Teaching Evolution: The Use of Social Networking Sites

Abstract

Online schooling has been adopted worldwide due to the COVID-19 pandemic. During quarantine, people go online for all kind of purposes, especially for amusement such as via social networking sites (SNSs). This study examined university physical education (PE) students’ SNSs usage intention using the Unified Theory of Acceptance and Use of Technology model II (UTAUT2) in Taiwan. Research respondents were selected from PE departments of 19 universities through purposive sampling method. A total of 707 questionnaires were collected, with a returning rate of 93%. Using Warp PLS 7.0 as the main instrument for data analysis, this research finds that performance expectancy, facilitating conditions, hedonic motivation, price value, and habit within the UTAUT2 model have significant positive effects on students’ intention to use social networking sites, and the model explains 63.4% of the variance in their intention to use SNSs. Among those variables, hedonic motivation had the highest impact (β = 0.24). Moreover, intention, facilitating conditions and habit have significant positive effects on students’ use of social networking sites, and the model explains 13.4% of the variance in their use of social networking sites. The moderating effects of gender, age and experience are found in some path analyses. These findings provide future university instructors a with better understanding of students using SNSs. We thus recommend for university PE instructors to create interesting and pleasant classroom learning experiences to attract students’ attention, and recommend that they may even manage a SNS as an aid for teaching to enhance students’ interests in learning.

1. Introduction

With the prevalence of smartphones and 5G, social networking is part of many people’s everyday life. However, what exactly is social networking? According to Psychology Today [1], the term “social network” refers both to a person’s connections to other people in the real world and to a platform that supports online communication, such as Instagram, Facebook, or Twitter. The term is now used more often in the second sense, and the Internet provides an opportunity for anyone to create an online identity, connect with friends, family, and strangers alike, acquire knowledge, and share ideas and information without having to be physically present. Instead, one’s presence is represented on social media by shared comments, photos, videos, and other images.

Behind social networking, social networking sites provide platforms for people to create personal profiles and communicates with each other. According to Donath and Boyd [2], a social networking site (SNS) is an online space where people publicly share the profile they set up, and which offers an opportunity for social interaction. Rheingold [3] stated that a social networking site is also a social aggregate of virtual communities where people establish and connect relationships. Simply put, a social networking site creates a great opportunity for people with similar interests to interact and share information [4].

According to Statistica [5], Facebook remains as the most popular social networking site worldwide as of October 2021. In Taiwan, according to Digital report [6], the top-five most-used social media platforms are YouTube (89.6%), Facebook (89.2%), Line (88%), Instagram (59.5%) and Facebook Messenger (59.3%). Among them, YouTube serves as a platform for people to watch various kinds of videos to enjoy entertainment, learn things, and tune in to real-time information, and users can also upload or even broadcast real-time videos to sharing information with others. As for Facebook or Instagram, they provide a platform for people to post their personal information, photos, films, etc., and share their information with friends and interact with friends as well. As for Line and Facebook Messenger, people can use them to communicate with each other or groups using texts, photos, pictures, icons, and voice messages. Additionally, Line and Facebook Messenger also provide real-time call-out services for talking through voices and images. In one way, these social networking sites provide a great opportunity to relieve stress or kill time, communicate with friends and workmates and share personal information with others. In another way, some social networking sites also provide platforms for people to learn things, acquire information, and make new friends. Social networking sites, regardless of features, therefore, have become a platform for entertainment, social connection, and even just for killing time.

Foreseeing the convenience of communication and knowledge acquisition via social networking sites, some teachers set-up personal profiles on them and encourage students to join designated learning groups. They use these to post course-related materials to allow students to obtain the needed information, to enable groups to communicate and to reply to students’ questions in a timely manner, which has been found to be a very effective teaching aid. The interaction motivates students and provides an opportunity to put their learned knowledge into practice. Studies have also shown that online social media used for collaborative learning had a significant impact on interactivity with peers, teachers and online knowledge-sharing behavior, which had a significant impact on students’ engagement and consequently contributed to students’ academic performance [7,8].

During the COVID-19 pandemic, many courses have turned to online learning. This is especially difficult for some hands-on training courses which especially need students’ active participation. For example, physical education (PE) students who need to follow an instructor’s lead to perform specific actions may find it difficult to follow an instructor’s orders online. If the instructors put their teaching materials and instruction videos on their exclusive SNSs in advance, will it enhance PE students’ motivation to engage in the online courses? According to previous studies [7,8], it certainly will. Therefore, this leaves us with the following question: what factors contribute to PE students’ usage behavior and intention to engage with SNSs? By identifying these factors, PE teachers can design and maintain their SNSs to attract PE students’ participation and make their learning more enjoyable, which may contribute to their academic performance.

Since SNSs incorporate cutting-edge ICT, it is therefore interesting to know how the users get to be familiar with the SNSs and become accustomed to using them. In other words, what factors affect their usage and behavior regarding this technology? Many researchers have studied intention and behaviors when using information technology from psychological and sociological perspectives. A number of factors that influence customers’ acceptance and use of technology have been discussed too. For instance, the technology acceptance model (TAM) is an information systems theory that David [9] introduced based on the theory of reasoned action (TRA). TAM emphasizes how users come to accept and use information technology, and suggests that when users are presented with a new technology, factors such as perceived usefulness and perceived ease-of-use influence their decisions about how and when they will use it.

TAM has been tested, verified, and accepted across disciplines by many researchers [10,11]. However, Venkatesh, Morris, Davis, and Davis [12] claim that it is however insufficient to explain individual use intention and behavior toward information technology from psychological and sociological perspectives alone, thus they proposed the unified theory of acceptance and use of technology (UTAUT) model. This model explores customers’ technology usage intentions by incorporating performance expectancy (PET), effort expectancy (EE), social influence (SI), and facilitating conditions (FC) as predicting factors. In addition, gender, age, experience, and voluntariness of use were considered as moderators among the four constructs. The model has a good explanatory power and predicts 70% of the variance in users’ intention to adopt information technology [13].

Venkatesh et al. later found out that UTAUT has been applied in many research disciplines, and they felt the need to extend the model so it could be applied in a consumer context. Therefore, they modified the UTAUT model and proposed the UTAUT2 model, which incorporated three addition variables, namely, hedonic motivation (HM), price value (PV), and habit (HT), to interpret users’ behavior and intention [14,15]. Compared with UTAUT, they found UTAUT2 provided a more satisfactory explanatory power than UTAUT.

While UTAUT has been used in various research areas, UTAUT2, however, is rarely found in recent studies, and should be further tested before it can be applied in some other areas of research [16,17]. Therefore, this present empirical study employed UTAUT2 to explore the impacting factors of university PE students’ usage intention and behavior toward SNSs. Furthermore, the study also explored how each individual variable (performance expectancy, effect expectancy, social influence, facilitating conditions, hedonic motivation, price value, habit, behavioral intention, and use behavior) correlates with each other, and tried to navigate the moderating effects of gender, age, and experience among those variables in the UTAUT2 model. The results can provide useful information for university PE instructors to coordinate attractive personal profiles on the web to improve teaching efficiency and encourage students’ engagement in classes.

2. Methods

The study applied the UTAUT2 model to explore university PE students’ usage intention and behavior toward SNSs. The study used a structured questionnaire based on UTAUT2 to collect information from university PE students. Later, the collected data were used to test the hypotheses proposed by UTAUT2. The following sections describe how the participants were selected, how the questionnaire was constructed and how the collected data were applied to test hypotheses derived from UTAUT2.

2.1. Participants

In this study, we invited 760 participants who studied in the PE departments of universities in Taiwan. Using a stratified quota sampling method, 10 students (five males and five females) in each grade were selected, with a total of 40 students from each university. We were able to obtain 707 valid sample with a return rated of 93%. Before investigation, participants agreed to sign a consent form. This survey was approved by the National Cheng Kung University Human Research Ethics Committee (REC). The REC number is NCKU-HREC-102-121. The data were also used in a previous study which employed the UTAUT model [18], and the results will be compared with the previous study in the discussion section.

2.2. Measurement

The questionnaire as a research instrument was designed to explore how university PE students use social networking sites. In order to increase the reliability and validity, the instrument was developed based on a review of literature sources. The content of the survey instrument was developed from Venkatesh et al. [16] and modified to fit our research needs. The first section of the instrument included the UTAUT2 model, which comprised performance expectancy (PET), effect expectancy (EE), social influence (SI), facilitating conditions (FC), hedonic motivation (HM), price value (PV), habit (HB), behavioral intention (UI), and use behavior (UB). The instrument comprised 33 items and was measured using 5-point Likert scale to measure respondents’ level of agreement, ranging from “1 = strongly disagree” on one end to “5 = strongly agree” on the other. The demographic information included gender, age and SNSs usage experience.

2.3. Data Analysis

A total of 16 hypotheses were tested using partial least squares (PLS), a regression method for studying complex multivariate relationships among observed and latent variables. Similar to structural equation modeling (SEM), PLS measures the correlation of constructs and analyzes the reliability and validity of each construct [19]. PLS is a popular statistical method in information discipline and is commonly used in UTAUT and UTAUT2 studies [12,18,20]. There are many statistical software packages that enable users to perform PLS. Warp PLS 7.0 developed by Kock [21] was used for the present study.

3. Results

3.1. Demographic Characteristics of the Respondents

Table 1. Demographic characteristics of the respondents (N = 707).

	Frequency	%		Frequency	%
Gender male	375	53	SNS usage experience less than 1 year	26	6.6
female	332	47	1–2 years	63	8.9
Age 18	96	13.6	2–3 years	137	19.4
19–21	504	71.4	Over 4 years	314	44.4
above 21	107	15

presents the descriptive statistics of the participants. The participants comprised 375 (53%) male students and 332 (47%) female students. As for age, most of them were aged 19~21 years old (504/71.4%), 96 (13.6%) were 18 years old and 107 (15.0%) were 22 years old or above. As for SNS usage experience, the majority had over 4 years’ experience (314, 44.4%), and the next was 2–3 years’ experience (137, 19.4%), followed by 1–2 years’ experience (63, 8.9%), and the rest had less than one year’s use experience (26, 6.6%).

3.2. Reliability and Validity Verification of the Measures

3.2.1. Common Method Variance and Multicollinearity Diagnostics

According to Avolio, Yammarino, and Bass [22], if data are collected from one single course, then single-source bias, which is a form of common method variance (CMV), arises due to overlapping variability. To address the issue of CMV, Podsakoff, MacKenzie, Lee, and Podsakoff [23] applied Harman’s single-factor test, one of the most widely used techniques that has been used by researchers. They load all of the variables in their study into an exploratory factor analysis and examine the unrotated factor solution to determine the number of factors that are necessary to account for the variance in the variables. As the first factor is extracted and its explanatory power is less than 50%, the result is not a significant common method bias [22]. In this study, we performed exploratory factor analysis and had eight concluded factors. The first factor’s explanatory power was 36.10%, which is less than 50%, hence indicating no significant common method biases.

Another technique to address the issue of CMV is to use confirmatory factor analysis with a single factor to retest [23]. The method loads all items into single factor for testing. If it is over 50 for all the items, the results indicate that CMV occurs [24].

Alternatively, the confirmatory factor analysis presents a slightly lower model fit compared with other statistical models [23]. In this study, confirmatory factor analysis was performed using Lisrel 8.80, and the results indicated that, compared with correlated eight-factor models, single-factor analysis presented a lower model fit. Chi-squared is also affected by the size of the correlations in the model: the larger the correlations, the poorer the fit. As shown in

Table 2. Assessment of model fit in CFA.

Model	χ2	DF	RMSEA	CFI	SRMR	PNFI	GFI	AGFI
Single factor analysis	2740.22	456	0.09	0.91	0.08	0.80	0.85	0.67
Correlated 8-factor models	1653.71	436	0.06	0.98	0.05	0.86	0.91	0.88
Assessment	Not significant		<0.08	>0.90	<0.05	>0.50	>0.80	>0.80

, the issue of CMV was not found in the present study.

Moreover, multicollinearity between latent variables, according to Warp PLS 7.0 statistical results, shows VIF ranges from 1.162 to 2.727. While VIF may be less than 10 or less than 5 depending on the research need, the ideal VIF should be less than 3.3 [25,26,27]. The research results showed the VIF value for latent variables was less than 3.3, indicating that the issue of multicollinearity was not presented.

3.2.2. Reliability

According to Fornell and Larcker [28], the composite reliability and the Cronbach’s α are acceptable if they are equal to or greater than 0.70. The composite reliability for latent variables, calculated using standardized factor loadings and the error variance of observed variables, is greater than 0.70, reflecting a good internal consistency. In this present study, the composite reliability of various research constructs and Cronbach’s α all exceeded 0.70, showing the reliability of each model was acceptable. Results are shown in

Table 3. Reliability analysis.

Dimensions	Composite Reliability	Cronbach’s α
performance expectancy (PET)	0.86	0.79
effort expectancy (EE)	0.91	0.86
social influence (SI)	0.81	0.72
facilitating conditions (FC)	0.82	0.72
hedonic motivation (HM)	0.89	0.83
price value (PV)	0.86	0.78
habit (HB)	0.86	0.79
behavioral intention (BI)	0.89	0.83
use behavior (UB)	1.00	1.00

.

3.2.3. Convergent Validity

Convergent validity examines the extent to which measures of a latent variable share their variance, and how different they are from others. According to Hair, Black, Babin, and Anderson [29], the factor loading of items to their corresponding latent variables should be higher than 0.5. If less than 0.50, the corresponding item has to be removed. As shown in

Table 4. PLS loading and cross loading.

Variables	PET	EE	SI	FC	HM	PV	HT	BI
PET1	0.81
PET2	0.75
PET3	0.84
PET4	0.73
EE1		0.81
EE2		0.84
EE3		0.88
EE4		0.84
SI1			0.73
SI2			0.67
SI3			0.81
SI4			0.67
FC1				0.69
FC2				0.76
FC3				0.81
FC4				0.66
HM1					0.83
HM2					0.83
HM3					0.85
HM4					0.75
PV1						0.75
PV2						0.78
PV3						0.77
PV4						0.79
HT1							0.83
HT2							0.83
HT3							0.73
HT4							0.74
BI1								0.79
BI2								0.84
BI3								0.84
BI4								0.80

Note: PET: performance expectancy; EE: effort expectancy; SI: social influence; FC: facilitating conditions; HM: hedonic motivation; PV: price value; HB: habit; BI: behavioral intention; UB: use behavior.

, the factor loading of assessed items fell within 0.66 and 0.85 (>0.50). The factor loadings of the study variables were all greater than the acceptable suggested standard, indicating a good convergent validity.

3.2.4. Discriminant Validity

Discriminant validity, according to Chin [19], is assessed by demonstrating the average variances extracted (AVE) among the latent variables. It is determined by comparing the square root of the AVE to the correlation of the latent variables. In addition, Venkatesh, Thong, and Xu [18] suggested that the AVE should be equal to or greater than 70. As shown in

Table 5. Discriminant validity of constructs.

	PET	EE	SI	FC	HM	PV	HT	BI	UB
PET	0.78	0.57	0.61	0.56	0.49	0.61	0.45	0.59	0.25
EE	0.57	0.84	0.52	0.57	0.26	0.39	0.39	0.40	0.26
SI	0.61	0.52	0.72	0.61	0.49	0.51	0.50	0.52	0.25
FC	0.56	0.57	0.61	0.73	0.50	0.58	0.48	0.56	0.27
HM	0.49	0.26	0.49	0.50	0.82	0.62	0.44	0.63	0.27
PV	0.61	0.39	0.51	0.58	0.62	0.77	0.57	0.68	0.26
HB	0.45	0.39	0.50	0.48	0.44	0.57	0.78	0.64	0.30
BI	0.59	0.40	0.52	0.56	0.63	0.68	0.64	0.82	0.32

Note: PE: performance expectancy; EE: effort expectancy; SI: social influence; FC: facilitating conditions; HM: hedonic motivation; PV: price value; HB: habit; BI: behavioral intention; UB: use behavior.

, the square root of the AVE of all constructs in the research model exceeded 70, and were also higher than correlation coefficients in the same column and row of the same construct. It is evident that the measurement model has demonstrated a very good convergent validity.

The statistical results reported that the issue of multicollinearity was not presented. The composite reliability and Cronbach’s α exceeded the suggested standard, representing an acceptable reliability. The measurement model passed the test of convergent validity and discriminant validity. The research model was reliable and valid, indicating it can further analyze the moderating effects of constructs.

Diagonals represent the average variance extracted (the square root of the average variance extracted in the parentheses) while the other entries represent the correlations.

3.3. The Structural Model and Hypothesis Testing

Following the validity and reliability tests, based on the UTAUT2 model, the research developed 16 hypotheses and used Warp PLS 7.0 to test the hypotheses. Results are shown in Figure 1. The following will describe the test results in detail.

Hypothesis 1 (H1).

Performance expectancy of university PE students using social networking sites had a significant positive effect on behavioral intention (β = 0.13, p < 0.05), indicating higher performance expectancy results in higher behavioral intention to use social networking sites..

Hypothesis 2 (H2).

Effort expectancy of university PE students using social networking sites had no effect on behavioral intention (β = 0.02,p > 0.05). The research findings did not align with the hypothesis.

Hypothesis 3 (H3).

Social influence of university PE students using social networking sites had no effect on behavioral intention (β = −0.02, p > 0.05). Thus, the hypothesis was not supported..

Hypothesis 4a (H4a).

Facilitating conditions of university PE students using social networking sites had a significant positive effect on behavioral intention (β = 0.07, p < 0.05). A higher level of facilitating conditions led to higher behavioral intention.

Hypothesis 4b (H4b).

Facilitating conditions of university PE students using social networking sites had a significant positive effect on use behavior (β = 0.12, p < 0.05), indicating a high level of facilitating conditions resulted in more frequent use of social networking sites.

Hypothesis 5 (H5).

Hedonic motivation of university PE students using social networking sites had a significant positive effect on behavioral intention (β = 0.24, p < 0.05). High a level of hedonic motivation resulted in a higher level of behavioral intention.

Hypothesis 6 (H6).

Price value of university PE students using social networking sites had a significant positive effect on behavioral intention (β = 0.20, p < 0.05). High a level of price value resulted in a higher level of behavioral intention.

Hypothesis 7a (H7a).

Use behavior of university PE students using social networking sites had a significant positive effect on behavioral intention (β = 0.27, p < 0.05). High a level of use behavior indicated a high level of behavioral intention.

Hypothesis 7b (H7b).

Price value of university PE students using social networking sites had a significant positive effect on behavioral intention (β = 0.13, p < 0.05), presenting that a high level of use habit led to a high level of use behavior.

Hypothesis 8 (H8).

Behavioral intention of university PE students using social networking sites had a significant positive effect on use behavior (β = 0.18, p < 0.05), indicating a stronger use habit led a high level of behavioral intention.

Hypothesis 9 (H9).

Gender had no significant negative moderating effect on the relationship between PE and BI (β = −0.09, p < 0.05). Males had less effect on the relationship compared to their female counterparts. Age, however, did not have a significant moderating effect on the relationship between PE and BI (β = −0.01, p > 0.05), which means the research findings did not support the hypothesis.

Hypothesis 10 (H10).

Gender did not have a significant negative moderating effect on the relationship between EE and BI (β = −0.01, p < 0.05). The hypothesis was not supported. However, students’ UB had a significant positive effect on the relationship between EE and BI (β = 0.14, p < 0.05). The more students use the social networking site, the stronger the relationship.

Hypothesis 11 (H11).

A significant negative moderating effect of gender was found to exist on the relationship between SI and BI (β = −0.06, p < 0.05). Males had less effect on the relationship than females. The findings supported the hypothesis. On the other hand, while age did not have a significant moderating effect on the relationship between SI and BI (β = -0.04, p > 0.05), a significant moderating effect of UB was found to exist on the relationship (β = 0.18, p < 0.05). Results indicated that frequency of using the SNSs had a positive impact on the relationship between SI and BI.

Hypothesis 12a (H12a).

A significant negative moderating effect of gender was reported to exist on the relationship between FC and BI (β = 0.06, p < 0.05). Males had less effect on the relationship than females. Furthermore, age had no moderating effect on the relationship between FC and BI (β = 0.01, p > 0.05), and the hypothesis was not supported. Nevertheless, students’ experience of using SNSs had a significant negative moderating effect on the relationship between FC and BI (β = −0.18, p < 0.05). More experience did not lead to a stronger relationship between FC and BI.

Hypothesis 12b (H12b).

Age did not significantly moderate the relationship between FC and UB (β = 0.01,p > 0.05), and the hypothesis was not supported. PE students’ experience of using SNSs did not have a significant moderating effect on the relationship between FC and UB (β = 0.04, p > 0.05), and the hypothesis was not supported, either.

Hypothesis 13 (H13).

There was no significant moderating effect of gender on the relationship between HM and BI (β = −0.05, p > 0.05). Thus, the hypothesis was not supported. Age did not significantly moderate the relationship either (β = 0.05, p > 0.05), and the hypothesis was not supported.

Hypothesis 14 (H14).

Gender, age, and use behavior, respectively, did not have any significant moderating effect on the relationship between PV and BI (β = 0.02, p > 0.05; β = 0.01, p > 0.05; β = −0.03, p > 0.05). Consequently, hypotheses were not supported.

Hypothesis 15a (H15a).

Gender did not have a significant moderating effect on the relationship between UB and BI (β = −0.04, p > 0.05). The hypothesis was not supported. In contrast, age significantly moderated the relationship between UB and BI (β = 0.08, p > 0.05), indicating the older students became, the more positive the relationship grew. Use experience also had a significant positive moderating effect on the relationship (β = 0.09, p < 0.05). Thus, accumulated use experience of SNSs led to a more positive relationship.

Hypothesis 15b (H15b).

Gender did not significantly moderate the relationship between BI and UB (β = −0.01, p > 0.05), and thus the hypothesis was not supported. No positive and significant moderating effect of age on the relationship was reported to exist (β = 0.03, p > 0.05). Thus, the hypothesis was not supported.

Hypothesis 16 (H16).

Use experience did not significantly moderate the relationship between BI and UB (β = 0.03, p > 0.05).

3.4. Explanatory Power (R²)

Explanatory power refers to the ability to generate testable predictions of the research model. It is the percentage of the variance of the endogenous variable explained by all exogenous variables. A high value indicates a better predictability [30]. According to Hair, Hult, Ringle, and Sarstedt [26], R² can be classified into one of three categories: weak (0.25), moderate (0.50), or substantial (0.75). As shown in Table 2, R² = 0.624 suggested a moderate explanatory power, indicating performance expectancy, effort expectancy, social influence, facilitating conditions, hedonic motivation, price value, and use habit together explained 62.4% of the overall variance of behavioral intention. R² = 0.134 presented weak explanatory power, suggesting facilitating conditions, use habit, and behavioral intention together explained 13.4% of the overall variance of use behavior.

4. Discussion

This present research applied UTAUT2 to verify university PE students’ use of social networking sites. Our findings reported that PET, FC, HM, PV, and UB had significant positive impacts on BI, and were consistent with El-Masri et al. [16], Chong and Ngai [31], Ferreira Barbosa et al. [32], Venkatesh et al. [18], and Chang, Liu, Huang, and Hsieh [33]. However, EE and SI were reported to have no significant impact on BI. EE in this particular study was a belief that the use of a particular technology would be easy and effortless, while SI referred to individuals who have influence on how PE students perceive the use of SNS. Venkatesh et al. [12] stated that as students accumulated experiences using SNS, they were more likely to perceive the use of a technology easy and effortless. In this study, 96.2% of students were reported to have at least one year of experience using SNS, and 44.4% had more than four years’ experience; therefore, no significant effect of EE and SI on BI was found to exist. Results were consistent with Wang, Cheng, and Hsu [34], and Chang et al. [33].

By degree of influence, variables that had impacts on BI were HB, HM, PV, EE, and FC. Results were similar with Escobar-Rodriguez and Carvajal-Trujillo [35] on buying airline tickets on the Internet. They discovered that consumers could easily have access to mobile devices and find a better rate on the Internet. It was reported 94.8% of students were using mobile devices, and finding things online is a simple and cost-effective task. The study results showed that students built a sense of achievement and happiness when using SNS, and results were consistent with Brown and Venkatesh [36] who suggested optimal experiences using technology are drivers towards accepting and using new technology.

Moreover, this research discovered that BI, FC, and HB all had significant impacts on UB, and BI had the most impact. This discovery was similar to early research [35,37,38].

Variables in this research model explained 63.4% of overall variance of BI, and 13.4% of UB. These results were similar to Escobar-Rodriguez and Carvajal-Trujillo [35] and Chang et al. [33]. Nevertheless, the explanatory power on UB was weaker than that found by Escobar-Rodriguez and Carvajal-Trujillo [35] and Venkatesh et al. [18], but similar to Chang et al. [33]. The difference in the explanatory power can result from cultural differences or technology. The empirical evidence based on research conducted in Taiwan usually presented much lower explanatory power.

Regarding moderating effect, the moderating effect of male students on the relationship between PET and BI was less strong than females. These findings were consistent with Wu and Lin [39], Liu [40], and Chang et al. [33]. These findings suggested that female students should be more open to new technology if they know how to find what they need easily and effectively. Moreover, the moderating effect of male students on the relationship between SI and BI was less strong than their female counterparts; this, however, was inconsistent with Liu [40] and Chang et al. [33]. The research results showed that female students were more concerned about peer opinion. There was also a less moderating effect of male students on the relationship between FC and BI than females, suggesting female students were more likely to use SNS if they are offered help to use SNS.

Our study results suggested that age is a crucial factor in SNS use. Senior students were more likely to use SNS because they had fewer classes to attend. SNS use is frequent; thus, their intention was higher.

As students started to accumulate experience with SNS, the relationship between EE and BI improved. Those who had much experience with SNS were more likely to build up their intention compared to those who had less experience. Results indicated that proper guidance and instruction should be provided to those with less SNS experience.

Using SNS for interaction has become a phenomenon. SI had an impact on BI among students with enough SNS experience. Furthermore, some students use SNS every day because they believe SNS has become part of their lives. Therefore, frequent use led to high levels of BI. Students with much SNS experience were less likely to need assistance, and thus their perceived FC had less impact on BI.

5. Conclusions and Recommendation

5.1. Conclusions

This study adapted the UTAUT2 model to explore the factors affecting university PE students’ social media use intention. Among those predicting variables in the model, except effort expectancy, the study found performance expectancy, facilitating conditions, hedonic motivation, social influence, price value, and habit had positive impacts on social media use intention. In addition, gender, age and experience had moderating effects among the predicting variables and social media use intention. The results indicate that using UTAUT2 in predicting university PE students’ social media use intention is applicable.

5.2. Recommendation

Out of seven variables in the UTAUT2 model, UB and HM had the most impact on BI among university PE students. In order to manage a class or a sports team, class instructors and sports coaches can use SNS to enhance their teaching in general. Instructors should encourage students to use SNS, and at the same time be aware if students become addicted. Instructors should first familiarize themselves with SNS, direct students to take advantages of SNS, and advise on potential issues if students become addicted. Instructors also need to plan ahead if they are incorporating SNS into the teaching curriculum and conducting assessments.

Smartphones and laptops are indispensable in students’ routine life. Furthermore, Wi-Fi is readily accessible in many public places and SNSs develop quickly due to easy internet connection. As mentioned earlier, SNSs not only provide entertainment but also foster interpersonal relationships. Through SNSs, students interact with each other anytime anywhere, and instructors can build a trusting rapport with students at the same time.

Results reported that the UTAUT2 model had a high explanatory power on BI, but it was low on UB, indicating there might be some other variables that have impacts on UB. It is recommended that future studies incorporate possible variables to build a research model that better explains students’ behaviors. Furthermore, the moderating effect of age was not significant, probably because the experience variable was categorized into 2–3 years, 3–4 years, and more than 4 years. Consequently, future research may develop a use experience measurement scale at half-year intervals, focus on continuous measurement, and analyze a linear regress of two variables.

Due to the COVID-19 pandemic, many courses were delivered through the internet. In the beginning, both instructors and students were not comfortable with the adjustments. For many instructors, using the web as teaching channel was new to them. Unwillingly, but forced by situations, instructors used Zoom, Microsoft Teams, Google Meets, etc., to perform online teaching, which boosted e-learning and helped students to be more familiar with the new technology. To comply with these needs, instructors were urged to design teaching materials suitable for e-learning. Therefore, we suggest using the UTAUT model to study how students and instructors accept of this new form of learning and teaching to further understand their adaptions to e-learning and e-teaching, which can provide insights into ways for both to comply with the situations under the pandemic.