1 Introduction

A problem that has affected our physical well-being in recent times and has alienated young people from physical activity is the excessive time spent in front of the television screen or video game console (Baranowski et al., 2008; Kracht et al., 2020). Consistent with this idea, increased time spent playing video games has been associated with higher rates of student attention problems, as reported by teachers (Swing et al., 2010), decreased sleep efficiency (Hisler et al., 2020) and reduce home study time, and potentially, academic performance (Abbasi et al., 2021; Barlett et al., 2009).

In addition, obesity and overweight issues are two of the world’s most significant health problems (Chooi et al. 2019; Xi et al., 2020). The number of overweight or obese people is increasing every year due to our lifestyle all around the world (Fan & Zhang, 2020; Hemmingsson et al., 2020; Kenney et al., 2020).

However, it has been proven that the moderate use of video games can improve students’ cognitive development (Rojas et al., 2021), academic performance (Ibáñez et al., 2020), and physical condition (Williams & Ayres, 2020). Further to this, it facilitates the initiation of physical exercise and sports for students with lower sports performance and generates adherence, thanks to the motivation generated (Rogers et al., 2020).

Therefore, the main objective of this article is to carry out a study and analyse the factors that influence the acceptance of video games as a tool for improving academic performance. If academic performance can be improved, it could lead to a better perception of the subject of physical education in students and a greater motivation for them to engage in physical exercise (Williams & Ayres, 2020).

Numerous research studies show how beneficial it is to use information and communication technologies for all students at different school stages: in preschool (del Carmen Ramírez-Rueda et al., 2021), primary education (Chai et al., 2011), secondary education (Fernández-Gutiérrez et al., 2020; Volman et al., 2005), and higher education (Estriegana et al., 2019). Learning is easy because the learner sets the pace of the process and experiences actions that are not easy to perform in practice. Once these actions have been learned in the virtual environment, it is easier to transfer them to real practice (Biddiss & Irwin, 2010). From this perspective, it is important to analyse the factors that influence the acceptance of video games as a tool for improving academic performance.

1.1 Video game-based Learning, Physical Activity, and enjoyment

The purpose of video game-based learning is to generate an attractive product for students to improve their education through a tool that generates motivation for its users (de-Marcos et al., 2016). Obtaining improvements by progressing through the video game and gaining rewards generates a satisfaction for the players that leads to motivation to continue playing and this condition is used not only to enjoy playing but also to encourage learning (de-Marcos et al., 2016). Several authors have investigated these practices and the positive effects generated by the motivation of obtaining rewards from video games, which influences the improvement of acceptance of this type of technology (Estriegana et al., 2019). The game-based learning environment demonstrates great potential to inspire students and engage them in an active learning process. For example, Dominguez et al., (2013) reported that students who completed the gamification experience achieved higher scores on the actual work. In a similar way, Hwang et al., (2012) implemented an experimental model using online game-based problem-solving activities and reported significant improvements in students’ learning attitudes, interest in learning and acceptance of technology.

Similarly, numerous research studies have analysed the impact of the use of video games in combination with physical activity and have shown improvement in academic performance (Ibáñez et al., 2020; Williams & Ayres, 2020). The use of sports video games is one of the ways to attract adolescents closer to sports (Biddiss & Irwin, 2010). By playing this tool, fun learning and global development of students are usually achieved (Peng et al., 2013). Different authors, such as Lanningham-Foster et al., (2009) and Vernadakis et al., (2015) have developed interventions where students are explained the different basics and sport tactics that can be developed in sports video games, and this generates a greater facility to perform that sport in a real situation. It has been concluded that an effective use of video games focusing on sports practice leads to a reduction in the number of hours of sedentary activity while increasing the hours of physical activity (Foley & Maddison, 2010).

When the enjoyment generated by sport and the enjoyment of video games are linked together, adherence to physical activity is generated. This adherence will make it easier for these young people to maintain practicing physical activity throughout their lives (Garn & Cothran, 2006). When children play games and practice sports, they do it to have fun or to pass the time and entertain themselves (Kracht et al., 2020). Enjoyment is understood as the opposite of boredom and when we try to define the scale that measures the distance between these two terms, motivation appears. The more motivation a game generates, the more fun it is and, conversely, the less motivation the more boring it is (Al-Adwan et al., 2020; Kendzierski & DeCarlo, 1991). Therefore, for children and teenagers to engage in physical activity, a high level of motivation must be achieved in order to generate fun (Standage et al., 2003). In addition, motivation and enjoyment generate attraction to physical activity and games, which means that as adults they will maintain these healthy habits, improving their quality of life (Marker et al., 2018).

1.2 Technology Acceptance Model (TAM)

The TAM model, formulated by Davis (1986; 1989), suggests that a person’s acceptance of any technology is determined by the beliefs he or she has about the consequences of its use. This model adapts the psychological Theory of Reasoned Action (TRA or ToRA) of Fishbein & Ajzen (1975), which aims to predict people’s behaviour based on their intentions and attitudes. The model (Davis, 1989) suggests that the Attitude Towards Use (ATU) of an information technology system is based on two prior factors: Perceived Usefulness (PU) and Perceived Ease of Use (PEU). The most widely accepted definition of attitude, although, according to the authors themselves, may be a highly ambiguous statement, which is “a learned predisposition to respond in a consistently favourable or unfavourable manner with respect to a given object” (Fishbein & Ajzen, 1975, p. 6). On the one hand, PU is considered as an extrinsic motivation to the person and Davis (1989) defines it as “the degree to which a person believes that using a technology will enhance performance” (p. 320). On the other hand, another factor of the TAM model, PEU, can be understood as “the degree to which a person believes that using a particular system would be free of effort” (Davis, 1989, p. 320).

In addition, Ajzen (1991) generated a correlation between students’ attitude towards the use of technology and their Intention to Use it (IU). As suggested by Eze et al., (2021), in order to know whether a technology will be used optimally, it is necessary to identify different external factors that may impact the usefulness and ease of use perceived by ICT users.

It should be noted that the TAM model has been revised and extended with the consideration of new factors (Escobar et al., 2014) that have been named as TAM2 (Venkatesh & Davis, 2000) and TAM3 (Venkatesh & Bala, 2008). In any case, although the TAM model has evolved over time, the model is still constituted at its core by a simple set of factors identified in the first formulation, which is possibly its great advantage in application (Davis et al., 2020).

2 Method

2.1 Research Model and Hypotheses

In the analysis of the model, it is suggested that certain external factors can serve as predictors and, therefore, influence the usefulness and ease of use perceived by students with respect to technology (Yong et al., 2010). It is true that different studies have analysed some of these factors, such as accessibility (Culnan, 1985; Bhattarai & Maharjan, 2020), enjoyment (Al-Adwan et al., 2020; Kendzierski & DeCarlo, 1991), technological competence (Cabero et al., 2016; Vanduhe et al., 2020), learning through videogames (Cabero & García, 2016; Himang et al., 2021), and final basketball grades (McMahon & McMahon, 2020; Klusemann et al., 2012); nevertheless, none of them have done it together in the same model in the context of physical education. To define the hypotheses of this study, each external factor of the TAM model is described below.

2.1.1 Accessibility

When we refer to accessibility, we can refer to the physical or cognitive level (Culnan, 1985). Furthermore, perceived physical accessibility suggests the extent to which a user has physical access to the technology. Also, we have perceived information accessibility, which indicates the ability to retrieve desired information from the system (Bhattarai & Maharjan, 2020). Numerous studies suggest that perceived accessibility can influence the use of communication technology (Culnan, 1985; Bhattarai & Maharjan, 2020).

2.1.2 Enjoyment

Perceived enjoyment alludes to the degree to which the use of a technology is seen to be enjoyable (Davis, 1989). It is the user’s perception of the fun and pleasure derived from using the application. It refers to the hedonic attractiveness, aesthetic beauty, perceived pleasure, playfulness, or fun derived from using a system or an interface. Using new apps or technology is motivating and users enjoy them. Various studies on perceived enjoyment (Al-Adwan et al., 2020; Hussain, et al., 2016) have shown that users’ happiness while using an application have a huge effect on their intention to use the application. Hussain et al. (2016) found that pleasure or perceived enjoyment is related to the time of use. Al-Adwan et al., (2020) observed that liveliness had a huge effect on the expectation to utilise a technology and user derived entertainment and fun from the use of an application imparts on their acceptance of the application.

2.1.3 Technological competence

Perceived control and perceived ability were first introduced by Csikszentmihalyi (1988). This author noted that users in a state of relaxation tend to show their intrinsic motivations while performing and completing specific tasks and feel coherence between activities and self-awareness. One of the possible approaches to understanding the state of flow is to consider it as a particular function of ability and control that is deeply related to the user’s perceptions and activities. Csikszentmihalyi (1988) defined perceived control and perceived skill as the user’s perception and behaviour about the degree of challenge involved in performing a given activity and the user’s skill in performing that activity. In the present study, perceived control and perceived dexterity are defined as users’ perception of how skilled the user is when playing the game.

Regarding the changing state in mobile technologies, the higher the level of skill in using mobile devices, the higher the degree of control of mobile devices (Vanduhe et al., 2020). Therefore, this study considered these two factors in one construct. Previous studies showed that perceived control and perceived dexterity are key factors in determining the perceived usability and enjoyment of specific technologies and services (Cabero et al., 2016; Vanduhe et al., 2020; Venkatesh & Davis, 2000).

2.1.4 Learning through videogames

One problematic situation that could be influenced is the environment in the face of the unknown. An in-depth review was done to prove that the students’ prior learning experience with video games and technology has an enormous influence on participation in an online learning program (Cabero & García, 2016; Himang et al., 2021) and playing video games can benefit social relationships and this may influence perceived accessibility (Greitemeyer & Osswald, 2011).

2.1.5 Final basketball grades

This factor is the result of the combination of three assessment resources on the practical and theoretical knowledge of basketball. Firstly, the Basketball Rules Test that consists of 25 multiple-choice questions according to the Official Basketball rules. Secondly, the Basketball Knowledge Test that has the same format as the previous one, but the students had to select the correct answer according to the technical-tactical knowledge of basketball. Finally, the Individual Technical Basketball Test (ITBT) for use by 16–17-year-old adolescents (Cárdenas & Moreno, 1996).

2.1.6 Hypotheses

Based on previous research, we hypothesised that ATU would positively impact on IU (H1). Accessibility would positively impact on PEU (H2), and on PU (H3). Thus, we hypothesised that enjoyment would positively impact on ATU (H4), on accessibility (H5), on PEU (H6), on IU (H7), on technological competence (H8), on PU (H9), and on learning through videogames (H10). PEU would positively impact on ATU (H11) and on PU (H12). At the same time, IU would be a significant factor in predicting variance on final basketball grades (H13). In addition, we hypothesised that technological competence would positively impact on accessibility (H14), on PEU (H15), on PU (H16), and on learning through videogames (H17). PU would positively have an impact on ATU (H18) and on IU (H19). Finally, we hypothesised that learning through videogames would positively impact on accessibility (H20), on PEU (H21), on final grades (H22) and on PU (H23).

Based on the literature review and the hypotheses proposed, the structural model shown in Fig. 1 has been developed.

Fig. 1
figure 1

Conceptual research framework

Full size image

2.2 Instruments used and their validation

2.2.1 Instruments

Regarding the diagnostic instrument to be used, the one usually developed for the TAM model by Davis (1989), will be used. This instrument is still used in all the investigations that fall within the TAM model, consisting of a series of questions with a Likert-type construction and seven response options. The research by Davis (1989) and Venkatesh & Davis (2000) was used to adapt the TAM scale of ATU, IU, PU and PEU. The learning through videogames questions were adapted from Cabero & García (2016) and Himang et al., (2021). Accessibility items were selected from Bhattarai & Maharjan (2020) and Culnan (1985). To analyse technological competence, we relied on Cabero et al., (2016) and Vanduhe et al., (2020).

The Physical Activity Enjoyment Scale (PACES) created by Kendzierski & DeCarlo (1991) was used to assess motivation and enjoyment. It was originally an 18-item scale developed to assess how much fun young adults had while engaging in physical activity (Kendzierski & DeCarlo, 1991). This instrument was validated for use in other population groups such as children (Paxton et al., 2008), adolescent girls (Motl et al., 2001), and older adolescents (Dunton et al., 2009).

The items used to assess academic performance were obtained from the ITBT by Cardenas and Moreno (1996), a Basketball Rules/Regulations Test and a Basketball Knowledge Test. The ITBT consisted of three game situations with different phases that were originally designed to measure individual technical-tactical improvement, time spent and satisfaction (Cárdenas & Moreno, 1996). This test was modified for the content and purpose of the present study looking for individual technique optimisation based on the study of Klusemann et al., (2012) and McMahon & McMahon (2020). The three game situations were merged into a single situation in which the categories observed were dribbling, passing, shooting, lay-up and time spent. The student starts on the right side from the baseline and performs a slalom by bouncing until he makes a shot at the opposing basket. Afterwards, he makes some passes with another assistant to the initial basket where he finishes with a lay-up.

The video game used to train the students was NBA 2K16 (Fig. 2), which was developed by Visual Concepts and released by 2 K Sports. During the gameplay of this resource students can experience playing basketball with professional NBA players and perform the most complex and realistic individual technical actions that can be achieved. In addition, it has play setting functions as if you were a coach making a tactical decision. Therefore, this experience allows you to experience in a virtual environment, action that can be brought to reality thanks to the model visualised and internalised by the students (Rogers et al., 2020).

Fig. 2
figure 2

Gameplay of NBA 2K16

Full size image

2.2.2 Participants

The sample of students (n = 166) belonged to a charter school in Alcalá de Henares, in the Community of Madrid (Spain). Their mean age was 15 years, with 84 girls (50.6%) and 82 boys (49.4%). The sample was selected purposively, due to the ease of access to the educational center. Arbitrary assignment was carried out by computerized generation of random student numbers and assignment of students to groups based on these numbers. 84 participants (50.6%) were assigned to the experimental group and 82 participants (49.4%) to the control group.

2.3 Data Analysis

To analyse the model, this study used a regression analysis of latent factors based on the optimization technique of partial least squares (PLS). SmartPLS 3.3.3 is used in this study. PLS is a multivariate technique for testing structural models that estimates model parameters to minimize the residual variance of the model’s dependent factors (Hair et al., 2014).

2.3.1 Justification of sample size

Although PLS is suitable for small samples, since it does not need parametric conditions (Hulland, 1999), it is necessary to determine the size of the sample. For this, Hair et al., (2014) recommend using programs like G * Power 3.0. (Faul et al., 2007).

Thus, it is necessary to specify the effect size (ES), the value of the alpha significance level (α) and the power (β). Generally speaking, an alpha level of 0.05 and a power of 80% is accepted. In this case, the multiple regression study has been carried out with nine predictors, a mean effect size (ES) 0.15, an alpha of 0.05 and a power of 0.95 (following Cohen 1994), which allows us to find out the size of the sample. When applying the priori analysis it is observed that the result is n = 166 subjects. The sample exceeds any requirement demanded to carry out the analysis of the measurement models and the structural model.

2.3.2 Measurement model evaluation

Based on our findings, the measurement model is adequate. Except for a few values that exhibit a slight degree of non-normality, both the kurtosis and skewness values of the indicators are within the acceptable range of -1 and + 1. (Hair et al., 2014). In the case of PU4, PACES 13 and PACES 16 presented slight variations the kurtosis and skewness and ACU2 only in kurtosis.

Carmines & Zeller (1979) indicated that all standardised loads (λ) must be greater than 0.707. As shown in Table 1, all values satisfy this condition, indicating that the reliability of the individual item is adequate.

Table 1 Outer model loadings
Full size table

The simple reliability of the measurement scales was calculated using Cronbach’s alpha values, Nunnally & Bernstein (1994) suggest that all of them are greater than 0.70.

The composite reliability indicates that a high level of internal consistency reliability has been demonstrated among the latent factors since all the Values are above 0.70 (Werts et al., 1974).

Fornell & Larcker (1981) suggest that the analysis of variance all the values of the average variance extract (AVE) should be higher than 0.50, as shown in Table 2.

Table 2 Cronbach’s alpha coefficients, Rho_A, construct reliability and average variance extracted (AVE)
Full size table

Furthermore, Fornell–Larcker Criterion matrix show the value in diagonal represent the Square-root of AVE. Therefore, the values of the diagonal are not all one (Abab et al., 2021). According to Fornell & Larcker (1981), discriminant validity is present when the shared variance for all constructs in the model is not greater than their AVEs (Canguende-Valentim & Vale, 2021). Table 3 shows the square roots of the AVE on the diagonal are greater than other correlation values between the factors.

Table 3 Discriminant validity matrix (Fornell-Larcker Criterion)
Full size table

Henseler et al., (2015) analysed the discriminant validity measures using the heterotrait-multitrait (HTMT) method, this method obtains the geometric mean of the average monotrait-heteromethod correlation of both factors. Specificity rates for HTMT 0.85 are close to 100% with construct correlations of 0.70. In Table 4, the HTMT ratio for learning through videogames (Learning) and ATU, at 0.847, were found to be less than the 0.85 cut-offs and significantly less than the 0.95 cut-off recommended for conceptually close constructs (Henseler et al., 2015). This shows a good fit of discriminant validity between our measures of learning through videogames (Learning) and ATU measures (Henseler et al., 2015).

Table 4 Discriminant validity matrix (Heterotrait-Monotrait Ratio Criterion)
Full size table

To assess the best fit of the model, it must be analysed using the value obtained from the residual mean square root (SRMR). In our case, the value was 0.067, which did not exceed the approved 0.08 (Hu & Bentler, 1998).

2.3.3 Structural model analysis

From the literature review, the model shown in Fig. 1 has been developed. Using a technique known as bootstrapping, the PLS program can generate T-statistics for significance testing of both the inner and outer models (Chin, 1998). In this procedure, a large number of subsamples (5000) are drawn from the original sample with replacement to yield bootstrap standard errors, which yield approximate T-values for structural path significance testing.

Following the completion of the bootstrapping procedure it was observed that the R-squared values are all between 0 and 1. Since R-squared should be high enough for the model to reach a minimum level of explanatory power (Falk & Miller, 1992).

In our case, the variance explained (R-squared) in the dependent constructs and the path coefficients for the model are shown in Fig. 1; Table 5. In our case, the R-squared values are greater than 0.10, with a significance level of t > 1.64.

Table 5 Structural model results
Full size table

3 Results

The standardization of the regression coefficients shows the hypothetical relationships between constructs. In the cases in which there is a significant change, the magnitude and statistical significance are greater than the T statistic of (t (4999), one-tailed test) 1.64. This process allows us to study and validate the hypotheses. From the results shown in Table 6, we can see that the relationships were positive and with high significance for the most part.

Table 6 Structural model results. Path significance using percentile bootstrap 95% confidence interval (n = 5000 subsamples)
Full size table

When percentile bootstrap is applied to generate a 95% confidence interval using 5.000 resamples, the hypotheses H3, H12 and H16 were not accepted because their confidence interval includes zero as shown in Table 6. Besides, Fig. 3 shows all the results that complete the basic analysis of PLS-SEM in our investigation.

Fig. 3
figure 3

Results of testing the model significance *P < .05. ** P < .01. *** P < .001

Full size image

Table 7 shows the amount of variance that each antecedent factor explains on each endogenous construct. Figures of R square are outstanding (Table 7) for almost all values. They are greater than 0.275 what is a high level. There is one value less than 0.150. Thus, cross-validated redundancy measures show that the theoretical/ structural model has a predictive relevance (Q2 > 0). As all values are greater than 0.124, they indicate it is a high level and predictive relevance.

Table 7 Effects on endogenous factors (extended model)
Full size table

From the data obtained in Table 7, accessibility is important in explaining both PU and PEU, as well as IU and ATU. Thus, perceived environment accounts for 50.547%, 49.7%, 63.58%, and 70.3% of the factors, respectively.

4 Discussion

In view of the structural model of the hypotheses and the analysis of the results, we can see that ATU impacts positively on IU in line with what is indicated by Ibáñez et al., (2020). Accessibility directly and significantly conditions PEU as indicated by Bhattarai & Maharjan (2020), but it does not influence PU, contrary to what these authors indicate. This is because students consider that the accessibility of the application does not have a significant direct relationship on perceived usefulness, since the degree of accessibility of the application does not influence the learning process. But if students consider that the application helps them to learn, they will accept it as a learning tool, thus increasing the perceived usefulness, as indicated by Wen et al., (2022). For his part, Haddad (2018) indicates that accessibility can influence PU depending on the quality of the learning management system.

Enjoyment influences on ATU, on accessibility, on PEU, on IU, on the acquisition of technological competence, on PU, and on learning through videogames, since the greater the enjoyment with technology, the higher the quality of learning of theoretical and practical concepts, and the lower the resistance to the integration of technology in the learning process, in line with what is indicated by Kendzierski & DeCarlo (1991), and Al-Adwan et al., (2020).

In the case of PEU and ATU, they are significantly influenced due to the predisposition that users have towards the use of technology in the training process from easy-to-use tools and user adaptability to it, which favours the perception of usability of the tool within the student’s training process, as confirmed by Cabero et al., (2016) Vanduhe et al., (2020) and Venkatesh & Davis (2000). However, PEU did not have a positive influence on PU, since students consider that, when an application is easy to use, it does not always generate a perception of usefulness on their learning process. Different authors study these two concepts independently. Venkatesh & Davis (2000), Venkatesh (2000), and Rose & Fogarty (2006) have studied these concepts independently, because the determinants that influence PEU are different from the determinants that influence PU. From this perspective, PEU can be unrelated to PU.

Furthermore, the final grades obtained are directly related to IU and learning through videogames as motivational tools in the learning process, and indirectly influenced by the accessibility and technological capabilities that help students in their learning process, as argued by Williams & Ayres (2020).

Thus, we can see that technological competence influences accessibility, PEU, and learning through videogames, so that students who have greater technological skills can better acquire learning through videogames, in line with what is indicated by Escobar et al., (2014), Ibáñez et al., (2020), and Yong et al., (2010). On the other hand, technological competence did not influence PU. This is due, as we can observe in the results, to the fact that students with higher technological competence make the game easier to play and can focus on the learning process. However, this fact does not make them have a higher perception of use of the application in the learning process, as indicated by Bourgonjon et al., (2010) and Esteve et al., (2014).

In addition, we can observe that learning affects PU, as it is more related to the practical aspect of the tool in the learning process, and accessibility and technological competences influence PEU, as they are more related to the technical aspects of the application (Bourgonjon et al., 2010).

Finally, learning through videogames influence accessibility, PEU, and PU because students who enjoy using new technologies in their learning process have a higher ATU and IU, compared to technology and its use in other subjects, in line with what Cabero & García (2016) indicate.

5 Conclusion, Limitations, and Future Research

It can be concluded that there is a direct relationship between the factors analyzed in this study, namely, accessibility, enjoyment, technological competence and learning through video games, with the acceptance of using video games in the context of physical education. The higher the level in the different factors, the higher the acceptance of using video games as a learning aid. In addition, this positive relationship leads to improved academic performance in physical education.

Our findings show how video games become a relevant resource for the development and improvement of cultural competences, social and cognitive skills by secondary school students in the context of physical education. The role of motivation and enjoyment is an inspiration to learn and develop thinking skills, and it influences the technological competence, learning and accessibility.

Regarding the contributions of this study, it should be noted that the analysis conducted here is one of the few research works conducted with adolescents that examine the degree the degree of improvement offered by sports training with video games on academic performance and how the different factors studied jointly influence in the same structural model.

As for the educational implications emerging from this study, much of the literature on adolescents has focused on differences in the rates of female and male students who choose educational paths leading to STEM disciplines (e.g., Rodán et al., 2016). This study suggests that training these skills in an academic context could increase the potential pool of students who successfully enter STEM careers. Thus, it becomes increasingly necessary to investigate the use of sports video games as a moderator of the relationship between physical activity and technological competence.

These results can be considered as a guide for future research in this area, although it is important to remember that, due to their quasi-experimental nature, the results do not allow for causal conclusions. Although these results should be taken as preliminary, due to the small sample size and consequent low statistical power, our training task was relatively short. Furthermore, as in most school studies, the groups have been randomly assigned, rather than selected subjects, which increases the risk that the groups may not be equivalent.