Does gamification mediate the relationship between digital social capital and student Performance? A survey-based study in Spain

The International Journal of Management Education 21 (2023) 100846 Available online 10 July 2023 1472-8117/© 2023 Published by Elsevier Ltd. Does gamification mediate the relationship between digital social capital and student Performance? A survey-based study in Spain Jos´ e M. Fortuna * , Gabriel de la Fuente, Pilar Velasco Universidad de Valladolid, Department of Finance and Accounting, Avenida Valle del Esgueva, 6, 47011, Valladolid, Spain ARTICLE INFO Keywords: Higher education Gamification Social networks Academic performance Learning technologies ABSTRACT This article aims to investigate the potential mediating role of gamification in the relationship between students’ digital social capital and their academic performance. Social networks are informal platforms where students can develop their own skills in order to adapt to the digital environment and where there is visibility in public rankings. Consequently, we argue that digital social capital developed through the use of social networks can promote student motivation and engagement in online gamification practices, which in turn might help to enhance their course performance. Our empirical study applies a survey-based approach on a sample of 133 undergraduate students enrolled in a hybrid course in corporate finance at a Spanish public university during the COVID-19 pandemic. Empirical evidence suggests that stronger digital social capital (i. e. a greater number of following contacts in social networks) increases a student’s propensity to participate in Kahoot! gamification. Additionally, digital social capital has a positive indirect effect on a student’s academic performance, with this relationship being mediated by Kahoot! participation. This educational research encourages links between different digital technologies to be exploited to a greater extent in order to strengthen student engagement and maximize their academic performance. 1. Introduction The COVID-19 pandemic –and subsequent social distancing restrictions– forced many universities to replace traditional face-to-face teaching with hybrid courses (having some students in class, and others following the class online) (Ives, 2021; Kortemeyer et al., 2023). Such a shift in the teaching environment has significantly accelerated the integration of digital technologies into the educational domain. Evidence of this can be found in the use of videoconferencing for classes and tutorials (Correia et al., 2020), and in the expansion of e-learning tools such as Blackboard and Moodle (Aljawarneh, 2020). Given the less direct interaction between teacher and student under the hybrid teaching model (Fang et al., 2023), one primary challenge comes from the need to ensure students’ motivation and their follow-up in the courses. Class attendance forces students to participate in onsite activities. However, hybrid teaching weakens monitoring and allows students to freely decide whether to physically attend classes or not, and which particular activities to join in with. Consequently, stimulating a student’s engagement in the learning process becomes of the utmost importance in this hybrid setting (Tang et al., 2021). Moreover, endowing students with a practical set of skills to succeed in real-world decision-making is taking on a key role in terms of boosting student employability in the job market, which again urges looking at active learning methodologies (Okolie et al., 2022; Mu˜ noz Miguel et al., 2023; Yesildag & Bostan, 2023). * Corresponding author. E-mail addresses: [email protected] (J.M. Fortuna), [email protected] (G. de la Fuente), [email protected] (P. Velasco). Contents lists available at ScienceDirect The International Journal of Management Education journal homepage: www.elsevier.com/locate/ijme https://doi.org/10.1016/j.ijme.2023.100846 Received 17 December 2022; Received in revised form 3 July 2023; Accepted 5 July 2023 The International Journal of Management Education 21 (2023) 100846 2 Among the range of innovative teaching strategies, gamification has increased in popularity in recent years. Gamification extends game attributes to non-game environments in order to influence people’s learning-based behaviour or attitudes (Landers, 2015; Landers & Landers, 2015; Sanchez et al., 2020). Recent studies reveal the benefits of using gamification in the classroom to support student learning, such as through greater motivation (G¨ oksün & Gürsoy, 2019; Ekici, 2021), stronger class cohesion (Candan & Basaran, 2023), enhanced academic performance (Dias, 2017; Ekici, 2021; Ortiz-Martínez et al., 2022), and richer learning outcomes from other teaching strategies (Ekici, 2021; Candan & Basaran, 2023), to name but a few. Again, such digital-based learning requires students to display enough motivation to materialize the benefits of gamification activities into learning outcomes. Some works alert to the need to improve gamification design frameworks (Mora et al., 2017; Murillo-Zamorano et al., 2023) and to tackle the potential exhaustion of student engagement after the repetitive use of gamification (Sanchez et al., 2020), which poses a particular challenge to teachers. In an effort to address this weakness of gamification, another strand of literature has highlighted the motivational role played by social networks in the higher education context, which is felt to improve the teaching-learning process (Ranieri et al., 2012; Hortigüela-Alcal´ a et al., 2019; Mishra, 2020). This becomes a key issue for the current student generation -the so-called Generation Z-whose daily life flows naturally around Internet and social media (Biro, 2014). Much of the research into gamification has focused on the impact of using Kahoot! on student performance, either considering gamification in isolation or in combination with other active learning methodologies such as flipped classroom (Eriki et al., 2021). In contrast, our study examines the implementation of Kahoot! in combination with student use of social networks, which are a kind of digital community platform where students develop additional digital and social skills by themselves. This becomes even more relevant in the hybrid-learning context given the lack of social interaction and subsequent feeling of isolation that the COVID-19 pandemic has induced amongst students (Elmer et al., 2020). Recent works point out that digital social capital 1 can boost student academic performance indirectly (Salimi et al., 2022), which leaves room to explore alternative channels through other variables. On the other side of the coin, other studies have also alerted to the harmful effects of ‘too much of a good thing’ in the form of an overuse of social networks, which can lead to student distraction and therefore, impair their academic performance (Zimmer, 2022). We investigate the association between social network use, gamification engagement, and academic performance. We focus on this three-pillar set of variables, since research has thus far mostly addressed them separately, although the current demands facing the higher education environment have led them to coexist in many university courses. Extending our knowledge of the potential complementarities of social networks and gamified techniques thus proves crucial vis-` a-vis maximizing student academic performance. Our teaching experience was conducted at a Spanish public university during the autumn semester of the 2020/2021 academic year, as a part of an innovative teaching project based on the use of gamification via Kahoot! in several undergraduate corporate finance courses. This teaching experience is targeted at improving teaching quality as well as promoting active student learning in the hybridlearning teaching context prompted by the COVID-19 pandemic. The rest of the article is structured as follows. Section 2 reviews the literature about the learning value of digital gamification and how digital capital from social networks reinforces student motivation. This constitutes the theoretical base to formulate our hypotheses. Section 3 explains the data collection process based on a survey carried out amongst students, and describes the sample, variables, models, and estimation method. Section 4 presents the empirical findings, while Section 5 provides a discussion thereof. Finally, Section 6 offers a number of conclusions, implications for teaching practice and future research avenues to improve our understanding about how to exploit the educational potential of gamification in full. 2. Theoretical background 2.1. Gamification The latest research in education underscores how important acquiring essential practical skills and active learning are nowadays (Okolie et al., 2022; Mu˜ noz Miguel et al., 2023; Yesildag & Bostan, 2023). Prior works point out the usefulness of a number of innovative educational methodologies, such as collaborative learning (Okolie et al., 2022; Mu˜ noz Miguel et al., 2023), service learning (H´ ebert & Hauf, 2015), simulation-based experiential learning (Tiwari, Nafees & Krishna, 2014; Bakoush, 2022), and movie analysis (Yesildag & Bostan, 2023), to name but a few examples. In the field of management education, many studies advocate the need to endow students with experiential learning experiences so as to engage them in the world of real-life business decision-making (Tiwari et al., 2014; Yesildag & Bostan, 2023). Bakoush (2022) applies simulation learning using a stock market analysis project and argues that this practical method boosts student satisfaction. This in turn is found to discourage students from surface learning, and provides them with an advantage to succeed in deep learning strategies. Active teaching strategies geared towards endowing students with first-hand experience and with assigning them a leading role in their learning process seem to have become a platform through which to improve academic performance in the current educational context. One active learning methodology to have become widespread in recent years is gamification (Subhash & Cudney, 2018; Candan & Basaran, 2023). According to the theory of gamified learning (Landers, 2015; Landers & Landers, 2015), gamification applies game-based elements outside the game context in order to influence students’ behaviour and attitudes, which can result in enhanced 1 Clouder et al. (2019) define digital social capital as “the benefit derived from the individual or group’s social connections and networks based on their socialisation into the use of technology and the investment of time in developing technical knowledge and competence”. J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 3 learning outcomes, either directly or indirectly. Among all the gamification platforms used for educational purposes, Kahoot! is one of the most popular in higher education 2 (Candan & Basaran, 2023; Sevim-Cirak & Islim, 2023). Kahoot! is a game-based learning application which combines Student Response Systems (SRS) developed in the sixties (Judson & Sawada, 2002), game-based learning methods (Gee, 2003), social learning (Sarkar et al., 2017), and student familiarity with digital devices (Wang, 2015) in order to increase student engagement in classroom activities and efficiently implement formative assessment (Sharples, 2000). Such enhancing engagement strategies prove particularly useful for online learning, which is where students are more prone to boredom (Baker et al., 2010). Extensive research documents a number of benefits attached to the application of gamification in the learning process, such as the increased level of student motivation, attentiveness and participation (Dias, 2017; Ekici, 2021; G¨ oksün & Gürsoy, 2019; Qiao et al., 2022; Subhash & Cudney, 2018); stronger knowledge retention (Putz, Hofbauer & Treiblmaier, 2020); and superior academic performance (Dias, 2017; Ekici, 2021; Ortiz-Martínez et al., 2022) Nevertheless, it is worth acknowledging that recent works also alert to certain concerns which might limit the effectiveness of applying gamification to learning. For instance, Sanchez et al. (2020) report evidence of a ‘novelty effect’ in gamification, which leads to its benefits for academic performance weakening over time. One plausible explanation is that learners might perceive the gamified activity as less enjoyable and useful as a result of its repeated implementation. Greater use of gamification may exhaust student motivation and prove counterproductive for student performance (Sanchez et al., 2020). These weaknesses suggest that gamification is not per se a never-ending source of motivation for students and that, therefore, paying closer attention to maintaining student engagement throughout its use over time is by no means a trivial matter. 2.2. Digital social capital Other student motivation drivers that are external to the academic environment –such as peer pressure and social relationships– might prove key to maintaining their involvement in gamified learning over time. In this regard, interestingly, one stream of works shows that social networks can strengthen student motivation in the teaching-learning process (Ranieri et al., 2012; Hortigüela-Alcal´ a et al., 2019; Mishra, 2020). The role played by social networks has become core amongst the current generation of students –the so-called Generation Z– for whom much of their communication in daily life takes place through the Internet and social media (Biro, 2014). Indeed, social media are even more motivating for them because they are digital natives who have lived in an instant-reaction world that is rife with social media rewards (Gabrielova & Buchko, 2021). Social networks allow people to connect with others who display common interests or goals. Such networks may serve as an escape mechanism to recharge student motivation in another digital –albeit more informal– environment which may relieve them from the pressure of a more formal academic atmosphere. Relationships between individuals forged within social network sites constitute a type of asset known as digital social capital (Clouder et al., 2019; P´ erez-Hern´ andez et al., 2023). Therefore, social networks are widely believed to provide students and professionals with digital (or online) social capital, which has supported learning (Ranieri et al., 2012), particularly since the outbreak of the COVID-19 pandemic (Salimi et al., 2022). Our research is also particularly timely because we focus on the context of the COVID-19 pandemic, whose health and safety restrictions aggravated students’ feeling of social isolation and impaired their motivation and enthusiasm. Mishra (2020) emphasizes how useful social capital is in terms of improving academic performance, especially in the case of minority students, who find it more difficult to access and integrate into the higher education system. In another work, Salimi et al. (2022) find that digital social capital improves student academic performance indirectly, through the mediation of knowledge sharing in the online setting. Complementarily, one group of studies underscores the relevance of also taking into account the quality of online interaction, which is viewed as a key factor in the development of digital social capital (Zheng et al., 2020). Previous works have also reported evidence concerning the motivation-enhancing effects prompted by digital social capital in other contexts such as entrepreneurship. For instance, P´ erez-Hern´ andez et al. (2023) show that digital social capital has a positive effect on entrepreneurial intention. 2.3. Hypotheses Drawing on evidence from earlier literature, we expect stronger student engagement in social networks to help develop their digital social capital. Our starting hypothesis is that this digital social capital might foster student motivation and engagement in online gamification practices, given that the latter aligns better with the digital environment and with the visibility of public rankings. In turn, this greater participation in Kahoot! is likely to lead to better student academic performance in courses. Consequently, we test whether digital social capital developed through an involvement in social networks might have an indirect effect on student academic performance mediated by their Kahoot! participation. Fig. 1 graphically illustrates our hypothesized relationships. Based on the previous discussion, we propose two hypotheses: H1.The greater the student social capital developed through social networks, the greater their participation in Kahoot! games. H2.Student participation in Kahoot! mediates the relationship between student digital social capital and their academic performance. 2 See Wang and Tahir (2020) for a recent literature review about the effect of using Kahoot! on learning. J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 4 3. Research methodology 3.1. Study design and sample Our study relies on a survey-based approach. One major advantage of this quantitative strategy is that it provides direct, recent and rapidly available information about students’ profile (Biart & Praet, 1987). Moreover, it allows us to facilitate replication studies and to conduct statistical analyses of a representative sample of undergraduate students so as to generalize the results to larger populations (McClintock et al., 1979; Knoke et al., 2017). Our research is based on cross-sectional data in the academic year 2020/2021. We choose Kahoot! as the gamification platform because it is the most widely used in the higher education context, as shown by recent research (Candan & Basaran, 2023; Sevim-Cirak & Islim, 2023). Sample students took part in Kahoot! games during the classes, which were carried out in hybrid mode due to the restrictions imposed by the COVID-19 health crisis. We conducted a survey amongst students at a particular point in time; namely, at the end of the autumn semester in January 2021. In this survey, students were asked a number of questions about their degree of engagement in social networks (in which networks they have a profile, how often they use them, average number of contacts, etc.), in addition to a set of questions about personal details and background (e.g. whether they combine their studies with a paid job, their preference between online/hybrid teaching and onsite classes). This survey was released through the course virtual campus (Moodle). Survey data were extended by adding data about academic performance records in the courses analysed as well as students’ personal details taken from the university’s databases (SIGMA). SIGMA software is the educational platform for academic and research data management at higher education institutions. Our sample consists of 133 undergraduate students (58% women and 42% men) enrolled in several corporate finance courses at a public university in Spain. These students belong to three different bachelor degrees at the Faculty of Economics and Business Administration: the Degree in Business Administration (100 students), the Joint Degree in Law and Business Administration (20 students), and the Degree in Finance, Banking and Insurance (13 students). 3.2. Empirical strategy: variables, models and estimation method Our empirical strategy consists of two stages. First, we analyse whether students’ digital social capital affects their likelihood of participating in Kahoot! Second, we adopt a mediation approach in order to evaluate whether the impact of digital social capital on students’ academic performance is mediated by their participation in Kahoot! gamification activities. Table 1 summarizes our research variables. 3.2.1. Dependent variable The first-stage dependent variable is student participation in Kahoot! gamification during the hybrid classes, which is captured by Fig. 1. Illustration of the mediating model. Table 1 Study variables. Variable Definition Label Participation in Kahoot! gamification A binary variable equal to 1 if the student has participated in Kahoot! games, and zero otherwise. KH_PARTICIP Percentage of Kahoot! games each student has taken part in. KH_GAMES Percentage of Kahoot! class sessions each student has participated in. KH_SESSIONS Academic performance Student’s overall mark in the exams. EXAM_PERFORMANCE Student’s overall mark in the course (considering exams and continuous assessment activities). COURSE_PERFORMANCE Digital social capital The natural logarithm of the average number of follower contacts FOLLOWERS The natural logarithm of the average number of following contacts FOLLOWING Control variables Teaching mode preference A binary variable equal to 1 if the student prefers online or hybrid teaching, and zero otherwise. TEACHMODE Repeat students A binary variable equal to 1 if the student is a repeater, and zero otherwise. REPEATER Student gender A binary variable equal to 1 if the student is female, and zero otherwise. GENDER Working status A binary variable equal to 1 if the student has a paid job, and zero otherwise. JOB Dummy variables of the university degree. J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 5 three alternative proxies: KH_PARTICIP (a dummy variable equal to 1 if the student has participated in Kahoot! games, and zero otherwise); KH_GAMES (the percentage of Kahoot! games each student has taken part in), and KH_SESSIONS (the percentage of Kahoot! class sessions each student has been involved in). In the second stage of the analysis, the dependent variable is student academic performance, which is approximated by two measures: their overall mark in the exams (EXAM_PERFORMANCE) and their final mark in the course (COURSE_PERFORMANCE). The final mark in the course considers both exam performance and active participation. 3.2.2. Explanatory variables In the first stage of the analysis, the explanatory variable is student digital social capital. Similar to the research literature on entrepreneurial finance through digital platforms (e.g. crowdfunding) (Colombo et al., 2015; Buttic´ e et al., 2017), digital social capital is measured by the natural logarithm of the average number of follower contacts (FOLLOWERS) and the natural logarithm of the average number of following contacts (FOLLOWING). In the second set of analyses, student participation in Kahoot! gamification serves as an explanatory variable in the model (i.e. mediating variable). As described earlier, we rely on the same alternative proxies: KH_PARTICIP, KH_GAMES and KH_SESSIONS. 3.2.3. Control variables In all the estimations, we control for a number of factors which might also influence student performance in some way, both in terms of participation in Kahoot! and academic performance. Our set of control variables is made up of: teaching mode preference (the dummy TEACHMODE, which takes the value of 1 if the student prefers online or hybrid-teaching, and zero otherwise), repeat students (the dummy REPEATER, which equals 1 if the student is a repeater, and zero otherwise), student gender (the dummy GENDER, which equals 1 if the student is female, and zero otherwise), and working status (JOB, which equals 1 if the student has a paid job, and zero otherwise). Additionally, we include dummy variables to control for the university degree in which the student is enrolled. 3.2.4. Empirical models and estimation method In the first stage of the analysis, in order to examine the effect of digital social capital on student Kahoot! participation, we specify the following equation [1]: KAHOOTi=β0+β1•FOLLOWERSi+β2•FOLLOWINGi+β3•CONTROLSi+ ε i[1] where KAHOOT denotes the alternative proxies for student participation in Kahoot!, FOLLOWERS and FOLLOWING indicate the two dimensions of digital social capital measurement, subscript i represents each student, and ε i is the random disturbance. Since KH_PARTICIP is a binary variable, when we use it as the dependent variable to proxy Kahoot! participation, a probit regression is applied to estimate the model. However, when we draw on KH_GAMES and KH_SESSIONS as alternative dependent variables, we run Tobit estimations since these dependent variables are censored (Amore & Murtinu, 2021). We then assess the potential mediating role of Kahoot! participation in the relationship between student digital social capital and student performance. For this purpose, Baron and Kenny (1986) posited a mediation approach 3 based on the fulfilment of four conditions: [i] a direct effect, namely a significant relationship between the independent variable (digital social capital) and the dependent variable (academic performance); [ii] a significant relationship between the independent variable (digital social capital) and the mediating variable (Kahoot! participation); [iii] a significant association between the mediator and the dependent variable; and [iv] the effect of the independent variable weakening (partial mediation) or losing its statistical significance (full mediation) once the mediating variable is included. Equation [1] previously indicated tests for condition [ii]. The remaining conditions can be examined by these three additional equations [2] to [4]: PERFORMANCEi=γ0+γ1•FOLLOWERSi+γ2•FOLLOWINGi+γ3•CONTROLSi+ ε i[2] PERFORMANCEi=δ0+δ1•KAHOOTi+δ2•CONTROLSi+ ε i[3] PERFORMANCEi= α 0+ α 1•FOLLOWERSi+ α 2•FOLLOWINGi+ α 3•KAHOOTi+ α 4•CONTROLSi+ ε i[4] where PERFORMANCE denotes student academic performance, KAHOOT represents the alternative proxies for student participation in Kahoot!, FOLLOWERS and FOLLOWING indicate the two dimensions of digital social capital measurement, subscript i represents each student, and ε i is the random disturbance. Equations [2], [3] and [4] test for Baron and Kenny’s (1986) conditions [i], [iii] and [iv], respectively. Given that the dependent variable is continuous and censored, we again apply a Tobit estimation procedure. Finally, it is worth noting that later studies such as Zhao et al. (2010) alert to a possible misapplication of Baron and Kenny’s former perspective to identify mediating effects. They point out that it is not essential to require the existence of a significant relationship between the independent variable and the dependent variable, as established in condition (i). Mediation can still apply in this case in the form of indirect-only mediation, in which only the indirect effect displays statistical significance. 3 This econometric approach to test mediation has been widely applied by previous studies (Müller & Wulf, 2022; Wittmann & Wulf, 2023). J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 6 4. Results 4.1. Descriptive statistics Table 2 summarizes the main descriptive statistics of our sample. Almost 82% of sample students took part in Kahoot! games in the hybrid courses. On average, each student was involved as a participant in approximately 64% of Kahoot! games/sessions. As far as digital social capital proxies are concerned, slightly higher dispersion was seen in terms of the number of followed contacts compared to following contacts. Almost a quarter of the students had a paid job, which they combine with their university studies. Table 3 shows pairwise correlations. All the proxies for Kahoot! participation display strong correlation (above 0.80), which supports their use as alternative proxies for the same construct. Interestingly, these variables of Kahoot! engagement have a positive and statistically significant correlation with the variables of student academic performance (both EXAM_PERFORMANCE and COURSE_PERFORMANCE), with the correlation ranging between 0.30 and 0.41. Digital social capital variables present no statistically significant correlation with either Kahoot! participation variables or academic performance variables. The pairwise correlation between FOLLOWING and FOLLOWERS is about 0.33 in our sample, such that there are no concerns about potential collinearity problems of adding both of them simultaneously. 4.2. Regression estimates Table 4 displays the results of the first stage of our analyses, in which we evaluate the influence of student digital social capital on their Kahoot! engagement. Columns (1) and (2) report the results using KH_PARTICIP as the dependent variable. Since KH_PARTICIP is a binary variable, we report probit estimation results. Our evidence suggests that stronger digital social capital –as measured by FOLLOWING– has a positive and statistically significant impact (β =0.2895, p <0.10) on student willingness to participate in Kahoot! gamification. In contrast, no statistically significant effect is found for FOLLOWERS (β = − 0.1663, p >0.10). In the subsequent columns of this same table, we run additional robustness estimations by considering KH_GAMES and KH_SESSIONS as dependent variables, alternatively. Since these variables are censored, Tobit estimations are applied in columns (3) to (6). Results remain similar when applying these alternative proxies. For instance, if FOLLOWING increases by one percentage point, student participation in Kahoot! games (KH_GAMES) rises by 0.08 percentage points. As far as the second part of the study is concerned, we examine whether Kahoot! participation plays a mediating role in the relationship between student digital social capital and student academic performance. For this purpose, we evaluate Baron and Kenny’s (1986) remaining conditions, in addition to the condition [ii] already tested in the results described previously. To do this, we run Tobit regressions, since the alternative dependent variables –either EXAM_PERFORMANCE or COURSE_PERFORMANCE– are continuous censored variables. Table 5 reports these estimations. Column (1) considers only control variables in the estimation. Column (2) tests for condition [i], columns (3) to (5) test for condition [iii], and finally, columns (6) to (8) assess condition [iv]. Panel A in Table 5 displays the results based on EXAM_PERFORMANCE. As regards the control variables in Column (1), only GENDER and JOB display statistical significance. A student’s academic performance decreases by about 1.12–1.40 points if they have a paid job, which is consistent with the idea that work commitment reduces the time available to devote to studying. Column (2) additionally enters the digital social capital proxies. As shown, digital social capital from social networks carries no significant effect individually on students’ overall mark in the final course exam. In columns (3) to (5), we consider Kahoot! participation as the explanatory variable, measured by the three alternative proxies. We find that KH_PARTICIP has a positive effect on students’ academic performance, which is statistically significant at the 1% level (δ =1.4082, p <0.01). This result strongly supports Hypothesis 1. This evidence remains robust to the use of the two alternative proxies; namely, KH_GAMES (δ =1.8870, p <0.01) and KH_SESSIONS (δ = 1.8281, p <0.01). Finally, columns (6) to (8) enter the digital social capital proxies and Kahoot! participation simultaneously. The lack of statistical Table 2 Summary statistics. No. of Obs. Mean Median Std. Dev. Min. 25th perc. 75th perc. Max. Kahoot participation KH_PARTICIP 133 0.8195 1 0.3860 0 1 1 1 KH_GAMES 133 0.6372 0.7308 0.3770 0 0.4091 1 1 KH_SESSIONS 133 0.6460 0.7500 0.3765 0 0.4286 1 1 Academic performance EXAM_PERFORMANCE 133 4.1218 4.1100 2.0785 0 4.1100 5.6200 8.5333 COURSE_PERFORMANCE 133 5.0060 5.3000 2.6449 0 3.1000 7 10 Digital social capital FOLLOWING 122 6.2813 6.3808 0.8444 0.3365 6.0426 6.8024 7.6009 FOLLOWERS 122 6.4297 6.4068 0.9989 4.0943 5.9914 6.8957 14.247 Control variables TEACHMODE 133 0.4060 0 0.4929 0 0 1 1 REPEATER 133 0.2932 0 0.4570 0 0 1 1 GENDER 133 0.5789 1 0.4955 0 0 1 1 JOB 133 0.2406 0 0.4290 0 0 0 1 J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 7 Table 3 Pairwise correlation matrix. 1 2 3 4 5 6 7 8 9 10 11 1. KH_PARTICIP 1.000 2. KH_GAMES 0.796*** 1.000 3. KH_SESSIONS 0.808*** 0.994*** 1.000 4. EXAM_PERFORMANCE 0.304*** 0.374*** 0.362*** 1.000 5. COURSE_PERFORMANCE 0.346*** 0.418*** 0.406*** 0.981*** 1.000 6. FOLLOWING 0.119 0.086 0.071 0.057 0.057 1.000 7. FOLLOWERS −0.120 −0.086 −0.107 −0.136 −0.128 0.328*** 1.000 8. TEACHMODE −0.209** −0.205** −0.195** −0.070 −0.070 0.057 −0.049 1.000 9. REPEATER −0.299*** −0.336*** −0.350*** −0.035 −0.065 0.058 0.099 0.107 1.000 10. GENDER −0.004 0.039 0.028 0.211** 0.223*** 0.131 0.076 −0.008 0.081 1.000 11. JOB −0.239*** −0.245*** −0.245** −0.185** −0.185** 0.083 0.187** 0.072 0.217** 0.052 1.000 ***, ***, and * denote statistical significance at the 1%, 5%, and 10% level, respectively. J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 8 Table 4 Kahoot! participation and digital social capital. PANEL A: Probit regressions PANEL B: Tobit regressions Dependent variable: KH_PARTICIP Dependent variable: KH_GAMES Dependent variable: KH_SESSIONS (1) (2) (3) (4) (5) (6) Constant 1.2987*** (0.2843) 0.5246 (1.1818) 0.9054*** (0.1140) 0.7018** (0.3342) 0.9265*** (0.1136) 0.7999** (0.3333) FOLLOWERS −0.1663 (0.1520) −0.0644 (0.0448) −0.0735 (0.0450) FOLLOWING 0.2895* (0.1581) 0.0828* (0.0461) 0.0798* (0.0459) Control variables TEACHMODE −0.1258 (0.3032) −0.1526 (0.3283) −0.0834 (0.0842) −0.1078 (0.0849) −0.0708 (0.0839) −0.0945 (0.0845) REPEATER −0.6695** (0.3054) −0.7792** (0.3305) −0.2611*** (0.0840) −0.2752*** (0.0854) −0.2711*** (0.0837) −0.2884*** (0.0849) GENDER −0.0916 (0.3001) 0.1032 (0.3230) 0.0222 (0.0758) 0.0906 (0.0780) 0.0125 (0.0755) 0.0816 (0.0776) JOB −0.7629** (0.3331) −0.5389 (0.3700) −0.2283** (0.0898) −0.1452 (0.0912) −0.2265** (0.0894) −0.1410 (0.0907) University degree dummies Yes Yes Yes Yes Yes Yes No. of obs. 100 91 133 122 133 122 Log likelihood −47.7128 −41.2615 −82.7580 −71.3002 −82.5377 −71.0089 P-value Chi2 0.0052 0.0382 0.0001 0.0006 0.0000 0.0005 Standard errors are reported in parentheses under the estimated coefficients. ***, ***, and * denote statistical significance at the 1%, 5%, and 10% level, respectively. J.M. Fortuna et al. The International Journal of Management Education 21 (2023) 100846 9 Table 5 Academic performance and digital social capital: the mediating role of Kahoot! participation. PANEL A Dependent variable: EXAM_PERFORMANCE (1) (2) (3) (4) (5) (6) (7) (8) Constant 5.9276*** (0.5221) 6.2347*** (1.5444) 4.3558*** (0.7133) 4.2665*** (0.6370) 4.2803*** (0.6410) 5.0729*** (1.5709) 5.0346*** (1.5098) 4.9306*** (1.5280) KH_PARTICIP 1.4082*** (0.4526) 1.2935** (0.4983) KH_GAMES 1.8870*** (0.4511) 1.7456*** (0.4909) KH_SESSIONS 1.8281*** (0.4554) 1.6781*** (0.4961) FOLLOWERS −0.2813 (0.1893) −0.1988 (0.1837) −0.2014 (0.1785) −0.1913 (0.1799) FOLLOWING 0.1818 (0.2160) 0.0759 (0.2145) 0.0750 (0.2080) 0.0844 (0.2088) Control variables TEACHMODE 0.3139 (0.3801) 0.1577 (0.4017) 0.3799 (0.3681) 0.4429 (0.3589) 0.4168 (0.3601) 0.2204 (0.3925) 0.3153 (0.3853) 0.2872 (0.3865) REPEATER 0.3467 (0.3754) 0.3688 (0.3992) 0.5965 (0.3717) 0.7514** (0.3661) 0.7572** (0.3692) 0.6148 (0.4007) 0.7677* (0.3966) 0.7746* (0.4006) GENDER 0.6398* (0.3426) 0.7476** (0.3673) 0.6775** (0.3314) 0.5936* (0.3224) 0.6123* (0.3239) 0.7121** (0.3584) 0.6181* (0.3519) 0.6386* (0.3531) JOB −1.4007*** (0.3994) −1.1169*** (0.4293) −1.1380*** (0.3952) −1.0650*** (0.3841) −1.0773*** (0.3859) −0.9667** (0.4226) −0.9103** (0.4131) −0.9243** (0.4148) University degree dummies Yes Yes Yes Yes Yes Yes Yes Yes No. of obs. 133 122 133 133 133 122 122 122 Log likelihood −269.0522 −247.4126 −264.3597 −260.8009 −261.4191 −244.1165 −241.3705 −241.9168 P-value Chi2 0.0000 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 PANEL B Dependent variable: COURSE_PERFORMANCE (1) (2) (3) (4) (5) (6) (7) (8) Constant 7.1246*** (0.6719) 7.4712*** (1.9867) 4.8236*** (0.9080) 4.7446*** (0.7984) 4.7631*** (0.8125) 5.7393*** (2.0024) 5.7224*** (1.9136) 5.5662*** (1.9390) KH_PARTICIP 2.0609*** (0.5762) 1.9252*** (0.6352) KH_GAMES 2.7027*** (0.5710) 2.5419*** (0.6220) KH_SESSIONS 2.6200*** (0.5772) 2.4499*** (0.6293) FOLLOWERS −0.3269 (0.2393) −0.2160 (0.2341) −0.2221 (0.2262) −0.2071 (0.2281) FOLLOWING 0.2231 (0.2778) 0.0657 (0.2733) 0.0677 (0.2636) 0.0809 (0.2648) Control variables TEACHMODE 0.4705 (0.4889) 0.2766 (0.5165) 0.5674 (0.4685) 0.6553 (0.4543) 0.6179 (0.4565) 0.3702 (0.5001) 0.5061 (0.4882) 0.4657 (0.4903) REPEATER 0.2111 (0.4829) 0.2566 (0.5134) 0.5764 (0.4730) 0.7907* (0.4633) 0.7993* (0.4679) 0.6224 (0.5105) 0.8373 (0.5024) 0.8488* (0.5080) GENDER 0.8716** (0.4406) 1.0176** (0.4723) 0.9268** (0.4218) 0.8053** (0.4081) 0.8320** (0.4104) 0.9646** (0.4567) 0.8289** (0.4458) 0.8584* (0.4478) JOB −1.7026*** (0.5139) −1.3601** (0.5521) −1.3185*** (0.5030) −1.2217** (0.4861) −1.2391** (0.4891) −1.1369** (0.5385) −1.0591** (0.5234) −1.0789** (0.5262) University degree dummies Yes Yes Yes Yes Yes Yes Yes Yes No. of obs. 133 122 133 133 133 122 122 122 Log likelihood −301.5871 −277.1362 −295.4493 −291.1909 −291.9691 −272.6809 −269.2719 −269.9545 P-value Chi2 0.0000 0.0004 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Standard errors are reported in parentheses under the estimated coefficients. ***, ***, and * denote statistical significance at the 1%, 5%, and 10% level, respectively. J.M. Fortuna et al.