Participants were recruited at a major university in Asia. An invitation email was sent to all undergraduate and postgraduate students’ (N=22,260) university email accounts inviting them to participate in the current study. Students who were between 17-25 years old, had adequate Internet literacy and a Facebook account, and were reachable via the local network were eligible to participate. The invitation email explained to participants that they would not only be required to complete a 3-week game on Facebook, but would also be asked to complete a set of Web-based, self-assessed questionnaires both before and after finishing the game. They were allowed to play the game anytime and anywhere with a computer connected to the Internet within the study period. Those who completed the game and the online questionnaires received cash compensation and were entered in a drawing for a tablet computer, supermarket coupons, and theme park tickets. Interested students replied to the invitation email and provided their Facebook ID for verification. The research team then confirmed their eligibility through email.

Descriptive statistics were used to illustrate the general picture of the data for each measurement. Any participants with missing data or who dropped out were not included in the completers analysis. In addition to completers analysis, data were analyzed through an intent-to-treat (ITT) analysis, using multiple imputation (MI) to address loss of follow-up data. Dropout participants who completed questionnaires only at the pretest were handled by applying the technique of MI with 5 imputations for missing data [42]. We conducted missing values analysis to investigate whether data were missing completely at random (MCAR). A within-subject Student’s t-test was used to analyze the statistical significance of changes in knowledge, and a Student’s t-test for independent samples was employed to test for gender differences.

Linear regression was used to examine the relation between knowledge enhancement and motivational constructs. By fitting knowledge enhancement with motivational constructs, the coefficient and 95% confidence interval could be presented. The regression was further adjusted by adding pretest score, gender, and age as control variables. Backward selection was employed in the regression analysis. All data analyses were conducted using SPSS software.

Ethics approval, from the Human Research Ethics Committee for Non-Clinical Faculties at the authors’ institution, was obtained before data collection. Participants provided informed online consent before they started the game. The informed consent form detailed information about the procedures of the study.

Participants were recruited from November to December 2011 at the authors’ institution. A total of 221 undergraduates agreed to participate in the study. Out of those who started the game (n=136), a majority completed the pretest (n=127). The other 9 participants did not fully complete the pretest due to technical problems, such as early termination of the questionnaire. Out of these 127 participants, 73 completed both the pre- and posttest. All 127 participants were included in the ITT analysis, whereas only data from the 73 participants who completed both assessments were used for the completers and regression analyses. Figure 4 shows the participant flow. A total of 31 males and 42 females, ranging from 17-25 years old (M=20.82, SD 1.81), completed both assessments.

Table 1 shows the number of correct responses participants achieved on the mental health literacy assessment both before and after playing the game, as well as their overall improvement levels. Both male and female participants achieved over half correct at pretest, suggesting participants’ possessed good background knowledge on mental health. The pretest scores of males and females were similar, indicating that their familiarity with mental health was comparable. The improvement of both genders was also statistically significant and consistent, revealing the effectiveness of the game play intervention for both males and females. In a t-test comparison between the results of the pre- and posttest, a statistically significant improvement was documented. On average, participants answered 2.21 more questions correctly after the intervention (total of 31 questions on the knowledge test). Participants demonstrated moderate improvements between pre- and posttests in terms of mental health literacy (d=0.65). Furthermore, there was not a statistically significant gender difference in knowledge improvement (P=.97). Apart from completers analysis, noncompleters were also included in the ITT analysis to provide more understanding about the effects of the intervention. Missing values analysis demonstrated that the hypothesis that the data were MCAR could not be rejected, χ ² _Little=3.75, (P=.15). ITT analysis showed that the improvement between pre- and posttest was statistically significant, with a moderate effect size (d=0.66). This result was consistent with the completers analysis (d=0.65).

Table 2 presents the descriptive statistics of the motivational subscales. Value focuses on the reasons why participants engaged in the educational game. Among the three subscales that made up the construct (intrinsic and extrinsic goal orientation and task value), both genders achieved the highest score on intrinsic goal orientation. On the other hand, extrinsic goal orientation was the lowest among the three value subscales. Expectancy refers to participants’ beliefs that they could accomplish tasks in the game. Satisfactory scores on both subscales of the expectancy construct demonstrated participants’ confidence in the acquisition of mental health knowledge. Affect concerns test anxiety associated with the knowledge test. It is reasonable that the anxiety scores were low in a gaming context since learning in a game should be enjoyable and relaxing. Moreover, since the test in this study was conducted online and without a time limit, test anxiety was likely alleviated. Indeed, participants’ scores on the anxiety measure were the lowest among the 6 motivational subscales. T-tests revealed no statistically significant gender differences on any of the 6 motivational subscales (Table 2).

Nonsignificant motivational subscales were eliminated in the regression analysis with backward selection. Nonsignificance of the three subscales within the value construct indicated that reasons for participating in the game did not affect knowledge enhancement. The influence of control of learning beliefs in the expectancy construct was also not statistically significant. Control variables, namely pretest score, gender, and age, were retained in the model regardless of their level of statistical significance. As expected, gender and age did not affect learning outcomes, whereas pretest score did significantly affect knowledge enhancement (see Table 3). This result is reasonable given that pretest score restricted the maximum possible improvement on the posttest. In other words, participants attaining high scores on the pretest, when compared to those with lower scores on the pretest, could not improve as much on the posttest. Consequently, pretest score negatively influenced knowledge improvement. Furthermore, self-efficacy for learning and performance significantly affected knowledge improvement. Test anxiety, on the other hand, negatively affected knowledge enhancement.

To our knowledge, this is one of the first studies to explore the use of DGBL through an SNS to enhance mental health literacy. Despite the limitations of this exploratory study, participants attained mental health knowledge in a continual and self-paced manner. With moderate improvements among participants, it seems players may be able to acquire knowledge in the game at their own learning pace and apply the skills to solve real-life problems. This finding provides supporting evidence that Web-based DGBL can effectively enhance mental health knowledge. The combination of gaming concepts and online education may also facilitate self-paced learning processes in higher-order thinking. It seems that this new type of learning initiative for mental health information is efficient for both male and female learners.

Thus, the current study reinforces Lieberman’s suggestion [28] that games in health education are beneficial in various aspects, including the support of a self-paced learning process. Furthermore, although the learning content of the electronic game was modified from a school-based mental health enhancement program for adolescents [29], it demonstrated its effectiveness in enhancing the mental health literacy of young people aged 17-25. Therefore, we may extend the findings to a slightly older age group on Internet-based intervention on health education. Previous literature suggested that Internet-based intervention on health education was more effective on adolescents than on adults, as such intervention for adolescents demonstrated small-to-moderate effect size, whereas similar interventions for adults usually yielded small effect size only [9]. With moderate effect size found in the current study, it is suggested that social and gaming features may enhance the effectiveness of Internet-based intervention on health education for young adults.

Participants’ learning motivation was generally positive. They believed in their own abilities to learn in the game and tailored their learning processes in the absence of a facilitator. Although participants who were outstanding in the knowledge test were rewarded with an extra prize, intrinsic goal orientation was still stronger when compared to extrinsic orientation. High intrinsic motivation indicates that participants primarily perceived that they took part in the game for its challenges, out of curiosity, and in an effort to attain mastery. Therefore, participants enjoyed playing the game and learning about mental health knowledge throughout the intervention, instead of simply pursuing rewards. With high expectancy, participants also believed that their efforts to learn throughout the self-paced game would result in positive learning outcomes. Self-directed learning was possible in the game. The positive influence of self-efficacy for learning and performance on learning outcomes implies participants’ expectations about their own performance and judgment generally reflected their learning outcomes. Participants were capable of enhancing their mental health knowledge and manage the learning process in the game. Moreover, the game-based environment provided an enjoyable learning environment for participants, evidenced by substantially reductions in their stress over the knowledge test. However, participants’ worries and concerns about the test could have still resulted in performance decrements. Although the knowledge test was conducted in a casual way, relationships between test anxiety and test results still existed.

In addition, the feedback collected from participants on learning motivation and game design provided additional information about the high intrinsic goal orientation. The feedback from participants was generally positive and encouraging. Participants appreciated the combination of gaming concepts and mental health knowledge. The transformation of mental health knowledge into a game made mental health knowledge easy to understand and attractive (eg, “The game helped me understand more about stress management”; “Overall the game was quite interesting as it used animations to show what stress and emotional distress are and solutions to manage them”). They also found that the game was good to raise their awareness in mental health (eg, “Informative about mental disorder”; “A good way to promote awareness in mental health”), and the game itself was interesting and interactive (eg, “Interactive and easy to play”). This may explain why intrinsic goal orientation remained high in the learning process. Some participants, however, pointed out the game sometimes had too many words, which may decrease the enjoyment of the game (eg, “More informative than interesting”; “Sometimes want to skip information and conversations to advance plots”). We therefore may need to think of balancing information and entertainment to achieve better learning motivation in our future intervention development.

There are some limitations of this exploratory study: lack of a control group, small sample size, high dropout rate, and a biased sample. Moreover, the social functions in the game might not have been fully utilized, as the total number of players was not large. Time limitations prevented the number of players in the game to grow to an adequate size. Since the number of players influences the interactivity of a social game, the game in this study might not have been as interactive as other social games for pure entertainment on the market. In addition, our DGBL study was domain-specific, a common issue in most previous studies.

In sum, DGBL in combination with an online social communication platform in health education should be advocated as a way to promote mental health awareness and equip people with domain-specific knowledge education, and warrants further examination. Future studies should involve a larger sample, incorporate a control group, and recruit different age groups in order to have a more comprehensive understanding of the effects of this Web-based electronic game on mental health education. In addition to augmenting the sample size, another future research direction includes investigating the effectiveness of social game education based on this study. Social games on SNS facilitate social interaction and communication, making them different from traditional online games. SNS could provide tools to embed communication platforms (eg, blogs, forums) into games, which may, in turn, substantially enhance levels of sharing and knowledge exchange among players [43]. These social elements can motivate players to learn and potentially foster the learning process. DGBL should not be restricted to learning inside the game but should extend outside the realm of the game (eg, discussing learning material on blogs or forums). The effectiveness of social game education needs further evaluation. Also, the scope of this research should be extended to different health issues, such as mood and eating disorders. Through such future work, DGBL can be fully utilized and evaluated in different aspects of health education.