Introduction

Problem based learning (PBL) has great potential to enhance learning outcomes and cognitive abilities in educational contexts and along with this potential, it has attracted great attention from researchers and educators (Arruzza et al., 2023; Yew & Goh, 2016). The effect of PBL on learning outcomes has been investigated by different researchers from all over the world. There are many meta-analysis studies investigating the effectiveness of PBL on learning outcomes including academic achievement, critical thinking, or attitude in pharmacology education (Liu et al., 2019; Zhou et al., 2016), nursing education (Sharma et al., 2023; Yan et al., 2023), radiology education (Zhang et al., 2018), and psychology education (Gao et al., 2020). Also, there are other meta-analysis studies focusing on different disciplines to investigate the effect of PBL like science education (Funa & Prudente, 2021; Üstün, 2012), mathematics (Fitriyah et al., 2022), and physics (Nora, 2020). Although there are other meta-analysis studies related to different disciplines, it is clear that the meta-analysis studies focusing on the different disciplines of medical education dominate the PBL literature. It can be said that this is not surprising because PBL has rooted in medical education and it has been widely used as an educational concept since the 1960s (Jones, 2006). There are also some other meta-analysis studies (Batdı, 2014; Dağyar & Demirel, 2015; Erdem, 2022; Liu & Pásztor, 2022) not focusing on any disciplines but examining the effectiveness of PBL on academic achievement or attitude as well as some cognitive abilities like critical thinking of students from different educational levels like high school or elementary school. All of these meta-analysis studies provided positive evidence regarding the effect of PBL on academic achievement and attitude as well as cognitive abilities of students from diverse disciplines and educational levels.

Though PBL can be seen as a relatively new concept for English as a Foreign Language (EFL) context (Ansarian & Mohammadi, 2018), the number of experimental studies investigating the effectiveness of PBL in EFL classrooms is increasing. Most of these experimental studies provide evidence regarding the positive effect of PBL on language skills although their effect sizes differ. However, there are also a few studies (Asmara, 2022) concluding that PBL is not more effective than traditional teaching to enhance academic achievement in EFL classrooms. Therefore, it can be said that the question of how effective PBL is to enhance academic achievement in EFL classrooms remains as an important problem to be answered which means that there is a clear need for a meta-analysis study to combine the results of previous studies and provide a comprehensive picture of the effectiveness of PBL in EFL classrooms. However, there are not any meta-analysis studies aiming to investigate the general effect size of PBL in EFL context although various meta-analysis studies have proved its effectiveness in diverse disciplines. This meta-analysis study addresses this gap and intends to contribute to a solution of this question and a better understanding of the effectiveness of PBL in EFL classrooms by providing meta-analytic evidence regarding the general effect of PBL on EFL students’ academic achievement. Addressing this issue and providing a general effect size of PBL in EFL classrooms is important because its effectiveness in EFL classrooms has not yet been proven.

Therefore, the first aim of this study was to examine the overall effect size of PBL on achievement in EFL classrooms. The second aim was to examine the variability of the overall effect size of PBL and then to investigate study characteristics that may explain this variability. To this end, answers to the following questions were sought for:

  1. 1.

    What is the overall effect size of PBL on achievement in EFL classrooms?

  2. 2.

    Does this effect size significantly differ by moderator variables, namely, language skills, education level of study group, treatment duration, location of study, learning environment, and document type?

This study intends to contribute to EFL literature by providing general effect of PBL on academic achievement of EFL students which can show us how effective PBL is to enhance academic achievement in EFL classrooms. Also, this study can significantly contribute to PBL literature by providing meta-analytic evidence regarding the effectiveness of PBL from an EFL perspective. In addition, this study is important because it intends to examine the possible moderators, including learning environment, on the effectiveness of PBL in EFL context. As researchers and teachers have started to be interested in online PBL and it has started to be implemented in blended or online learning environments for the last few years (Valtonen et al., 2022), investigating the effectiveness of learning environment as a possible moderator is critical. In short, this study is important, intends to contribute to the related literature, and differs from other meta-analysis studies because it focuses on the effectiveness of PBL on academic achievement of EFL students and includes some possible moderator variables which have not been investigated before.

Literature Review

Problem Based Learning

PBL may be seen as a student-centered strategy that offers contextualized, ill-structured, and real-life problems, allows for collaborating in small groups, discussing and applying new knowledge, interaction with peers, and active participation, and provides learners with guidance, instruction, and opportunities to reflect as they increase their content knowledge (Arruzza et al., 2023). Also, Yew and Goh (2016) define PBL as a student-centered approach which uses problems derived from real life to engage learners actively in learning process. The main point of PBL is that learning starts with a problem, query, or question that learners try to solve (Charlin et al., 1998) and during this process, learners actively process and organize information, research, use their prior knowledge, and interact with meaningful content (Gallagher, 1997). It can be said that PBL does not merely refer to creating problem solving opportunities and it makes problem solving the main reason to learn. Learners are required to seek out, research, think, analyze, examine, and use the knowledge needed to generate ideas to find a solution for a specific problem (Alghamdy, 2023) which constitutes the real power of PBL because learners also acquire the knowledge required to solve similar problems while they are trying to find a solution for a specific problem (Jones, 2006).

PBL originates from constructivism which indicates that learners actively construct knowledge through dialogue and social interaction among learners (Saqr & Alamro, 2019). Figure 1 shows the PBL framework proposed by Savery and Duffy (1995) and its instructional principles through constructivism. The first step of the framework is “ownership for the task” which means that learners should prize the goals and relevance of the tasks instead of seeing them as simple assessment activities that should be passed. This can be assured by using real-life problem scenarios and within authentic learning environments. PBL gives freedom for learners while solving a problem which makes the learning process personalized. Also, teachers should not tell the students what to do or how to solve a problem. Instead, they should guide and facilitate learning and inquire at the “leading edge” of the learner’s thinking by scaffolding and challenging thinking abilities of learners. The learning in PBL is multidirectional which means problems do not have one simple solution and learners should test some solutions and ideas against alternative ones. The last step of the framework is “reflection” which refers to satisfying the goal of developing self-regulation and independence abilities by reflective process.

Fig. 1
figure 1

The principles of PBL proposed by Savery and Duffy (1995) through a constructivist framework

Full size image

The potential of PBL to enhance academic achievement has been investigated and proved by many meta-analysis studies focusing on different education levels or diverse disciplines. In his meta-analysis study, Erdem (2022) concluded that PBL has a large effect (ES = 0.82) on the secondary school students’ academic achievement. Similarly, Batdı (2014), in his meta-analysis study, concluded that PBL has a large effect on academic achievement (ES = 1.30). Also, Sugano and Nabua (2020) revealed that PBL has a large effect on students’ chemistry achievement (ES = 1.27) in their meta-analysis study. In a similar way, Funa and Prudente (2021) found that PBL has a large effect on secondary school students’ achievement in science (ES = 0.87). In short, previous meta-analysis studies have provided evidence regarding the large effect of PBL on academic achievement related to different disciplines.

Problem Based Learning in EFL Context

The adaptation of PBL in language classrooms does not date back more than two decades and it can be seen as a relatively new concept for EFL context (Ansarian & Mohammadi, 2018). Teachers were skeptical about the effectiveness of PBL in language classrooms at the beginning; however, it is seen as an innovative approach for language learning today (Ansarian et al., 2016). PBL is a suitable approach for language classrooms to create meaningful language teaching/learning activities, trigger engagement, and enhance motivation resulting in a meaningful language learning environment (Bosuwon & Woodrow, 2009). It remains evident that PBL has great potential for improving language skills, communication opportunities, and content knowledge in language classrooms and language learners have a chance to incorporate and utilize factual knowledge and target language into their lives during PBL activities (Sadeghi et al., 2016) because they are exposed to useful, meaningful, and real-world content knowledge (Boaler, 1997). In language classrooms, learners need to research, discuss, communicate, produce, and present their results in the target language during collaborative problem solving activities. Also, learners have a chance to share their findings and results regarding the problem with their friends in the target language while their friends have a chance of listening and sharing their ideas in the target language at the same time which creates a lot of opportunities to use the target language meaningfully resulting in better language skills (Mathews-Aydinli, 2007).

In his study, Alghamdy (2023) concluded that PBL significantly increased EFL learners’ writing skills as well as their motivation. Similarly, in their quasi-experimental study, Jumariati and Sulistyo (2017) and Ibnian (2023) revealed that PBL significantly increased EFL learners’ argumentative and expository writing skills. In a quasi-experimental study, Lin (2018) concluded that PBL significantly increased students’ reading comprehension abilities. Also, there are some other studies providing evidence regarding the positive effect of PBL on speaking (Ansarian et al., 2016), listening (Lin et al., 2019), and vocabulary (Fard & Vakili, 2018) skills. On the other hand, Asmara (2022) conducted a study to investigate the effect of PBL in EFL classroom and concluded that although PBL significantly increased the academic achievement of EFL students, there was not a significant difference between PBL group and non-PBL group. In short, related literature includes some studies providing evidence regarding the positive effect of PBL on language skills although their effect sizes differ while some of them concluded that PBL-students did not outperform their non-PBL counterparts.

Method

In this meta-analysis study, updated PRISMA 2020 guidelines proposed by Page et al. (2021) were followed.

Collection of Studies and Inclusion/Exclusion Criteria

Some online databases including Web of Science, Google Scholar, Scopus, and ERIC were systematically searched to find out the studies investigating the effect of PBL on achievement in EFL classrooms. The literature review was carried out using the search pattern below which was determined based on the PICOS (Population, Intervention, Comparison, Outcomes, Study Design) framework for systematic reviews (Methley et al., 2014).

(“language learn*” OR “language teach*” OR “L2 learn*” OR “L2 teach*” OR “language student*” OR “foreign language learn*” OR “EFL” OR “language class*”)

AND

(“problem based learning” OR “PBL” OR “problem-based learning”)

AND

(“empirical study” OR “experiment*” OR “randomized controlled trial” OR “RCT” OR “randomized trial”)

As it can be seen in Fig. 2, after the literature review, 8473 studies were revealed in total by the last search on 31st of December 2023. No limit was established for the publication date. Five criteria were determined for the studies to be eligible for inclusion in the meta-analysis. First, studies must be in English or Turkish because of the researcher’s language limitation. Second, an experimental or quasi-experimental design with a control group must be employed in the studies. Third, studies must be related to the effectiveness of PBL on EFL academic success. Fourth, the traditional method must be employed in the control group while PBL was employed in the experimental group. Fifth, studies must report the necessary data to calculate effect size.

Firstly, the studies were examined through their abstracts and titles, and 7780 studies were excluded due to some reasons (e.g., not related to PBL or EFL classroom, not in Turkish or English, or duplicates, etc.). After that, 693 studies were assessed for eligibility and they were examined by two researchers using the inclusion criteria abovementioned and 657 studies were excluded. 264 of these studies did not employ experimental design and 306 of them were not related to language success in EFL classrooms although they were experimental studies. Also, 28 of them did not have a control group and the traditional method was not employed in the control group in 19 studies. Besides, 40 studies did not report necessary statistical data to calculate effect size. The authors of these 40 studies were contacted via e-mail and necessary statistical data were requested from them, but they did not respond to the e-mails. Finally, 36 studies were included in the meta-analysis. As some of the studies shared more than one data (e.g. two study groups from different educational levels, two different language outcomes like grammar and vocabulary, etc.), meta-analysis was conducted with 41 data (extracted from 36 unique studies). For the included studies, online supplementary material can be seen. While reviewing the primary studies, all discrepancies between two researchers were resolved before going on to the next phase.

In addition, the reference lists of the collected studies were investigated to uncover any other potential studies that may fulfill the inclusion criteria. However, no eligible study was found. As a result, the total sample number of the included studies was 1294 for the control group and 1290 for the treatment group. All of the studies targeted L2 learning and while the participants’ native language varied, the target language was English in all of the studies.

Fig. 2
figure 2

Flow diagram. Note k = the number of studies; EFL = English as a Foreign Language

Full size image

Coding of Studies

Before starting to extract the data from the studies, the researcher of this study developed a detailed coding manual and the studies were coded by two researchers independently. Both of these researchers had a PhD degree and meta-analysis experience. One of them works in the field of English Language Teaching while the other one works in the field of Educational Sciences. Full consistency was seen between the coders (r = 1.00). Name of the study, year published, author(s), location of study, education level of study group, examined language skills and components, learning environment, treatment duration, mean, standard deviation, and sample number of the treatment and control groups were extracted from the studies. For the descriptive results of the included studies and detailed coding results, online supplementary material can be seen.

Data Analysis and Interpretation

The Comprehensive Meta-Analysis (CMA) package program (Borenstein et al., 2022) was used to analyze the data. Mean and standard deviation values based on the results of immediate posttests and sample number of the treatment and control groups were used to calculate the effect sizes. Hedges’s g coefficient was used for the effect size calculations and the confidence interval was determined as 95% in all calculations. Publication bias was investigated using funnel plot and Duval and Tweedie’s trim-and-fill method. Also, Q statistics and I2 values were investigated together to check the heterogeneity among the studies. Q value which is greater than the critical limit in the χ2 table at k− 1 degrees of freedom value (k represents the number of effect sizes) shows that there is heterogeneity among studies. Also, I2 value ranges of 25%, 50%, and 75% indicate low, medium, and high heterogeneity, respectively (Cooper, 2017). As homogeneity hardly exists among the studies from social sciences and the random effects model should be preferred while conducting meta-analyses including studies from social sciences (Schmidt & Hunter, 2015), the overall effect size was calculated using the random effects model in this study.

Subgroups of language skills (grammar, reading, speaking, etc.), education level of study group (K12 and university), treatment duration (less than 7 weeks and more than 8 weeks), location of study (Asia and Middle East), learning environment (in-class, online, and blended), and document type (unpublished and published) were determined as categorical variables. For the education level of study group, studies were grouped under two categories which are K12 and university because most of the studies were conducted with university students (n = 19). For the treatment duration, some of the studies (n = 6) did not report a specific duration of treatment and instead of it, they just mentioned the academic year when the treatment occurred which can be seen as an important deficiency. The other studies which reported a specific duration of treatment were grouped under two categories, namely, less than 7 weeks and more than 8 weeks because less than 7 weeks can be seen as short/medium term while more than 8 weeks can be regarded as long term for experimental studies in educational contexts and similar categorization was used in previous meta-analysis studies (e.g., Lin & Lin, 2019; Liu & Pásztor, 2022; Ngo et al., 2024). For the location of study, nearly half of the studies were carried out in Indonesia (n = 16), China (n = 6), Malaysia (n = 5), and Taiwan (n = 4). Previous PBL meta-analysis studies focusing on different disciplines have also mentioned that Indonesia is the top location for PBL studies (Erdem, 2022). Therefore, the included studies were grouped under two categories, namely, Asia and Middle East. For the document type, while proceedings and thesis were grouped as unpublished studies, articles were grouped as published studies.

Moderator analyses were performed using Analog ANOVA with a mixed-effects model. Analog ANOVA produces different Q-statistic values like between-group (QB), within-group (QW), and total (QTOTAL). QB statistic shows the between-group homogeneity and can be used to decide if the categorical variable is a real moderator or not (Lipsey & Wilson, 2001). If QB value is significant and greater than the χ2 table critical value, it means that the effect size of PBL varies between categories of the moderator variable.

Results

Results on Publication Bias

The funnel plot shown in Fig. 3 was investigated using the trim-and-fill method by Duval and Tweedie (2000) to check the publication bias.

Fig. 3
figure 3

Funnel plot

Full size image

In the funnel plot, symmetrically distributed studies around the general effect size indicate the lack of publication bias. Figure 3 shows that the funnel plot seems symmetric. Also, the funnel plot was investigated using the trim-and-fill method and it was seen that only two imaginary studies are needed to add to eliminate publication bias totally. Also, when these two imaginary studies are included in the analysis, the overall effect size rises to 1.12 from 1.06 which can be seen as a small difference. Therefore, we can say that there is no publication bias in the results of this study.

Results on the Effect Size of PBL on Academic Achievement

Table 1 shows that Q(df=40) value was calculated as 184.760 (p < 0.05) which is greater than the χ2 table critical limit with 40 degrees of freedom and in 0.05 confidence interval (df = 40, χ2(0.05) = 55.759). Therefore, it can be said that the heterogeneity exists among the studies. In addition, the calculated I2 value (78.35%) means a high level of heterogeneity. According to the random effects model, the effect size of PBL was calculated as 1.067 (95% CI = 0.88–1.24). Based on Cohen’s (1992) classification, values higher than 0.8 indicate a large effect. Therefore, this meta-analysis study showed that PBL has a positive and large effect on academic achievement in EFL classrooms.

Table 1 Effect size of PBL and the heterogeneity test results in random effects model
Full size table

Also, as it can be seen in the Fig. 4, out of 41 effect sizes, only one was negative. Therefore, it can be said that PBL has a positive effect on EFL success in the majority of the studies.

Fig. 4
figure 4

Forest plot

Full size image

Results on the Moderator Analyses

Table 2 shows that the heterogeneity value of the subgroup of skill (QB = 28.171, p < 0.05) is greater than the critical value of χ2 table which means that it is a real moderator on the effect size of PBL on EFL success. PBL has the greatest effect on grammar (g = 2.70) and grammar is followed by speaking (g = 1.71), listening (g = 1.60), writing (g = 1.04), vocabulary (g = 1.03), reading (g = 0.74), and general language success (g = 0.74). On the contrary, the heterogeneity value of the subgroups of study group (QB = 2.003, p > 0.05), treatment duration (QB = 2.011, p > 0.05), location of study (QB = 1.716, p > 0.05), learning environment (QB = 4.633, p > 0.05), and document type (QB = 0.049, p > 0.05) are lower than the critical value of χ2 table indicating that these variables are not real moderators on the effect size of PBL on EFL success.

Table 2 Comparison of effect sizes across categorical moderators
Full size table

Conclusion and Discussion

This meta-analysis study showed that PBL has a large and positive effect on academic achievement in EFL classrooms (Hedges’s g = 1.067). Also, investigation of publication bias using various methods revealed that there is no publication bias. This result is important because it shows us that PBL is highly effective to enhance the language achievement of students in EFL classrooms although it is a relatively new concept and its use in language classrooms does not date back more than two decades.

Previous meta-analysis studies (Batdı, 2014; Dağyar & Demirel, 2015; Erdem, 2022) concluded that PBL has a large effect on enhancing academic achievement of students from different education levels and this effect size does not differ according to disciplines like science, mathematics, or social sciences. Also, other meta-analysis studies (Funa & Prudente, 2021; Gao et al., 2020; Sugano & Nabua, 2020) focusing on specific disciplines apart from EFL concluded that PBL has a large effect on academic achievement related to diverse disciplines like science, chemistry, or medical psychology. This meta-analysis study also echoed similar results regarding the large effect of PBL on EFL students’ academic achievement which shows that PBL is as highly effective as it is in other disciplines to enhance academic outcomes in EFL classrooms.

PBL can be seen as a suitable and innovative approach for language learning (Ansarian et al., 2016) because it creates meaningful language learning activities, triggers engagement, and enhances motivation which results in a highly meaningful language learning environment (Bosuwon & Woodrow, 2009). As language learners are exposed to meaningful, useful, and real-world content knowledge during PBL (Boaler, 1997), they have a chance of utilizing and incorporating factual knowledge and target language into their lives (Sadeghi et al., 2016). Also, PBL creates many opportunities for learners to use the target language meaningfully (Mathews-Aydinli, 2007). Therefore, it can be said that the result of this meta-analysis study regarding the large effect of PBL on the academic achievement of EFL students aligns with this theoretical background.

Moderator analyses revealed that the language skill is a real moderator on the effect size of PBL on EFL success. PBL has the greatest effect on grammar and grammar is followed by speaking, listening, writing, vocabulary, reading, and general language success although all of these effect sizes refer to large effect. During collaborative PBL activities, language learners have to research, discuss, communicate, produce, and share their findings regarding the problem situation with their friends in the target language while their friends have to listen, comment, and share their ideas in the target language at the same time which creates a lot of opportunities to use the target language meaningfully resulting in better language skills (Mathews-Aydinli, 2007). This can explain the result of this study indicating speaking and listening are two skills which have the largest effect sizes after grammar. However, it can be said that PBL produces large effect sizes regarding productive and receptive language skills as well as language components like vocabulary and grammar which may be seen as an important result because it shows us that PBL can be implemented to enhance various language skills and components.

Moderator analyses also revealed that the study group, treatment duration, location of study, learning environment, and document type are not real moderators on the effect size of PBL. Previous meta-analysis studies investigating the effectiveness of PBL on academic achievement in different disciplines revealed similar results regarding study group (Dağyar & Demirel, 2015; Funa & Prudente, 2021; Üstün, 2012), treatment duration (Batdı, 2014; Dağyar & Demirel, 2015; Erdem, 2022; Üstün, 2012), location of the study (Erdem, 2022), and document type (Dağyar & Demirel, 2015) are not significant moderators on the overall effect size of PBL. Therefore, it can be said that the results of this meta-analysis study are confirmed by similar meta-analysis studies.

On the other hand, in his meta-analysis study investigating the effect of PBL on academic achievement and motivation in science, Üstün (2012) concluded that the location of the study was a significant moderator. However, he categorized the location of the study as Türkiye, the USA, and others. Also, in their meta-analysis study investigating the effectiveness of PBL on students’ mathematical problem-solving skills, Musna et al. (2021) concluded that the treatment duration including the categories of two-four, five-six, and more than six meetings was a significant moderator. Therefore, it can be said that these meta-analysis studies focusing on different disciplines concluded contradictory results regarding some moderators which can be explained by the different categorizations of the researchers.

The result indicating that the study group is not a significant moderator on the effect size of PBL shows us that PBL can be effectively used with the students from K12 and tertiary level. Also, the result indicating that the learning environment is not a real moderator provides critical insights into PBL in online environments. Although PBL has been subject to different studies since the 1960s, researchers and teachers have started to be interested in online PBL for the last decade (Valtonen et al., 2022). Thanks to massive development in internet and computer technologies, online learning environments have started to be an alternative to traditional learning environments all over the world (Chao et al., 2022) and various usable technological devices and internet tools have emerged (Donnelly, 2010). Today, many potentially beneficial online learning platforms that offer functionality including embedded search engines, document sharing, and access to online materials are available for researchers and educators and these platforms provide great online learning environments for PBL. Therefore, PBL has started to be implemented in online or blended learning environments for the last few years. This meta-analysis study shows that online PBL has a large effect on the academic achievement of EFL students indicating online learning environments are as promising as traditional classrooms. This result is important because it provides evidence regarding the effectiveness of online learning environments for PBL and shows that blended or online PBL can be an alternative to face-to-face PBL although traditional learning environments (face-to-face) have a longer history of implementation for PBL.

In short, this study showed that PBL has a large and positive effect on academic achievement in EFL classrooms. Also, the language skill is a real moderator on this effect size of PBL on EFL success while the study group, treatment duration, location of study, learning environment, and document type are not real moderators. As this study shows that PBL is highly effective to enhance academic outcomes in EFL classrooms, language teachers can include PBL as an effective way of teaching EFL in their classrooms. Also, school administrators should encourage their EFL teachers to conduct PBL activities to enhance their language teaching quality. Besides, this study provides clear implications for education policies to develop highly effective learning environments and design a more efficient EFL curriculum by integrating PBL into EFL teaching/learning activities. In addition, this study shows that online and blended PBL has a large effect to enhance academic achievement of EFL students and they are as highly effective as face-to-face PBL. Therefore, it can be suggested that language teachers can move their PBL activities into online or blended learning environments to diminish the challenges of traditional classrooms.

This study has significantly contributed to the EFL literature because it presents evidence regarding the effectiveness of PBL on enhancing academic achievement of EFL students although it is a relatively new concept in language classrooms. Also, this study has made an important contribution to the PBL literature from the EFL context because the literature lacks a meta-analysis study focusing on EFL classrooms although it has various meta-analysis studies focusing on other disciplines and proving the effectiveness of PBL. In addition, this study is important because it reveals that PBL is as highly effective as it is in other disciplines to enhance academic outcomes in EFL classrooms by echoing similar results regarding the large effect of PBL on academic achievement with previous meta-analysis studies focusing on different disciplines. Moreover, this study is important because it examined the possible moderators, including learning environment which has not been investigated before, on the effect of PBL and showed that online and blended PBL has a large effect on the academic achievement of EFL students indicating online learning environments are as promising as traditional classrooms.

Limitations and Recommendations for Future Studies

This study is not free from limitations. First, there might be other possible moderator variables that were not considered in this study. So, it may be suggested that other possible moderator variables can be examined in future meta-analysis studies. Also, only categorical moderator variables were examined in this study; however, there might be some other continuous moderator variables like % female or mean age of the participants, etc. which can be examined using meta-regression. Therefore, future studies should investigate the effect of possible continuous moderator variables. Second, for the learning environment variable, the number of studies is limited for online (n = 4) and blended learning (n = 1) environments while it was 36 for traditional in-class environments and this can be shown as a limitation. Therefore, other researchers can examine the effect of this moderator more deeply when the number of studies for online and blended PBL increases. Although this meta-analysis study searched some international databases, most of the included studies were carried out in Indonesia which can be shown as location bias.