Introduction
In recent years, self-efficacy (SE) as an individual characteristic of learners has been widely explored for its effect on learning skills (Bruning et al., 2013; Komarraju & Nadler, 2013; Sanders-Reio et al., 2014; Wang & Bai, 2016). Hayes (2000) stated that the writing process is a communicative and generative effort involving both cognitive and affective factors. Therefore, self-efficacy and self-regulation strategies, as they are influential for learners, are assumed to play an important role in writing according to cognitive process theory. Previous studies have shown that motivational variables exert a great impact on writing behaviour and performance (Kormos, 2012), and self-efficacy related to motivation reflects learners’ self-judgement about their writing ability (Pajares, 2003; Pajares & Valiante, 1997).
With English for Academic Purposes (EAP) teaching playing an increasingly important role in higher education within China, many scholars have discussed project-based learning (PBL) writing,* as it is widely used in university academic writing English courses (Wang et al., 2016; Cai, 2019; Zhao & Lv, 2019; Zheng, 2019). As a particular teaching model of academic writing, project-based writing (PBW) requires students to work in groups in order to find a problem within the discipline and search for solutions by reading, researching, and writing. Although scholars have tried to establish the relationship between self-efficacy and academic writing, such as summary writing or source-based writing (Golparvar & Khafi, 2021), studies on this topic are far from sufficient. For complicated academic writing, few studies have ever focused on the effect of SE on PBW that demands higher and deeper competencies, such as collaborative discussion, academic source searching, literature reading, summary writing, and synthesizing.
Therefore, the current study intends to explore the effects of SE on the use of SRS and PBW in order to fill this research gap and provide suggestions for the improvement of both learners’ SE and PBW teaching.
Theoretical frameworks and related literature
According to the cognitive process theory of writing (Hayes, 2000), both cognitive and affective factors, such as self-efficacy and self-regulation, are vital for successful writing outcomes. Social cognitive theory of writing also states that learners with higher levels of self-efficacy are more likely to make stronger efforts, exercise more persistence, and thus have better writing performance (Usher & Pajares, 2008); self-regulation refers to reciprocal environmental, behavioral and personal processes (Bandura, 1986). The present study will explore the interdisciplinary collaboration between cognitive science, psychology, and writing (Chien, 2012) in academic contexts within the above two theoretical frameworks.
Background of self-efficacy (SE) in writing
Many scholars have studied how self-efficacy affects learners’ writing performance. Prat-Sala and Redford (2012) discovered that self-efficacy in both reading and writing contributed independently to writing performance for college students. Pajares and Valiante (2001) also found that self-efficacy was positively correlated with both language achievement and other constructs such as academic self-concept, self-regulation, achievement goals, value, task goal orientation, and performance approach. Conversely, self-efficacy negatively correlated with apprehension and performance-avoidance goals. Self-efficacy was also correlated with the use of language learning strategies (Magogwe & Oliver, 2007) that metacognitively affect learners’ writing performance. Woodrow (2011) and Zhang and Guo (2012) discovered the mediational effect of writing SE on the association between writing proficiency and affective constructs (i.e., anxiety or motivation).
Most of the above studies show that the higher the level of personal efficacy is, the greater the effort made and the more time spent achieving learning goals. However, the question concerning the effect of self-efficacy on academic writing performance awaits further research in how self-regulation strategies function in writing.
Background of self-regulation strategies (SRS) in writing
Traditional theories of learning strategies focused on the different aspects of language learning (listening, speaking, reading, writing) and grouped the strategies based on educational and psychological theories - behaviorist learning theory, cognitivist learning theory, and constructivist learning theories (O’Malley & Chamot, 1990; Oxford, 1990). Later, the inconsistency of the definition of strategies made Rose et al. (2018) urge researchers to move away from language learner strategies (metacognitive strategies, cognitive strategies, socio-affective strategies) to self-regulation strategy research. A large body of research has suggested positive relationships between writing SRS use and students’ writing proficiency (Asmari, 2013; Ching, 2002; García-Sanchez & Fidalgo-Redondo, 2006; Teng & Huang, 2019). Graham and Harris (2000) identified sixteen SRSs in writing, including goal setting and planning, seeking information, record keeping, organizing, transforming, self-monitoring, reviewing records, self-evaluating, revising, self-verbalizing, rehearsing, environmental structuring, time planning, self-consequence, seeking social assistance, and self-selecting models. These SRSs, functioning as intermediate factors, help writers flexibly regulate metacognitive, cognitive, and socio-affective performance (motivation, personal behaviour, cooperation, etc.) to direct their writing process, which finally affects their writing outcomes and self-efficacy.
Teng and Zhang (2016) proved that the SRS model is constructed of cognitive, metacognitive, interpersonal support, and motivational regulation aspects. The empirical evidence offers preliminary support to a transfer of self-regulated learning (SRL) theory from educational psychology to the field of L2/EFL education, particularly L2/EFL writing. Although the SRS model proposed by Teng and Zhang (2016) might not be directly applied to PBW that especially involves students’ complex individual and interactive activities with long-text academic outcomes, the present study adapted their strategy model of SRL by specifying the use of strategies in the concrete tasks.
Writing self-efficacy (SE) and writing self-regulated strategies (SRS)
In the domain of writing, studies also revealed that writing self-efficacy had a positive correlation with SRS use (Lavelle & Guarino, 2003; Schunk & Zimmerman, 2007; Zimmerman & Risemberg, 1997). Self-efficacious students were found to use more cognitive and metacognitive strategies and persist longer in the face of adversity compared to their less efficacious counterparts (Pajares, 2007; Wang & Bai, 2016). Steward, Steifert and Rolheiser (2015) found that an enhancement of undergraduate students’ writing self-efficacy was related to an improvement in their perceptions of the utilization of metacognitive writing strategies. Ching (2002) argued that instruction in self-regulation strategies increases ESL students’ self-efficacy beliefs. However, in Graham et al. (2005)’s study, after receiving self-regulated strategy development (SRSD) instruction, the students’ self-efficacy beliefs were not found to have improved.
Therefore, the complicated interaction among SE, SRS, and academic writing indicates a need for further exploration.
PBW: Project-cased writing in academic contexts
PBL is learner-centered (Wilkerson & Gijselaers, 1996), and it involves students learning cognitively and interactively in a group. As a PBL teaching model in writing courses, PBW is a type of task-based and process-oriented writing that is frequently used to solve problems in disciplines. It requires students to form groups, work together to discover a problem they are interested in, formulate a project that they will research, and finally compose academic reports or arguments laying out solutions. In the whole writing process, teachers provide scaffolding by introducing searching skills, research methods, and writing knowledge that PBW needs to support students.
The process of PBW requires the ability to use discipline-specific norms (including vocabulary, sentence patterns, discourse structures, and styles within the field) (Capraro et al., 2013); metacognitive competence (Jordan 1980 cited in Ong 2014; 1989 cited in Stewart et al. 2015; Robinson 1980); an understanding of research methods, research procedures, value judgement, and demonstration methods (Hyland, 2014). For example, the British Association of Lecturers in English for Academic Purposes (BALEAP) categorizes postgraduate academic English competencies as follows: language and discourse skills (e.g., understanding the problems frequently encountered in the course and mastering the commonly used terms in the discipline), academic cognition and metacognitive abilities (e.g., understanding the purpose of reading and writing, critical analysis of text), subject research skills (subject-specific knowledge building), and the practical skills related to the discipline (such as how to find and sort out the literature) (Sun & Wang, 2020).
For the particularity of PBW in academic contexts, the effects of self-efficacy on the use of SRS and academic writing are worth exploring in depth. Therefore, based on the above-documented skills and competencies required by PBW, self-efficacy is here grouped as the following three factors: language self-efficacy, literature search and read self-efficacy, and research self-efficacy. The current study intends to explore how and to what degree the three self-efficacies affect PBW performance in these areas.
Specifically, the research questions of the study are as follows:
How does writing self-efficacy influence self-regulation strategy use in project-based writing (PBW) activities?
How does self-efficacy influence PBW?
Methodology
Participants
The sample for this study was selected from undergraduate students at a Chinese economics university enrolled in academic writing courses. The sample consisted of 107 students from four academic writing in English classes taught by one of the researchers. The writing course was guided by one of the researchers for eight times in one semester with the themes of academic writing skills. Besides academic writing courses, participants studied four hours of integrated English per week. The participants’ general language proficiency ranged from High B1 to High B2 according to the Common European Framework of Reference for Languages (CEFR). The mean age of the respondents was 20.5, ranging between 19 and 22. The sample consisted of thirty-nine males and sixty-eight females majoring in different fields of study. This skewness reflects the gender ratio of the university, which is favoured by female students who are perceived to perform better in foreign language learning. All the participants took part in the survey voluntarily.
Subjects
| Subjects ID | Year 1 and 2 university students |
| No. | 107 |
| Average age | 20.5 |
| Gender | 39 males; 70 females |
| English proficiency | CEFR High B1-High B2 |
| Courses taken | Two terms of academic writing; integrated English course |
Data collection procedures
This study was designed to collect data from undergraduates enrolled in an academic writing course at an economics and business university in Shanghai in 2020. The course ran from early February to early July. Students received eight sessions of PBW instruction. They were asked to do small-scale research in groups of three to four and submitted individual academic reports of 2,000-2,400 words at the end of the term. At the end of the course, the questionnaire was sent to participants through an online platform (Wen Juan Xing, Changsha Ranxing Company) by one of the researchers, who was also teaching the course. The students were informed clearly about the purpose of the study and told that their participation in the study was voluntary. Then the teacher told the participants how to complete the questionnaire. The confidentiality and anonymity of their responses were ensured. The students completed the questionnaire during their regular writing class time.
Instruments
PBW self-efficacy scale (PBWSS)
PBW involves three main activities: literature search and reading, academic research, and writing, and therefore the PBW self-efficacy scale (PBWSS) includes the three parts accordingly. The writing self-efficacy items were adapted from the writing self-efficacy questionnaire (WSQ) (Pajares, 2007). The questionnaire was modified in order to tap into PBW ability, and some items were reworded to make them clearer and more straightforward for the respondents. Literature searching self-efficacy was measured by three questions from the aspect of searching, reading, and synthesizing. Academic researching self-efficacy is measured from problem awareness, research method designing, data collecting, data analysing, and data presenting ability. The final version of the modified scale consisted of eleven items: six items from Pajares (2007)’s WSQ scale, five items from the Pintrich and de Groot (1990), and one item developed by us. After the items were selected, the scale was further verified by researchers specializing in L2 education, with two of them having more than fifteen years of experience in second-language teaching. The scale was tested through a pilot study to collect feedback about item wording, testing administration, and reliability of the questionnaire. The final PBWSS was a six-point Likert scale rating from 1 (not very true of me at all) to 6 (very true of me). Participants were asked to indicate the extent to which they thought the statement was consistent with their behaviour in the learning process. The Cronbach’s alpha reliability index for this scale was 0.85. The factor loading of all items were above 0.60, and the Cronbach’s alpha of the subscales were all above the .70 threshold value based on the sample size, corroborating the internal consistency reliability (DeVellis, 2012) (see
PBW self-efficacy scale of eleven items
| Factors | Items | Cronbach’s alpha | Factor loading |
|---|---|---|---|
| Literature search and read self-efficacy | 1. I can find academic literature (in English or in Chinese). | 0.78 | 0.60 |
| 2. I can understand the academic resources in English. | 0.75 | ||
| 3. I can synthesize academic resources. | 0.80 | ||
| Research self-efficacy | 4. I can find problems that need to be studied. | 0.791 | 0.60 |
| 5. I can design reasonable research methods. | 0.64 | ||
| 6. I can analyse research data. | 0.66 | ||
| 7. I can support my opinion with collected data. | 0.91 | ||
| Language self-efficacy | 8. I can write complex sentences. | 0.846 | 0.71 |
| 9. I can use proper terms used in writing. | 0.89 | ||
| 10. I can write articles with clear discourse structure. | 0.74 | ||
| 11. I can adjust and use proper language according to the aimed readers. | 0.66 |
PBW strategy scale (PBWSS)
Respondents’ writing strategies were drawn from a writing strategy questionnaire developed by Teng and Zhang (2016). The writing strategy scale in this study contained twenty-two items in two categories: cognitive strategies (idea planning strategies, text processing, and revising strategies), and metacognitive strategies (goal management and task-planning strategies). The overall reliability of the whole scale as determined by Cronbach’s alpha was 0.91.
PBWSS of twenty-two items
| Cognitive strategy (14 items) | Idea planning strategies (7 items) |
| Text processing (3 items) |
|
| Revising strategies (4 items) |
|
| Metacognitive strategy (8 items) | Goal-oriented strategies (5 items) |
| Task-planning strategies (3-items) |
Academic writing scoring
Students’ writing performance was measured by their written work after a whole term of researching and writing. Students chose the problem to explore in small groups and wrote their reports individually. As PBW is widely practised in Chinese universities, especially in Shanghai, in order to prepare students for EMI (English mediated instruction) courses or learning abroad, PBW can be regarded as a typical academic writing assessment format. The topics were freely chosen by students on which social and economic problems they considered to be worth exploring, such as the factors influencing managing behaviour or the prospect of E-commerce business. Students searched the library or online for sources, collected their data, and wrote their reports.
The overall quality of the reports was scored using Good et al.’s (2012) discipline-specific writing rubric. These analytical writing criteria measure five aspects of writing performance: focus, content, organization, style, and English language convention. Two raters in applied linguistics attended a training workshop to standardize their understanding of the writing rubric. Inter-rater reliability and intra-rater reliability were at acceptable levels: .80 and.90 respectively. In total, 107 reports were collected, with an average score of 21.58 (SD = 1.58) out of 25.
Statistic analysis
We took two steps to apply a structural equation modelling (SEM) analysis in the study. It is believed that SEM analysis can identify sources of the misfit of data models and prevent problems of nonconvergence (Mueller & Hancock, 2008). Firstly, we evaluated the measurement models of the two questionnaires. The reliability and validity of the two questionnaires were checked with CFAs (confirmatory factor analyses) using software, i.e., AMOS program Version 22.0 (Arbuckle, 2013). This was designed to check whether the data we collected fit the hypothesized measurement models. Informed by the CFA results, we proposed a mediation model and evaluated it by SEM. Several goodness of fit indices was reported in order to check model fits, including the ratio of chi-square (χ2) divided by the degree of freedom (df), the comparative fit index (CFI), the Tucker-Lewis index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). We applied Hu and Bentler’s (1999) cut-off values and found that CFI and TLI were lower than .95; RMSEA lower than .06, and SRMR no larger than .08, showing a good model fit. In addition, chi-square/df ratio values lower than 2.5 were considered in our study to represent an acceptable model (Byrne, 1991).
Results
CFA model fit results
CFA model fit results
| Model | x2 | df | X2/df | p | RMSEA | SRMR | CFI | TLI |
|---|---|---|---|---|---|---|---|---|
| Model 1a | 12.432 | 12 | 1.036 | .412 | .018 (.000, .102) | .037 | 0.997 | .999 |
| Model 1b | .000 | 0 | .000 | .000 | .000 (.000, .000) | .000 | 1.000 | 1.000 |
| Model 1c | .158 | 2 | .079 | .924 | .000 (.000, .064) | .005 | 1.000 | 1.026 |
| Model 1 | .000 | 0 | .000 | .000 | .000 (.000, .000) | .000 | 1.000 | 1.000 |
| Model 2a | .679 | 3 | .226 | .878 | .000 (.000, .080) | .015 | 1.000 | 1.000 |
| Model 2b | .000 | 0 | .000 | .000 | .000 (.000, .000) | .000 | 1.000 | 1.000 |
| Model 2 | .000 | 0 | .000 | .000 | .000 (.000, .000) | .000 | 1.000 | 1.000 |
| Model 3a | 1.642 | 4 | 1.607 | .169 | .076 (.000, .179) | .029 | .987 | .973 |
| Model 3b | 4.531 | 1 | 1.642 | .200 | .078 (.000, .284) | .018 | .996 | .977 |
| Model 3c | .000 | 2 | 2.265 | .104 | .009 (.000, .018) | .000 | .098 | .939 |
| Model 3 | .000 | 0 | .000 | .000 | .040 (.000, .099) | .000 | 1.000 | 1.000 |
Note: Model 1a = idea planning cognition; Model 1b = text processing; Model 1c = text revising; Model 1 = cognitive strategy; Model 2a = goal oriented strategy ; Model 2b= task planning ; Model 2=meta cognitive strategy; Model 3a=linguistic self-efficacy; Model 3b=literature self-efficacy; Model 3c=research self-efficacy; Model 3=self-efficacy mean strategy
Descriptive statistics
Descriptive statistics
| V1 | V2 | V3 | V4 | V5 | V6 | V7 | V8 | V9 | |
|---|---|---|---|---|---|---|---|---|---|
| V1. | 1 | .504** | .658** | .756** | .558** | .704** | .290* | .312** | .231* |
| V2. | 1 | .637** | .460** | .228* | .430** | .147 | .259** | .205* | |
| V3. | 1 | .559** | .399** | .652** | .229* | .203* | .308** | ||
| V4. | 1 | .524** | .649** | .302** | .376** | .220* | |||
| V5. | 1 | .650** | .328** | .621** | .026 | ||||
| V6. | 1 | .285** | .253** | .116* | |||||
| V7. | 1 | 1 | .746** | .140* | |||||
| V8. | 1 | .203* | |||||||
| V9. | 1 | ||||||||
| Mean | 5.4 | 5.22 | 5.35 | 5.10 | 5.19 | 5.23 | 3.99 | 4.04 | 21.37 |
| SD | 0.8 | 1.04 | 0.98 | 1.05 | 0.99 | 0.93 | 0.82 | 0.71 | 1.58 |
Note: V1 = idea planning; V2 = text; V3 = Text revising; V4 = Task planning; V5 = Goal managing; V6 = Language self-efficacy; V7 = Literature self-efficacy; V8 = Research self-efficacy; V9 = Writing score.
Mediation model evaluation
The next step was to specify an SEM to evaluate how self-efficacy, being mediated by writing strategies, affected academic writing performance. A critical condition for testing the mediation effect is that there should be significant relationships between predictors and dependent variables (Hayes, 2013). Therefore, we checked the correlations of these variables, including self-efficacy, writing strategies, and academic writing performance.
A partial mediation model with both mediation and direct paths from self-efficacy to writing performance SE = self-efficacy; CS = cognitive strategy; MCS = metacognitive strategy; AW = academic writing; LaS = language self-efficacy; RS = research self-efficacy; Lis = literature searching and reading self-efficacy; IP = idea planning; TeP = text processing; TeR = text revising; GM = goal managing; Tap = task planning; AW = academic writing
The SEM results revealed an unacceptable model fit, and we modified the model by adding covariance between a pair of cognitive strategies (related to text processing and revising), and adding covariances between metacognitive strategies to latent cognitive strategy considering goal managing, task planning activities, and cognitive activities happening at the same time. The modified model was acceptable with χ2 (20) = 23.636; χ2/df = 1.18; p = 0.259>0.05; CFI = .99; TLI = .98; RMSEA 90% confidence interval = .04 [.00,.09]; SRMR = .06. On this basis, we retained the partial mediation structure of self-efficacy on writing performance.
A partial mediation model with both mediation and direct paths from self-efficacy to writing performance SE = self-efficacy; CS = cognitive strategy; MCS = metacognitive strategy; AW = academic writing; LaS = language self-efficacy; RS = research self-efficacy; Lis = literature searching and reading self-efficacy; IP = Idea planning; TeP = Text Processing; TeR = text revising; GM = goal managing; Tap = task planning; AW = academic writing. All path coefficients are standardized. * p > .05, ** p > .01.
To test the significance of the mediator effects, we used a bootstrapping procedure with bias-corrected confidence estimates. In the study, 95% confidence interval of the indirect effects was obtained using 2,000 bootstrap resamples (Preacher & Hayes, 2008). We evaluated the following mediation paths: self-efficacy ➔ cognitive strategies ➔ writing performance. We found there was a significant mediator effect (β= .11, p > .006); SE = .133; Z = 2.21; lower bounds= .084; upper bounds = .618.
To summarize, in this study, self-efficacy had significant strong positive effects on both cognitive and metacognitive strategies. In addition, self-efficacy had a partial mediation effect on writing performance through cognitive strategies.
Discussion
This study aimed to examine the effect of self-efficacy on cognitive and metacognitive writing strategies, and the effect of self-efficacy on PBW performance mediated by cognitive and metacognitive strategies.
Writing self-efficacy and self-regulated strategy use
To address the first question, our findings show that self-efficacy strongly influences both cognitive and metacognitive strategies in PBW activities, which is consistent with previous studies (Bandura, 1986; McCann & Garcia, 1999; Bai, 2018). According to social cognitive theory, students have cognitive abilities to self-organize, self-reflect, and self-regulate according to the changes in the learning tasks and set their own goals (Wang & Bai, 2016). Pajares (2007) found positive correlations between self-efficacy, cognitive and metacognitive strategies use, and learning persistence. He also stated that self-efficacy is part of the metacognitive knowledge and belongs to the most prominent part of the self-regulation process. In the present study, more self-efficacious students participated more actively in the group discussion in the beginning and more frequently regulated their plans and turned to teachers and peers to find help when coming across difficulty in searching the literature. They tried to exhaust traditional and non-traditional research resources, more frequently used effective cognitive writing strategies, and were better at consciously reflecting on their project completion and writing process. This highly effective self-regulation process is of great value to the research-based PBW tasks.
Writing self-efficacy, self-regulated strategies and PBW
Regarding the second question, our findings show that self-efficacy directly influences academic writing and indirectly influences academic writing through cognitive strategies. The result basically agrees with other researches. Extensive research has shown that writing self-efficacy has a significant predictive effect on both first-language writing (Bruning et al., 2013; Pajares et al., 1999; Pajares & Valiante, 1997; Zimmerman & Risemberg, 1997) and second-language writing (Sun & Wang, 2020; Teng & Zhang, 2020; Woodrow, 2011; Zabihi, 2018). The present study’s results confirmed the findings of previous research on PBW tasks. It can be explained that more successful self-efficacious students tend to dedicate more attention, effort, and time to the writing task, attach more value to the writing tasks they are involved in (Bandura, 1986; Pajares, 2003), and are actively committed to making different levels of revisions (Chen & Zhang, 2019). In contrast, less successful self-efficacious students are often reluctant to take risks in high-level revisions, which are more challenging in comparison with surface-level text revisions (Kormos, 2012).
Meanwhile, it is not difficult to figure out how writing self-efficacy indirectly affects academic writing through cognitive strategies. Through answering the first question - writing self-efficacy directly influences learners’ cognitive strategy use - it is natural to conclude that cognitive strategy use influences writing performance because strategy use always serves a certain learning or performing task. Cognitive strategies such as idea planning (e.g., I search literature to get research inspiration before writing), text processing (e.g., I refer to text books for linguistic and discourse help when writing) and revising strategies (e.g., I refer to text books for rhetorical and linguistic help when writing) work through the whole academic writing process, and affect the final writing performance. It is found that such cognitive strategies involved in ‘before writing’, ‘in writing’, and ‘after writing’, such as reviewing records, seeking opportunities, and self-evaluation strategies, significantly predicted learners’ writing performance (Sun & Wang, 2020). According to Roca de Larios et al. (2008), high-achieving students invested more time in planning and revising than in formulating.
Unexpectedly, our findings showed that metacognitive strategies played a different role compared with previous theories, as Yang and Plakans (2012) and Yang (2015) have shown that metacognitive strategies (evaluation and planning) assume an indirect predictive role in L2 writing performance, so the indirect influence of self-efficacy on writing performance via metacognitive strategy use could be predicted. However, in the present study, a non-significant negative influence (-0.19) is displayed. The reasons may be as follows: first, the categorization of metacognitive and cognitive strategies adopted from Teng et al. (2018) in the present study is unclear to some extent. The confusion about categorization is not difficult to be explained - metacognitive and cognitive strategies are hardly differentiated in complicated academic tasks - because the complexity of academic writing requires the continuous manipulation of cognitive strategies to carry out metacognitive activities, such as evaluating one’s own writing paper in order to revise it. Secondly, PBW tasks place on students a comparatively heavy cognition load (Ye, 2020) and the participants cognitive resources, i.e., long-term memory, process of reading, and working memory (Chenoweth & Hayes, 2001) were already taxed. Therefore, it is understandable that these first-year non-English majors found it difficult to use metacognitive strategies (e.g., setting goals, monitoring while writing, maintaining or improving motivation and reflecting on progress) to make up for insufficiency of discipline knowledge and cognitive writing strategies in PBW.
Limitations and implications
The present study contributes to cognitive process theory and social cognitive theory of writing by providing further evidence from a PBW task in university-level English in China. The non-significant negative relations between metacognitive strategy use and writing performance call for further studies on the effectiveness of metacognitive strategies between different proficient groups and on the relationships between cognition load of writing tasks and the effectiveness of metacognitive strategies.
The study has implied that PBW tasks might not be appropriate for the beginning stage of college students because of their high cognitive demands. In fact, PBW tasks are designed to teach the third-year and fourth-year college students in Waseda University of Japan and some other universities around the world (Wang, 2019). In addition, more systematic PBL course development and arrangements for academic writing programmess need to be widely explored and administered in EAP teaching of Chinese higher education.
This study also has other pedagogical implications. In Educational Psychology, Bohlin et al. (2018) cited Erikson (1980)’s finding that students’ SE will increase with encouragement from teachers and successful examples from peers, and in turn, their diligence and efforts will be greatly exerted to achieve their own goals. Similarly, based on the present study, scaffolding and encouragement from teachers and peers (Ong, 2013; 2014; Ong & Zhang, 2010; Parks & Raymond, 2004) are suggested to help improve students’ SE in PBW courses. Since Ching (2002) argued that instruction in SRS increases ESL students’ SE beliefs, we have reason to propose that teachers should give students writing strategy instruction and practice (self-regulation of metacognitive strategy) to improve their PBW performance, especially for the students from lower SEC (social standing and economic characteristics) at high-school and university levels (Ren & Xin, 2013). These students especially require help and encouragement from teachers, and they can also acquire more confidence from their peers’ support and from good examples.