = [0.003, 0.91], t(63) = 2.013, p = 0.048). There was a negative effect of social media use on self-esteem (use of social media causes lower self-esteem) and the effect was significant (R² = 0.06, F(1, 63) = 4.05, p < 0.05). However, though the effect was significant, it must be noted that only 5 people reported not using social media and since the sample size of those not using social media was very low, the results may not be correct.

To test Hypothesis 2, I created a model to predict self-esteem from both physical activity and social media use (Model 3). The addition of the variable of social media use cause the the relationship between physical activity and self-esteem to be halved, which indicates a partial mediation effect (β = 0.18, 95% CI = [-0.06, 0.45], t(62) = 1.52, p = 0.13). Unlike Model 2, the relationship between social media use and self-esteem is insignificant (p = 0.065), as remained the relationship between physical activity and self-esteem.

For Hypothesis 3, the model added an interaction effect between physical activity and social media usage (Model 4). Though there was an interaction effect (β = 0.19, 95% CI = [-1.44, 1.53], t(61) = 0.066, p > 0.05), it was insignificant. However, it is interesting to note that those who did not use social media generally had higher self-esteem regardless of physical activity than those that did use social media (Figure 3).

Table 1. Predicting self-esteem from physical activity frequency, social media use, and their interaction effect.

Note: * p < .05, ** p < .01. 95% Confidence Intervals reported in brackets, degrees of freedom for the F-test reported in parentheses (model, residual).

Figure 1. Relationship between standardized physical activity frequency and standardized self-esteem.

Figure 2. Relationship between social media use and standardized self-esteem.

Figure 3. Relationship between physical activity and self-esteem with social media use (+1 SD in black) and no social media use (-1 SD in red).

Discussion

The results of this study found that self-esteem and physical activity did not have a significant relationship. Social media use was a partial mediator in this relationship, and the main effect relationship between social media use and self-esteem was found to be significant. However, the number of participants who did not utilize popular social media platforms was very small (n = 5), so it is difficult to conclude that social media use significantly impacts self-esteem. When a possible interaction effect was tested between physical activity and social media usage, no significant effects were found. Even though these results were not significant, a positive trend was present in the relationship between physical activity and self-esteem, which is consistent with past research. This study does not adequately demonstrate a strong positive connection between physical activity and self-esteem, and further research would be needed to confirm or deny a positive relationship between the two variables, particularly in the Western world.

The most problematic limitation of this study was the small sample size (n = 65). With more time and resources, I would attempt to survey a random, larger sample that would have more equal amounts of male and female participants (50/50 instead of 40/60). I would also ask participants for their race/ethnicity because I suspect, due to the way I recruited participants and my social circle composition, that most of the participants in my study were concentrated in a single racial identity (Asian), which may have biased the results. However, I cannot be certain of this since I did not have the foresight to ask participants to identify their racial/ethnic background. Also, I struggled to get responses from non-college age students as there are few within my network, which may have biased the results as well. Finally, most of my participants were concentrated in California and Arizona so I would need to expand the survey area to better understand the general population and prevent bias due to location. Because of the small sample size and the possible biases of the data in age, race/ethnicity, and location, I am unable to generalize my results to the overall population.

The data could have been impacted negatively or skewed because there were many potential variables that I could not account for. For example, stress could have affected self-esteem at the time participants were taking the survey due to the various commitments they might have had, which may have minimized the relationship between physical activity and self-esteem. Another variable that could have potentially explained the trends in the data is gender, which I did not account for during my analysis though I included it in my survey. It would have been interesting to see the impact of gender on the relationship between physical activity and self-esteem, especially since the split between male and female was much more equal than that between those who use social media and those who do not.

If I were to conduct this study again, I would attempt to gather a larger amount of data that has participants across all ages, races/ethnicities, and geographical location to better represent the population. In my survey, I would include questions about race/ethnicity as well as city that the participant is located in. Additionally, I would have to disseminate the survey through more reliable and random means, such as a paid service like Amazon’s Mechanical Turk. Overall, though my study did not produce statistically significant results, I believe that there is a connection between the three variables of physical activity, social media use, and self-esteem that should be explored further.

References

• Blair, S. N., & Nichaman, M. Z. (2002). The public health problem of increasing prevalence rates of obesity and what should be done about it. Mayo Clinic Proceedings, 77(2), 109-113.
• Hawi, N. S., & Samaha, M. (2017). The relations among social media addiction, self-esteem, and life satisfaction in university students. Social Science Computer Review, 35(5), 576-586.
• Mooney, E., Farley, H., & Strugnell, C. (2009). A qualitative investigation into the opinions of adolescent females regarding their body image concerns and dieting practices in the Republic of Ireland (ROI). Appetite, 52(2), 485–491. https://doi.org/10.1016/j.appet.2008.12.012.
• Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.
• Rousseau, S. (2015). The Celebrity Quick-Fix. Food, Culture & Society, 18(2), 265-287.
• Zamani, S. H., Fathirezaie, Z., Brand, S., Pühse, U., Holsboer-Trachsler, E., Gerber, M., & Talepasand, S. (2016). Physical activity and self-esteem: testing direct and indirect relationships associated with psychological and physical mechanisms. Neuropsychiatric Disease and Treatment, 12, 2617-2625. doi:10.2147/NDT.S116811

Appendix

Full Scales

Rosenberg Self-Esteem Scale (RSES):

• I feel that I’m a person of worth, at least on an equal plane with others.
• I feel that I have a number of good qualities.
• All in all, I am inclined to feel that I am a failure.
• I am able to do things as well as most other people.
• I feel I do not have much to be proud of.
• I take a positive attitude toward myself.
• On the whole, I am satisfied with myself.
• I wish I could have more respect for myself.
• I certainly feel useless at times.
• At time I think I am no good at all.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.

Physical Activity Scale:

• I exercise almost every day for 30 minutes or longer
• In a typical week, I am involved in moderate-intensity activity (i.e. walking, bicycling, gardening, etc.) that causes small increases in breathing or heart rate for at least 30 minutes continuously almost every day.
• I am more active than I am sedentary in any given day.
• I participate in vigorous physical activities (i.e. running, sports, lifting weights, etc.) that cause large increases in breathing or heart rate for at least 10 minutes continuously almost every day
• My lifestyle is mostly sedentary.
• I do not exercise.
• I do not engage in activities that are moderate-intensity and make my heart rate increase or make me breathe harder than normal (i.e. carrying light loads, bicycling, walking, etc.)
• I am not involved in vigorous physical activities that make me sweat or puff and pant (i.e. aerobics, heavy lifting, jogging, etc.)