Using Role Models and Game-Based Learning to Attract Adolescent Girls to STEM

Abstract

Various pedagogical approaches have been proposed to attract more female students to STEM (science, technology, engineering, and mathematics), targeting student beliefs and perceptions that are linked to STEM study intentions. The current study, which took place in a secondary school in Cyprus and employed a quasi-experimental design, aims at contributing to this literature. Responses to pre- and post-questionnaires by 69 experimental and 27 control students show that students in the experimental group, who participated in an intervention in which they learned about the lives and accomplishments of STEM/STEAM role models via a game-based learning approach, improved their STEM ability perceptions compared to control group students, who reported similar pre-post survey levels of STEM ability perceptions. Female students benefited more compared to their male counterparts, and using a game-based learning approach contributed significantly to the effectiveness of students’ exposure to the role models. Taken together, study findings support the use of role models and learning games as tools to attract more female students to STEM.

1. Introduction

Research in the EU and other parts of the world [1,2] shows that adolescent female students are less likely than males to pursue studies and careers in STEM (science, technology, engineering, and mathematics) fields, which have more and better-paying employment opportunities. Women’s underrepresentation in STEM, which perpetuates economic gender inequalities and limits the diversity of the STEM workforce, cannot be attributed to differences in cognitive abilities and academic achievement [3]. According to recent TIMSS and OECD reports [4,5], the gender gap in science and mathematics has been narrowing, and in many countries, adolescent girls have caught up with and even outperform adolescent boys. However, based on a recent OECD report [2], on average across OECD countries, only 14% of the girls who were top performers in science or mathematics reported that they expected to work as professionals who use science and engineering training, as opposed to more than 26% of top-performing boys. In other words, female students who perform well in science and mathematics are highly unlikely to pursue studies in STEM fields.

Research has established that students’ career-related ambitions and academic choices are not dependent solely on academic achievement but are shaped by complex psychological processes which, in male-typed domains such as science, engineering, and computing, are influenced by stereotypes and gendered social-role expectations, leading female and male students to different academic and career paths [3,6]. Based on this research, scholars and educators have used a variety of pedagogical approaches to attract more females to STEM, targeting student motivational beliefs and other perceptions that are linked to STEM study intentions [7,8]. The present research, which involves an intervention that uses role models and game-based learning with secondary school students, aims to contribute to this literature.

1.1. Ability Perceptions, Gender Stereotypes, and Role Models in STEM

Students’ perceived competence in relevant academic domains, which has been conceptualized as expectancy beliefs [9], academic self-concept [10], or self-efficacy [11], is an important predictor of their academic and career choices. Students are more likely to choose a particular school subject or field of study in college if they think that they are competent and can succeed at it. Although ability perceptions are enhanced by academic accomplishments, they may not necessarily reflect students’ actual competencies because they are influenced by students’ own interpretations of their experiences, which are often based on gender stereotypes (e.g., “girls are not good at math”; “computing is not for girls”) [9,12,13]. For example, several studies have shown that, as a group, adolescent girls tend to underestimate their performance and to express lower confidence in their math and computer abilities even when they perform equally well as adolescent boys [12,13].

Gender stereotypes are expectations of the characteristics of men and women, which are attributed to biological differences [14]. Women are considered better suited for low-status roles that involve caring for others and for academic fields that are believed to require empathy and hard work, such as nursing and education [15,16]. They are also expected to place more priority on their family than on their professional career, while the opposite is expected from men, who are considered suited for high-status roles [17]. STEM careers are regarded as more appropriate for men because they are supposed to require agentic traits as well as abstract and logical thinking, which are considered “masculine” qualities [15,16]. Although research has challenged these gender stereotypes [17], they are still abundant in mass and digital media [18] and can be found even in educational textbooks [19,20]. Their pervasiveness, combined with the paucity of STEM female scholars and professionals in educational materials, communicates the implicit message that women have not made significant contributions to these fields [21].

Several studies, e.g., [22,23], have demonstrated that gender stereotypes create psychological obstacles for women that keep them out of STEM. Female students who internalize gender stereotypes about STEM may underestimate their ability to succeed in STEM fields or decide that STEM fields are not appropriate for them. Research has identified links between women’s endorsement of gender stereotypes and their STEM self-concept, career expectations, and actual career attainment [22,23,24]. In a longitudinal study, Dicke et al. [23] found that women who had endorsed traditional gender-role views in adolescence acquired lower levels of education in adulthood and were less likely to have occupations within male-typed STEM domains at age 42 compared with women who did not endorse such stereotypes when they were adolescent students.

Exposing students to successful female scientists and STEM professionals who are perceived as role models is considered an effective approach to attracting more female students to STEM [7]. According to social cognitive theory [11], observing competence in individuals with which learners identify can increase learners’ self-efficacy because they may think that they can also be successful. Also, based on social role theory [14], young people’s perceptions of male and female characteristics and gender-appropriate behaviors are based on their everyday observations of men’s and women’s activities. Observing role models that have counter-stereotypical characteristics, such as successful female scientists, may challenge gender stereotypes and encourage girls to think that they can also succeed in STEM and that STEM careers are appropriate for women [25].

1.2. Using STEM Role Models with Secondary School Students

Although research on STEM role models has tended to focus on university students [26,27], several studies have addressed adolescent students. In most of these studies, role model interventions had a positive impact on variables that are important predictors of girls’ and boys’ future academic and career choices. These variables include students’ academic performance, e.g., [28,29], ability perceptions, e.g., [30,31], and interest in one or more STEM subjects or future careers in STEM, e.g., [32,33], as well as science identity, e.g., [34], and perceptions of female scientists’ personality characteristics and abilities, e.g., [30,35]. Although most studies focused on female students and used female role models, there is evidence that female role models can have a positive impact on boys as well, e.g., [36]. However, based on a recent systematic review [7], this finding is not consistent, as there are some studies showing that role models from underrepresented groups in STEM (e.g., female or non-white scientists) had a positive effect on ingroup students (e.g., girls) but no effect on the students belonging to the majority-group (e.g., boys, white students). In addition, role models are more likely to be effective when they represent an incremental instead of an entity view of ability, communicating to school students that success can be achieved with hard work and is not dependent on talent, which is commonly perceived as a fixed characteristic, e.g., [28,36,37].

Only in some studies did students have the opportunity to meet real STEM researchers, university students, or professionals [30,32,34,35]. In many successful interventions, role models were fictional personalities, and students learned about them via short stories or interviews [31,36,37] or by interacting with computer-based interface agents [29]. It appears, therefore, that even short interventions that do not involve face-to-face interactions with STEM role models can have a positive influence on variables that determine students’ future study and career choices.

Overall, the above findings show that STEM role models can benefit female and, also, male secondary school students. There are, however, certain gaps in this research. Based on a recent relevant review [26], 81% of studies on STEM role models have been carried out in the U.S. This highlights the need to investigate how adolescents raised in other sociocultural contexts respond to role model interventions and which role models are most effective for them. In addition, interventions that did not involve student interactions with real STEM professionals and scholars relied on text, video, and audio presentations of role models [31,33,36,37]. However, technology can support several alternative approaches that may be more appealing to school students and engage them more actively in learning, such as games and computer-based interface agents [29].

1.3. Digital Game-Based Learning and STEM Role Models

Digital game-based learning (DGBL) is an educational approach that integrates digital games, either commercial (e.g., SimCity) or learning games, into the educational process to complement or transform conventional teaching and assessment [38,39]. Learning games, also known as educational games, have the characteristics of entertainment games, such as challenge, competition, rules, and fantasy, but have been designed to support specific learning goals [40]. When a game is selected for the classroom, it should be embedded in lessons or educational scenarios and used along with learning activities that support students to explore relevant topics and to connect what was learned in the game world to the real world and to curriculum concepts [39,41].

There are several arguments in favor of DGBL, particularly for serious games that are complex and can support higher-order learning [42]. Most of these arguments point to the impact that games may have on the quality of learning by enhancing student motivation [42,43,44]; motivated learners invest more time and effort, do not quit difficult tasks, and use better learning, reasoning, and problem-solving strategies. An important characteristic of digital games is that they allow learners to fail without having to suffer the actual consequences of failure, which protects their self-efficacy, encourages exploration, and supports the development of self-regulation [44,45].

Recent systematic reviews and meta-analyses that focused on STEM subjects, e.g., [38,46,47], have shown that DGBL is associated with increased motivation and learning of STEM content. Based on these reviews, however, relatively few studies explored the potential of games to improve student perceptions and attitudes regarding other STEM-related issues, such as gender inequalities and stereotypes [8]. Barrera Yanez et al. [48] reviewed digital resources, including games, designed to educate about gender equality, but nearly all of them addressed issues of gender violence and abuse. Also, in most studies where games were used to attract female adolescents to STEM, students participated in game-construction activities rather than playing a digital game [8]. We identified two studies involving role models and gameplay or gamification activities. Ertl [49] used an instructional learning scenario enriched with gamification elements (not an actual game), the story of which was based on a fictional female CEO from Silicon Valley who wanted to recruit a team of females for a new company branch. Ertl [49] found that after participating in the game challenge, which required the design of a new gaming app, female students reported higher soft skills and ability levels. Also, Kapp et al. [50], who evaluated “Serena Supergreen”, a digital game in which the main character was a girl who solved technical quests, found that playing the game improved both boys’ and girls’ perceived technical competence. The results of the above two studies are very promising, but more research is needed on using DGBL as an approach to expose adolescent students to STEM role models.

1.4. Context and Purpose of the Present Study

The study took place in the context of an EU-funded project that aimed, using role models and DGBL, at fighting stereotypes and inspiring adolescent students, particularly females, to pursue studies and careers in STEM/STEAM (STEM + the arts) fields. An important intellectual output of this project was FemSTEAM Mysteries, a learning game that introduces students to important female and male scientists and artists, who are expected to act as role models. After the game was developed, teachers in three secondary schools, one from each partner country (Cyprus, Greece, and Spain), working in multidisciplinary teams, designed and implemented STEAM education scenarios that integrated the game. In this article, we report findings from a quasi-experimental study that was carried out in the partner school located in Cyprus, where pre- and post-survey data were collected to examine changes in student perceptions as a result of their participation in the role-model, game-based learning scenarios. The study contributes to the literature because, as highlighted in the previous sections, research carried out outside the US is limited, and there is a dearth of studies that have exposed students to role models using interactive technologies such as game-based learning. We investigated the following questions:

• Can participation in the STEAM role-model, game-based learning scenarios influence secondary school students’ STEM-related ability perceptions, study intentions, and gender stereotypes?
• Are any changes in students’ STEM-related ability perceptions, study intentions, and gender stereotypes related to students’ evaluations of the game and the role models?

2. Materials and Methods

2.1. The Design of the FemSTEAM Mysteries Game

FemSTEAM Mysteries is a digital escape room game. In digital escape rooms, which are similar to physical escape rooms, players have to solve puzzles (i.e., problems, challenges, or activities) in a limited amount of time so as to achieve a goal that is embedded in a story [51]. FemSTEAM Mysteries includes eight escape rooms, each dedicated to a contemporary personality with a successful career in a STEM/STEAM field and active involvement in the promotion of gender equality. To influence female students’ perceived similarity with the role models so as to increase their effectiveness [7], two role models from each project partner country were selected, and six of them were females.

Game players are invited to find the identity of eight famous professionals and scholars because a great magician has managed to erase their memory during a big STEAM conference. To do so, players must solve the mysteries of eight escape rooms that were used by these eight role models before their memory accident. After entering a room, players can move around (e.g., look at something from different angles, zoom in on an object) and interact with objects (e.g., email messages, photos, invitations, cabinets) to collect clues and biographical information about the personality staying in that room. For a detailed description, see Vekiri et al. [52]. Figure 1 presents a snapshot of the game, which was developed by Challedu (Greece) with contributions from all FemSTEAM Mysteries project partners.

2.2. The STEAM Education Scenarios

STEAM education is a new pedagogical approach that combines the arts (the visual, performance, and other expressive arts as well as other non-STEM disciplines, including the liberal arts, humanities, and social sciences) with STEM subjects and encourages a holistic, trans-disciplinary investigation of authentic problems [53]. This approach is expected to help students develop a diverse set of transversal skills that are needed in contemporary societies [53] and is considered effective for attracting students, particularly females, to STEM because it may help them appreciate the contributions of STEM to the solution of complex real-world problems [54].

At a secondary school in Cyprus, nine teachers collaboratively developed five STEAM scenarios, each requiring five to seven class periods for implementation, with those lessons being spread across a single week for each scenario. Students in grades seven and eight participated in two scenarios, while students in grade nine worked on one scenario. The scenarios enabled students to explore a topic using concepts and skills from various STEAM fields. Hence, they were not limited to science and technology but contained many elements of art, graphic design, advertisement copywriting, and creative writing. Also, in each scenario, students learned about the life and accomplishments of a specific female personality whose work was relevant to the topic. For example, in the scenario entitled “Mapping the school”, year seven students had to create an app for school navigation and learned about American mathematician Gladys West and her contributions to the mathematical modeling of Earth’s shape, and in the scenario entitled “Holding on to your passion in life”, students investigated the challenges faced by Australian television presenter Stephanie Bendixsen as she took a role on a TV show about gaming, a male-stereotyped computing activity. Students in grade eight learned about DNA and the work of Rosalind Franklin in discovering its structure, along with the invaluable contribution of Margaret Hamilton to the Apollo 11 moon landing through her meticulous approach to software design. Students in the ninth grade investigated the impact of advertisements in the early days of the home computer revolution on the proportion of girls and boys who took up programming and game development, along with learning about some female pioneers in the earliest days of the video games industry.

Each scenario included learning activities requiring students working in pairs in the classroom to explore at least one of the FemSTEAM Mysteries role models during the time that scenario was being implemented. Students were allowed to explore freely in the game without prior instruction from the teachers so as to not bias their interactions or introduce preconceptions regarding the experience and outcomes. The game contained sufficient instructions and background information to allow for this without introducing any barriers or hindrances in the playing experience.

Scenario lessons were taught with as much continuity across classrooms and teaching staff as possible, with all scenarios being taught by the same teachers in Computing and Art, though timetabling constraints made it necessary for Mathematics, Science, and English classes to span several teachers in those departments. While the title of the project and its aims were known to the students, care was taken to present the project as a means by which all interested students, regardless of gender, can be encouraged to investigate a STEAM career, and the scenarios were designed and presented to be balanced and credible so as to avoid artificially estranging the male participants.

2.3. Study Participants

The study employed a pre-post-questionnaire quasi-experimental design. Participants were secondary school students (13- to 16-year-olds) attending the same school. The STEAM education scenarios were offered only to students in grades seven through nine. However, we also recruited students from grade 10 to serve as the control group. These students attended their regular school program, meaning that they did not have any involvement in the project learning activities and did not play the game. A total number of 123 students participated in the STEAM scenarios and the gameplay activities, and 32 students were enrolled in grade 10. Students responded anonymously to the online questionnaires, but they used a personal code that was assigned to them so that pre- and post-questionnaire cases could be matched later. Due to technical problems and practical difficulties, the pre- and post-questionnaires were answered by a smaller number of students, and we obtained 108 and 122 valid questionnaires, respectively. Complete data, including both pre- and post-survey responses, were obtained for 96 (69 experimental and 27 control) students. The experimental group included 34 boys and 34 girls, while the control group included 14 female and 11 male students. Three students chose not to provide gender information.

2.4. Pre- and Post-Questionnaires

Both questionnaires included 15 Likert items (1 = completely disagree to 5 = completely agree) addressing students’ (a) perceptions of ability in STEM fields (e.g., “I am confident in my ability in math”), (b) study and career interest in STEM fields (e.g., “I am interested in careers that use science”), and (c) gender stereotypes about females’ ability in STEM fields and about the suitability of STEM careers for women (e.g., “Girls are more inclined and better suited for the arts and humanities and boys for science and technology”). For the development of these items, we relied on previous scales assessing student confidence and interest in STEM fields, i.e., [55,56], and gender stereotypes about STEM fields, i.e., [22,57,58].

The post-questionnaire for the experimental group students (but not for the control group students who did not participate in the STEAM scenarios and did not play the game) included 8 additional Likert items addressing (1) students’ evaluation of the game, namely whether it was enjoyable and engaging and improved their learning and interest in STEM fields and careers and (2) students’ views on the role models that they had learned about in the project (whether they were inspiring, interesting, and relevant). The latter items were developed based on the questions used in the Shin et al. study [26], while items assessing students’ game evaluation were based on the dimensions of scales EGameFlow [43] and MEEGA+ [59]. Also, the pre-questionnaire included additional questions requesting gender and grade-level information as well as whether students knew anyone who had worked (a) in a math or science field, (b) as an engineer, or (c) in the information technology sector.

We carried out exploratory factor analysis and calculated the values of Cronbach’s alpha to evaluate the validity and reliability of each measure. Four (out of eleven) items addressing gender stereotypes relative to STEM were deleted because they did not correlate significantly with other variable items and/or did not have factor loadings above 0.5. Also, some of the stereotype items were reversed so that high values reflect stereotype endorsement. The values of Cronbach’s alpha for each measure are provided in

Table 1. Cronbach’s alpha values for pre- and post-intervention variable measures.

	Pre-Survey	Post-Survey
Perceived ability in STEM (4 items)	0.64	0.62
Interest in STEM studies and careers (4 items)	0.66	0.73
STEM gender stereotypes (7 items)	0.85	0.85
Game evaluation (5 items)	-	0.94
Role-model evaluation (3 items)	-	0.80

.

2.5. Research Procedure

Both questionnaires took approximately 10 min to complete. A brief paragraph in the beginning informed students about the purpose of the study, addressed anonymity and confidentiality, and stressed that participation was voluntary. The experimental group students responded to the pre- and post-questionnaires prior to and after the implementation of the STEAM DGBL scenarios, which took place in October–November 2022. The control group students, who did not participate in the STEAM DBGL scenarios, also completed both questionnaires at the same time. The questionnaires were available via Google forms, and students were given QR codes to access them using their own tablets or laptop computers during the morning registration period. Both times, students responded anonymously, using a personal code that was given to them privately by their form tutors. Permission for carrying out the study was obtained from the Cyprus National Bioethics Committee.

3. Results

3.1. Student Ability Perceptions, Career Interest, and Gender Stereotypes Relative to STEM

Data analysis examined between- and within-group differences in students’ (a) perceptions of ability in STEM, (b) study and career interest in STEM, and (c) gender stereotypes about STEM studies and careers before and after the intervention. Based on the value of the Kolmogorov–Smirnov test, the normality assumption was satisfied for only some of these six (three pre- and three post-intervention) variables, so the Wilcoxon sign rank and the Mann–Whitney tests were used to examine within-group and between-group differences, respectively, when the t-test was not appropriate. Pre- and post-scores for the two student groups are presented in

Table 2. Summary of student scores for the pre- and post-intervention study variables.

	n	Perceived STEM Ability		Interest in STEM Careers		STEM Gender Stereotypes
		Mean	SD	Mean	SD	Mean	SD
Control group
Pre-questionnaire	27	3.39	0.75	3.09	0.96	2.23	0.86
Post-questionnaire	27	3.41	0.70	3.11	0.88	2.31	1.08
Experimental group
Pre-questionnaire	69	3.65	0.71	3.14	0.86	2.36	0.83
Post-questionnaire	69	3.79	0.62	3.18	0.90	2.35	0.78

.

There were no significant between-group differences in perceived ability in STEM (t(94) = −1.607, p = 0.111), interest in STEM (Mann–Whitney U = 946.0, z = 0.119, p = 0.906), and STEM gender stereotypes (t(94) = −0.681, p = 0.498) prior to the intervention. There were also no differences between the two groups in the post-intervention scores on interest in STEM (t(94) = −0.362, p = 0.718) and STEM gender stereotypes (t(94) = −0.228, p = 0.820). However, the experimental group students expressed significantly higher perceived ability in STEM than the control group students in the post-questionnaire (Mann–Whitney U = 1196.5, z = 2.175, p = 0.030, r = 0.22). Further, there were no significant pre-post-questionnaire changes in interest in STEM, as well as in STEM gender stereotypes, both in the control (t(26) = −0.146, p = 0.885 and t(26) = −0.798, p = 0.432) and in the experimental group students (Wilcoxon signed rank z = −0.367, p = 0.714 and t(68) = −0.163, p = 0.871). Also, the control group students reported similar pre- and post-levels of perceived ability in STEM (Wilcoxon Signed Rank z = −0.598, p = 0.550). However, the experimental group had significantly higher scores on perceived ability in STEM after the intervention, and the effect size was medium (Wilcoxon signed rank z = −2.428, p = 0.015, r = −0.29).

Prior to the intervention, females in the experimental group scored lower than males on perceived ability in STEM (t(66) = −2.203, p = 0.03, r = 0.26); however, there was a significant improvement in female students’ scores (Wilcoxon Signed Rank z = 288.5, z = 2.930, p = 0.003), representing a large effect size (r = 0.50), and the gender difference on perceived STEM ability was eliminated after the intervention (Mann–Whitney U = 670, z = 1.138, p = 0.255) (see

Table 3. Summary of gender differences in experimental students’ scores.

	n	Perceived STEM Ability		Interest in STEM Careers		STEM Gender Stereotypes
		Mean	SD	Mean	SD	Mean	SD
Pre-questionnaire
Females	34	3.46	0.66	3.11	0.76	1.98	0.62
Males	34	3.82	0.72	3.15	0.97	2.79	0.80
Post-questionnaire
Females	34	3.68	0.59	3.19	0.89	1.98	0.65
Males	34	3.87	0.63	3.15	0.92	2.76	0.68

). Male students’ scores on perceived STEM ability, on the other hand, did not improve significantly (Wilcoxon Signed Rank z = 165.5, z = 0.451, p = 0.652). In addition, female and male experimental group students tended not to endorse gender stereotypes about the abilities of women and the suitability of STEM fields for them. However, both before and after the intervention, male students were significantly more likely to agree with gender stereotypes (Mann–Whitney U = 912, z = 4.114, p < 0.001, r = 0.50, and t(66) = −4.843, p < 0.001, r = 0.64, respectively). Finally, there were no significant gender differences in pre- and post-intervention scores on students’ interest in STEM studies and careers.

In summary, the above findings show that there was a significant improvement in experimental group students’—and, more specifically, in female students’—ability perceptions relative to STEM fields, which can be attributed to the STEAM educational game-based intervention, but no changes were observed in students’ interest in STEM or in their gender stereotypes regarding STEM.

3.2. Relations between Student STEM-Related Perceptions and Their Game and Role Model Views

It appears that although girls (M = 3.32, SD = 0.78) tended to like the game more than boys (M = 3.00, SD = 0.95), this difference did not reach statistical significance (Mann–Whitney U = 457.5, z = −1.510, p = 0.131). However, girls (M = 3.43, SD = 0.64) were significantly more likely than boys (M = 3.02, SD = 0.85) to report that the role models that they learned about in the project were inspiring, interesting, and relevant to them (Mann–Whitney U = 414, z = −2.055, p = 0.040). Anecdotal information provided by the classroom teachers who implemented the STEAM scenarios is consistent with these findings. Teachers observed that boys and girls tended to perform equally in making progress through the game and to be generally positive about the game. Most negative comments related to how “basic” gameplay was when compared to a modern game by a large studio. Teachers noted that students derived varying benefits from different types of role models. Some students could relate to the struggles and achievements of role models such as Stephanie Bendixsen, while other students were inspired by role models who were part of monumental achievements that changed the world. However, some students also reported feeling excluded by the project’s focus on girls and female role models, which may explain the above gender differences.

Backward multiple regression analyses were conducted to examine whether other study variables could predict students’ views about the game and the role models. As

Table 4. Predictors of experimental group students’ game evaluation and role model evaluation scores as well as of their post-intervention perceived ability in STEM.

Game Evaluation Score			Role Model Evaluation Score			STEM Perceived Ability 2
	b	R2		b	R2		b	R2
	Model 1			Model 1			Model 1
Parent in STEM	−0.010	0.415	Parent in STEM	−0.026	0.483	STEM perc. ability 1	0.770 **	0.631
Gender stereotypes 1	−0.002		Gender stereotypes 1	−0.072		Game evaluation	0.120
STEM perc. ability 1	0.003		STEM perc. ability 1	−0.093		Role model evaluation	0.065
Interest in STEM 1	−0.089		Interest in STEM 1	0.306 *
Role model evaluation	0.667 **		Game evaluation	0.590 **
	Model 5			Model 4			Model 2
Role model evaluation	0.639 **	0.408	Interest in STEM 1	0.249 *	0.469	STEM perc. ability 1	0.776 **	0.629
			Game evaluation	0.606 **		Game evaluation	0.161 *
Fmodel1 = 8.935, p < 0.001 Fmodel5 = 46.140, p < 0.001			Fmodel1 = 11,749, p < 0.001 Fmodel4 = 29.115, p < 0.001			Fmodel1 = 36.556, p < 0.001 Fmodel2 = 55.091, p < 0.001

Note: ** p < 0.001, * p < 0.01.

shows, the only factor that appeared to be a significant predictor of students’ liking of the game (the only predictor that was retained in the final model) was students’ evaluation of the role models, which was an important aspect of game design. Interest in STEM studies and careers prior to the intervention and students’ game evaluation were significant predictors of students’ evaluation of the role models.

Finally, backward multiple regression analysis (see Table 4) showed that students’ game evaluation was a significant predictor of their STEM ability perceptions at time 2 after controlling for their pre-intervention STEM ability perceptions. Role model evaluation score could not explain a significant part of the outcome variable, possibly due to its correlation with the student’s game evaluation score.

4. Discussion

The findings of the present study show that secondary school students who participated in an intervention in which they learned about the lives and accomplishments of STEM/STEAM role models via a DGBL approach improved their STEM ability perceptions compared to control group students. As a group, girls benefited more than boys, whose group level of perceived STEM ability did not improve significantly. It was interesting that before the intervention, female experimental group students had reported significantly lower confidence in their STEM abilities than their male counterparts, but after the intervention, the confidence levels of female and male students were similar. These findings suggest that exposing adolescent girls to role models of female STEM scholars via a DGBL approach is an effective way to improve their confidence in their STEM abilities. This is important because female students tend to underestimate their abilities in STEM subjects [12,13], while improving their ability perceptions may have long-term positive consequences for them and ultimately influence their career aspirations [6]. Ability perceptions influence several academic behaviors, including learning activity and school subject choices, but also persistence and learning effort [11], which prepare students for their future endeavors and increase the options that will be available to them.

Another interesting finding was that students’ post-intervention STEM ability perceptions were predicted by their evaluation of the FemSTEAM Mysteries game, which is consistent with theoretical perspectives that highlight the importance of game enjoyment and engagement as mediators for learning [43,44]. Students’ evaluation of the game was not related to other variables besides their evaluation of the role models that they had explored, indicating that the game appealed to a wide range of students, regardless of their interest in STEM and their perceived STEM abilities, degree of STEM gender stereotype endorsement, or having a parent in STEM, which could have made the game content more relevant to them. It seems that girls liked the role models and the game more than boys (although the latter difference did not reach statistical significance), which may explain why the intervention did not have an effect on male students’ ability perceptions. It is likely that boys liked the role models and the game less than girls because most of the game role models (six out of eight) were females, and, therefore, the perceived similarity with them was lower for boys compared to girls. This interpretation is supported by anecdotal information provided by the classroom teachers, who observed that some male students reported feeling excluded by the project’s focus on female role models. The above findings are also consistent with the conclusion of a recent literature review [7] that although role models belonging to groups that are underrepresented in STEM (e.g., women) may also influence majority-group students (e.g., males) in a positive way, this is not a consistent research outcome, suggesting that demographic similarity with role models may be important for all students.

The intervention did not have a direct effect on female students’ interest in STEM studies and careers, possibly because the information that was presented to students about the professional activities of the role models was not adequate to inspire interest and curiosity or to increase students’ appreciation of the value of the role-models’ work. In other studies, e.g., [30,33], in which student interest and aspirations in STEM increased, STEM role models talked to students more extensively about their careers and professional experiences. In the present study, there was also no effect on student gender stereotypes regarding the abilities of women and the suitability of STEM studies and careers for women, possibly because students, especially girls, did not endorse stereotypes in the pre-questionnaire. Also, previous research with adolescent and university female students has shown that such stereotypes are difficult to change after only one intervention [25,27].

The findings of the present study provide evidence for the positive impact that quality learning games may have on student affective outcomes, and more specifically on student STEM ability perceptions, and contribute to the relevant literature on DGBL and on STEM role models by showing that digital games can be an effective way to expose secondary school students to STEM role models. This is important because although enabling students to interact with actual STEM scientists and professionals in real time may have several advantages, it is not practically easy to implement this approach in most school settings. Another advantage of learning games compared to other alternative approaches, such as presenting interviews with STEM professionals in video, audio, or text format [36,38,41,42], is that learning games are self-contained digital resources that can be more easily integrated into classroom instruction by teachers themselves. Unlike previous role model interventions targeting secondary school students that were typically designed and led by researchers, e.g., [28,29,30,31,32,33,37], in the present study, teachers were actively involved in the intervention because they integrated the [FemSTEAM Mysteries game into STEAM educational scenarios that they designed and implemented with their own students.

Although the findings of the present research are very promising, the study did not use a follow-up design to assess whether the intervention had an enduring effect on female students. In addition, the study was quasi-experimental, and students were not assigned randomly to the experimental and control groups. Also, for practical reasons, the students of the two groups did not belong to the same grade level and attended the same school. These conditions may have compromised internal validity due to the lack of control over possible age differences and the potential influence of experimental group peers on control group students. Finally, it was not possible to use more than one experimental condition to evaluate the unique effects of utilizing role models, STEAM education, and game-based learning. However, carrying out the study in classroom settings and using educational scenarios designed by in-service teachers strengthens the study’s ecological validity and supports the generalizability of its findings to similar school contexts.

Future research needs to focus on the characteristics of games and the role models that contribute to their attractiveness and effectiveness for secondary school students and to examine more thoroughly the aspects of role-model characteristics that can impact students’ STEM interest and gender stereotypes about STEM studies and careers.