The relationship between media use and sports participation behavior: A meta-analysis

Eligibility criteria

Because of the significant variation in research across the literature, the criteria required for inclusion in the meta-analysis were: (a) the language was Chinese and English; (b) the study specified information about the study population, sample size, and measurement-related procedures; (c) the data results demonstrated the relationship between media use and sports participation behavior and were not purely theoretical or review articles; (d) the data analysis presented Pearson effect sizes; (e) the medium used was specifically mentioned; (f) at least one sport participation behavior was included; and (g) if the same study was published multiple times, only one instance was included in the analysis. A total of 12 literature that met the meta-analysis requirements were finally obtained. The specific inclusion processes are shown in Figure 1.

OPEN IN VIEWER

Figure 1.

Flow diagram for inclusion in meta-analysis.

Information sources and search strategy

Literature related to this topic was collected through a literature search. The studies were mainly collected from EBSCO, JSTOR, PubMed, SPECIASCI, ELSEVIER, Google Scholar, Embase, CINAHL, Sports Discus, and CNKI. Relevant manual searches were conducted on the references of published review articles to ensure that the data in the meta-analysis were as representative and complete as possible. The selected keywords related to media use were: “media,” “media use,” “mass media,” “Internet,” “computer,” “TV,” and “smartphone,” while those related to sports participation were “sports,” “sports participation,” “physical activity,” “exercise”, “sports consumers,” “online,” and “offline.” The keywords were searched one by one and in combination.

Selection process and data collection process

For the extraction of literature data, in order to ensure the quality of the literature, the selection process was completed by three researchers. The first researcher extracted articles according to the structured table, then the second researcher checked their accuracy. Their work was not carried out simultaneously but rather was sequential. When there was a disagreement, firstly, the two researchers would negotiate to come to a resolution. In situations where an agreement could not be made, the corresponding author made the final decision.

Data items

The 12 literature included in the meta-analysis were coded as follows: literature information, sample size, gender, study population, sports participation behavior, media, media instrument, media measurement method, culture, and effect size. According to the previous section, sports participation behaviors were divided into online physical behaviors, offline physical behaviors, online consumption behaviors, and offline consumption behaviors. Media were uniformly divided into two categories: new and traditional media. Media tools concretized new media and traditional media, including the Internet, television, and newspapers. According to the classification of related studies, the measurement of media use were divided into two categories: type (type of media) and time (frequency of media use). Culture was mainly divided into three types based on region: East Asia, Western Europe/North America, and others. ³⁷ East Asia mainly includes Asian countries such as China (including Hong Kong, Macau, and Taiwan), Japan, and South Korea. Europe/North America includes countries such as the United Kingdom, Switzerland, Germany, the United States, and Canada, while others refer to other countries not included in the above classification, such as Thailand, the Philippines, Turkey, and Iran. The study population was divided into primary and secondary school students, college students, and others (others mainly included adult groups and mixed groups of adults and students). Effect sizes for each independent sample were coded independently according to the sample itself or separately if more than one independent sample existed in a study. The effect sizes for this study were all direct Pearson effect sizes, as explicitly presented in the original literature, and were not converted otherwise. This study ultimately obtained 17 independent studies from 12 literature, as shown in Supplemental Table 1.

Research quality assessment

Literature quality was assessed using the Medical Education Research Study Quality Instrument (MERSQI). ³⁸ Although MERSQI was developed for medical education, it can also be applied in non-medical teaching research because of its relative neutrality and objectivity. ³⁹ The ten items of the MERSQI cover six domains: (a) study design, (b) sampling, (c) data types, (d) validity, (e) data analysis, and (f) outcomes. The maximum score for each domain is three, with a total score of 18. A mean of 14.26 was obtained with a standard deviation of 0.35 and a median of 14, which indicates the overall sound quality of the literature included in the study. The specific MERSQI values for each piece of the literature are listed in Supplemental Table 1.

Synthesis methods

After comparing different data processing software in this study, a more comprehensive CMA 3.0 (Comprehensive Meta-Analysis 3.0) was selected for analysis. The sample size (k), weighted effects (r), and 95% confidence intervals were calculated to test the study hypotheses. Studies were tested for significant differences in mean effect sizes by Cochrane's Q and I2 (heterogeneity test), and effect sizes were determined to show significant heterogeneity if the I2 exceeded 75. ⁴⁰ In addition, this study used a random-effects model, which is more suitable for meta-analysis than a fixed-effects model (in a random-effects model, the selected studies are considered as a random sample from a larger population as a way to generalize the findings). ⁴¹ In the literature included in this study, the subjects were of different ages. They used different measurement instruments and methods, making the use of a fixed-effects model inappropriate, and the need for the study to discuss the role of moderating variables in this, making a random-effects model more appropriate. To test for publication bias, we used funnel plots, Rosenthal’s Fail-Safe N test, and Egger’s test after synthesizing related studies

Heterogeneity test

After filtering the literature, correlation operations were performed on 12 literature with 17 effects. The forest plot in Figure 2 shows a more pronounced heterogeneity. Figure 3 shows the Q-value was 916.42 with a degree of freedom of 16, p < 0.001, indicating that the effect sizes in the meta-analysis were heterogeneous. The I-squared value was 98.254, indicating that 98.254% of the observed variation was highly heterogeneous due to actual differences in the effect values. A T-squared value of 0.094 indicated that 9.4% of the study space variables were available for the weight calculation. The heterogeneity test results indicated that the random-effects model was accurately chosen for the study.

OPEN IN VIEWER

Figure 2.

Study forest map.

OPEN IN VIEWER

Figure 3.

Heterogeneity test results.

Publication bias test

Figure 4 shows publication bias tests were performed on 17 effect sizes, and according to the funnel plot, the studies were relatively symmetrical overall. However, they were only partially distributed on the left side. An Egger's test (Figure 5) was conducted to obtain more accurate and relevant data, and it showed that there was no publication bias in the studies, p  =  0.47586 > 0.05. In addition, Rosenthal’s Fail-Safe N results (Figure 6) showed that an additional 1704 references involving media use and sports participation behaviors would need to be included for the total effect to be insignificant; therefore, there was no severe publication bias.

OPEN IN VIEWER

Figure 4.

Research funnel diagram.

OPEN IN VIEWER

Figure 5.

Egger's test results.

OPEN IN VIEWER

Figure 6.

Rosenthal's fail-safe N test results.

Main effect analysis

According to the results of the random-effects model, Figure 7 shows the main effect between media use and sports participation behavior was r  =  0.192, with a 95% confidence interval of (0.047,0.329), p  =  0.01, thus indicating a significant positive relationship between media use and sports participation behavior.

OPEN IN VIEWER

Figure 7.

Main effect results.

Test for moderating effects

Form of media (Internet, TV, Radio, Newspaper, Magazine), media measurement methods (type, time), study subjects (college students, primary and secondary school students, others), and culture (East Asia, Europe/North America, others) were tested for moderating effects, and the results are shown in Supplemental Table 2.

Correlation coefficients for the form of media indicated that media-specific form differences produced a moderating effect between media use and sports participation behavior; that is, they influenced the relationship between the two and showed a significant positive relationship (Q  =  19.927, r  =  0.063, p  =  0.015 < 0.05). The moderating effect was most pronounced, and the correlation was most significant for newspapers (r  =  0.365, p  =  0.000), television (r  =  0.330, p  =  0.015), and the Internet (r  =  0.242, p  =  0.009).

The media measurement method correlation coefficients indicate that differences in the way media performance is measured can only moderate the effect between media use and sports participation behavior (Q  =  1.556, r  =  0.198, p  =  0.006 < 0.01). However, it was found through the data that when measured by the type of media used, a positive relationship was presented, that is, the more types of media used, the more positive the sport participation behavior (r  =  0.232, p  =  0.003 < 0.01). However, when measured by the duration of media use, it was found that the longer the time spent using media, the more it inhibited sport participation (r  =  −0.036, p  =  0.861 > 0.05). However, there were no significant correlations between the two variables.

A significant positive effect for the study subjects’ indicated that differences in study subjects created a moderating effect between media use and sports participation behavior (Q  =  12.285, r  =  0124, p  =  0.009 < 0.01). However, primary and secondary school students showed the opposite (r  =  0.146, p  =  0.279 > 0.05), indicating that elementary and middle school students did not have a moderating effect on the relationship between media use and sports participation behavior.

Culture correlation coefficients indicated that cultural differences have a significant positive moderating effect on the relationship between media use and sports participation behavior (Q  =  12.470, r  =  0.171, p  =  0.001). This was most significant for East Asian countries (r  =  0.129) than European/North American countries (r  =  0.042), indicating that in terms of culture, Eastern countries have a stronger correlation between media use and sports participation than Western countries.

The relationship between media use and sports participation behavior

This study combined previous research on the relationship between media use and sports participation behavior and found a significant positive relationship, consistent with previous research.

Media use influences physical participation in sports. This means that watching TV or playing games may replace high-intensity physical training among young people. In theory, because a person's time is limited, apart from work, study, or rest, it is difficult to do one thing if one chooses to do another. Once an individual invests in media, it becomes difficult to participate in sports activities. ⁴² Related studies in Germany have shown that intensive use of digital media is incompatible with sports participation. Moreover, the more time spent on the media, the less time spent exercising. ⁴³ However, in fact, when social media is used in sports, their time increases. ⁴⁴ Young people can reach 78.6% participation in sports and present good attitudes toward sports after watching online videos or articles. Sports celebrities on the Internet also positively influence youth participation in sports. ⁴⁵ Overall, there is an ambiguous relationship between media and physical activity in sports, where the influence of participation or independence counteracts alternative utility in quantitative studies. However, the use of social media can enhance sports participation. ⁴⁶ In addition, informational and trend-based sports are more beneficial in social networks than organized sports and are integrated into “medicine and physical activity.” ⁴⁷

In terms of sports consumer engagement, marketers use social media to engage youth and actively introduce them to the benefits of sports, creating a desire to buy sports products. Sporting goods companies advertise their products on forums, and sports training companies advertise their programs and promotions through pictures and videos. ⁴⁸ For example, sports drink merchants use media to deepen the psychological perceptions of adolescents, which in turn leads to consumer behavior, with some studies showing a positive correlation between the two. ⁴⁹ This is true not only for physical sports consumption but also for some virtual sports consumption. Viewing behavior is a type of sports consumption participation, as social media determines, to some extent, the attitudes of sports fans towards sports programs and events. Furthermore, sports fans obtain information about games and players through social media and use social media to discuss them extensively, ⁵⁰ thus forming a so-called “circle culture,” which leads to deeper sports participation. Currently, because sports fans’ media use is not limited to one medium, organizations should utilize more than one social media platform, which will deepen the perceived power of the brand ⁵¹ and thus, sports consumption behavior. However, this is not absolute; one study that analyzed the relationship between attitudes toward social media and sports consumption among a group of adolescent sports fans born after 2000 found that attitudes toward social media do not influence their desire or willingness to consume sports. ⁵² Instead other external environmental seem to influence their desires, suggesting that, to some extent, the medium also has limitations.

Meanwhile, the value of r  =  0.19 also shows that, although the relationship between media use and sports participation behavior was significant, it was relatively weak. This means that although media use improves sports participation behavior, its influence is not strong. The main reason for this is that, whether sports activities or sports consumption, it is mainly purposeful behavior, which is often generated independently by humans for reasons of health or weight loss or to satisfy inner desires. For example, a study conducted a cross-sectional survey of 2527 Norwegian teenagers found that changing their bodies was an important reason for their participation in sports activities. ⁵³ That is, a considerable number of people, even without the role of the media, still engage in sports participation behavior, and the impact of the media on their sports participation behavior is very weak. What truly affects the use of media is that those who originally did not have the intention to participate in sports begin to participate in sports activities or to engage in sports consumption through the media. However, this is problematic. In addition to the media, there are many other factors that may lead to sports participation, such as friends, leaders, and promotions. Research has shown that the participation of 15-year-old children in sports activities is largely influenced by family culture, ⁵⁴ so the media is only one of the inducing factors and its influence is limited. These two reasons result in a relatively weak correlation between media use and sports participation, although the correlation is in fact significant.

The role of moderating variables in media use and sports participation behavior

In addition, as suggested in a previous study, some moderating variables also played an essential role in the final results. First, in terms of the form of media, both new media and traditional media show a positive relationship, but the specific magnitude varied. Although the results of this study show that traditional forms of media have more influence than new media, combined with previous research, it indicated that new forms of media (in many forms, including the Internet) are more significant than traditional forms of media, such as TV. ³⁶ This is primarily because the Internet, as the most widely used form of media, has broken the barriers of time and space and significantly increased the rate and frequency of information dissemination, thus deepening people's perceptions and implementing these perceptions into behavior. Second, the measurement of media presents conflicting results, with some studies finding that the more time spent on media per day, the less time spent doing exercise per week, indicating that excessive media use may lead to low activity levels and decreased motor skills.^33,34 Such results only apply to physical participation in sports and are not evident in sports consumption participation. There is also variability in the results for primary and secondary school students, which can be related to the time mentioned above, where primary and secondary school students have limited time due to their studies and cannot choose between media and sports functions, thus creating a relationship between the inverse effects of both. Finally, looking at the cultural domain, the characteristics of different cultures create differences in preferences for physical activity. In general, Western countries are more prominent in the field of sports, especially physical participation in sports, compared to Eastern countries. The r-values in various studies showed that Europeans and North American are more susceptible to media images influencing their sports participation than East Asians (0.043 > 0.129). However, there are also studies showing that Eastern countries are mostly collectivist cultures where adolescents may have a more solid social group than Western individualistic adolescents whose decisions are more independent. Eastern collectivism is more susceptible to media influence than Western individualism. ⁵⁵ Thus, the role of moderating variables is not absolute and can produce numerous variations.

In summary, based on the research results, compared with previous studies, we obtained the following new findings: First, although there is a significant relationship between media use and sports participation behavior, unlike previous studies, this study found the relationship to be weak. The main reason for this is that some people are subjectively willing to participate in sports rather than being influenced by external factors such as the media, while others may be influenced by external factors but not necessarily by the media; thus, the role of the media is not significant. Second, the research suggests that some factors may play a moderating role in media use and sports participation behavior, which has not been previously proposed, which constitutes a new finding. In terms of the form of media, although traditional media still has a greater influence than new media, with the passage of time, the new media will become more important than the traditional media. In terms of the measurement of media use, the longer the time spent on media, the less time spent participating in sports. However, this applies only to sports activities and not to sports consumption, and does not apply to primary or secondary school students. Due to school arrangements, their sports activities are obligatory and fixed and will not be affected by the time spent using the media. Third, in terms of culture, Europeans and North Americans are more susceptible to media influence than East Asians are, which has not been reported in previous studies.