System 95 (2020) 102366

Contents lists available at ScienceDirect

System
journal homepage: www.elsevier.com/locate/system

Relationship between self-efficacy and language proficiency:

A meta-analysis
Chuang Wang a, *, Ting Sun b
a
Faculty of Education, University of Macau, Avenida da Universidade, Taipa, Macau, China
b
Cato College of Education, University of North Carolina at Charlotte, 9201 University City Blvd., Charlotte, NC, 28223, USA

a r t i c l e i n f o a b s t r a c t

Article history: This study aims to estimate the average effect size of the relationship between self-efficacy
Received 14 May 2020 and language proficiency and to examine factors that moderate the relationship. A meta-
Received in revised form 17 August 2020 analysis of the 493 effect sizes for the relationship between language learners’ self-efficacy
Accepted 11 September 2020
beliefs and their language proficiency from 74 published journal articles, book chapters,
Available online 16 September 2020
and dissertations revealed an average of small to medium effect size. The use of hierar-
chical linear models indicated 76.7% of the variance between the published works, sug-
Keywords:
gesting that the effect size varies across studies. Effect sizes reported in studies with East
Meta-analysis
Self-efficacy
Asian students were larger than those reported in studies with students in Western cul-
English language learners ture. Parameter estimates of fixed effects noted that the effect sizes reported from zero-
Language proficiency order correlations inflated the effect size without consideration of the language, culture,
grade-level, sample size, and data analysis procedures. This study contributes to the
literature by summarizing studies from the last two decades in a systematic way and
cautions researchers when interpreting effect sizes reported in the literature.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction

English is the official language of more than 80 countries and is spoken as the first language by approximately 380 million
people all over the world (Duffin, 2019). As a result, it becomes a “world language” and occupies a central position in in-
ternational business, economic, political and cultural communications (McKay, 2002). In China, for example, English language
proficiency serves as a threshold for job application. In addition, it plays a significant role in the college admission process in
the United States and other English-speaking countries.
Many factors impact English learning processes, such as motivation (Zhang & Guo, 2012), strategy (Bai, 2018), and
cognition (Swanson et al., 2016). As one of the cognitive and motivational constructs, self-efficacy has gained increasing
attention in education (Usher & Pajares, 2008). Self-efficacy was defined as “beliefs in one’s capabilities to organize and
execute the course of action required to manage prospective situations” (Bandura, 1997, p. 2) and was believed to have a
facilitative influence on human agency (Bandura, 1986). According to the social cognitive theory, human attainments depend
on the reciprocal interactions between personal, behavior, and environmental actions (Bandura, 1986). Therefore, the beliefs
people hold about their abilities determine their behavior and cognition (Bandura, 1997). In the academic setting, efficacious
students usually have higher motivation (Zhang & Guo, 2012), experience less apprehension (Woodrow, 2011), and thus
perform better in measurements of learning outcomes (Grigg et al., 2018; Komarraju & Nadler, 2013).

* Corresponding author.
E-mail addresses: [email protected] (C. Wang), [email protected] (T. Sun).

https://doi.org/10.1016/j.system.2020.102366
0346-251X/© 2020 Elsevier Ltd. All rights reserved.
C. Wang and T. Sun System 95 (2020) 102366

Numerous studies have linked student academic achievement to self-efficacy beliefs (Diseth, 2011; Komarraju & Nadler,
2013; Pajares & Graham, 1999; Zimmerman & Kitsantas, 2005). The significant relationship between self-efficacy beliefs and
English language proficiency was also found for English language learners in many countries such as Norway (Diseth, 2011),
South Korea (Kim et al., 2015), Singapore (Liem et al., 2008), Botswana (Magogwe & Oliver, 2007), Vietnam (Truong & Wang,
2019), and China (Wang & Bai, 2017). However, the magnitude of the relationship between self-efficacy and English language
proficiency varies across tasks and contexts (Bong, 2002; Klassen, 2004). Some studies failed to find a statistically significant
relationship between self-efficacy beliefs and language proficiency when previous performance was controlled (e.g., Chea,
2012). The discrepancies may be the outcomes of various research designs and statistical data analysis procedures. There-
fore, a meta-analysis of previous studies is necessary to summarize the findings in a scientific method by considering the
methodology, covariates, and contexts. The purpose of this study is to synthesize the findings from studies for the last two
decades and to give directions for future studies.

2. Literature review

2.1. Definition of self-efficacy

Self-efficacy was defined as one’s subjective beliefs of how well one can perform a specific task with a judgment of his/her
own competence (Pajares, 1996). Since Bandura’s seminal paper on self-efficacy (Bandura, 1977), self-efficacy has become one
of the most studied topics in psychology. As Bandura and other researchers have discussed, self-efficacy can have an impact
on a person’s psychological states, behaviors, and motivation (Bandura, 1997). Self-efficacy can influence a person’s behavior,
either positively or negatively, based on one’s perception of his/her abilities as related to a particular task. It influences the
choices that a person makes, the effort that he/she puts forth, and his/her persistence when facing obstacles and failure
(Usher & Pajares, 2008).
Bandura (1997) posited that there were four sources of self-efficacydmastery or enactive experience, vicarious experi-
ence, social persuasion, and physiological and emotional states. Mastery or enactive experience refers to the idea that pre-
vious experiences exercise facilitative or undermining influence on one’s self-efficacy (Bandura, 1997). Previous successes
raise one’s self-efficacy while failures lower it. Self-efficacy is also affected by observations of others doing the same task,
which is known as vicarious experience (Bandura, 1997). Vicarious experience provides students with opportunities to
observe and compare themselves to high-achieving students. This creates an environment where “competent models
transmit knowledge and teach observers effective skills and strategies” (Bandura, 1997, p. 88), through which both self-
efficacy and competence would be elevated. Social persuasion also plays a role in the development of one’s self-efficacy
(Bandura, 1997). Encouragement and positive feedback would strengthen one’s self-efficacy, while punishment and nega-
tive comments would weaken one’s self-efficacy (Bandura, 1997). This contention was confirmed by a study conducted by
Jalaluddin et al. (2010) who contended that teacher’s feedback and assistance would influence students’ self-efficacy in
writing. More specifically, Bong and Skaalvic (2003) examined the influence of various types of teacher feedback on students’
self-efficacy beliefs and concluded that only constructive feedback had a positive impact. This positive influence was effective
only when the source of information was reputable and reliable. Therefore, verbal encouragement needs to be realistic so that
students’ self-efficacy can be promoted. Students’ physiological and emotional states also exert influence on the development
of self-efficacy (Bandura, 1997). For example, apprehension was evidenced to be negatively correlated with self-efficacy
(Woodrow, 2011). Therefore, self-efficacy could be enhanced by reducing students’ aversive arousal such as apprehension
or by increasing their positive sentiments.

2.2. Self-efficacy and academic outcomes

A large body of research has been conducted on the relationship between self-efficacy and academic achievement in
diverse disciplines (Diseth, 2011; Grigg et al., 2018; Komarraju & Nadler, 2013; Pajares & Graham, 1999; Wilson & Narayan,
2016; Zimmerman & Kitsantas, 2005). For example, self-efficacy was found to be a predictor of academic achievement among
students in computer science, social psychology (Wilson & Narayan, 2016), and mathematics (Grigg et al., 2018; Pajares &
Graham, 1999). Among a group of motivation variables (i.e., intrinsic motivation, extrinsic motivation, task value, control
learning, self-efficacy, and test anxiety), self-efficacy was the only variable that significantly predicted GPA among college
students (Komarraju & Nadler, 2013). Moreover, the significant correlation between self-efficacy and GPA was partially
mediated by effort regulation (Komarraju & Nadler, 2013). Self-efficacious students were more likely to exert more effort
regulation and display more self-disciplined behavior, which would lead to a higher GPA. However, some studies reported
mixed results in terms of the relationship between self-efficacy and academic success. For example, Galla et al. (2014)
conducted a longitudinal study with 135 elementary students and noted predictive effects of self-efficacy on reading
scores, but not on mathematics scores. This non-significant result may be attributed to the use of standardized tests rather
than GPA as the mathematics outcome measure, which is supposed to be more closely associated with students’ efforts and
self-discipline (Galla et al., 2014).
Self-efficacy does not only have a predictive effect, but also plays a mediational role in the relationship between students’
learning outcomes and other factors such as academic attitudes (Bandura, 1997), prior academic achievement (Diseth, 2011;
Wilson & Narayan, 2016; Zimmerman & Kitsantas, 2005), and quality of homework (Zimmerman & Kitsantas, 2005).

2
C. Wang and T. Sun System 95 (2020) 102366

2.3. Self-efficacy and language proficiency

Self-efficacy is a significant predictor of academic attainments in general (Diseth, 2011; Komarraju & Nadler, 2013; Pajares
& Graham, 1999; Wilson & Narayan, 2016; Zimmerman & Kitsantas, 2005) and language proficiency in particular (Bruning
et al., 2013; Nevill, 2008; Sanders-Reio et al., 2014). Nevill (2008) found that self-efficacy was predictive of reading
achievement in a sample of 84 fourth, fifth, and sixth graders in a rural school district in North Central Pennsylvania. The
predictive effect of self-efficacy on language learning was also replicated with college students (Prat-Sala & Redford, 2012). In
addition, writing self-efficacy and reading self-efficacy contributed independently to writing performance for college stu-
dents (Prat-Sala & Redford, 2012). However, writing self-efficacy and self-efficacy for self-regulation were not significantly
related to writing proficiency in Sanders-Reio et al.’s (2014) study.
Self-efficacy was not only positively correlated with language achievement, but also positively correlated with other
constructs such as academic self-concept, self-efficacy for self-regulation, achievement goals, value, task goal orientation, and
performance-approach goals. Conversely, self-efficacy negatively correlated with apprehension and performance-avoidance
goals (Pajares & Valiante, 2001). Self-efficacy was also correlated with learning beliefs (Sanders-Reio et al., 2014) and the use
of language learning strategies (Magogwe & Oliver, 2007).

2.4. Self-efficacy and English proficiency in the EFL context

Self-efficacy was not only studied extensively in the context of English as a first language (e.g., Pajares, 1996; Zimmerman
& Kitsantas, 2005) but is also well documented in the English as a foreign language (EFL) context (e.g., Bai et al., 2019; Bai &
Wang, 2020; Wang et al., 2013; Wang & Pape, 2007; Woodrow, 2011; Zhang & Guo, 2012). We conducted a series of studies of
self-efficacy in the EFL context, particularly in Korea, China, and Germany (e.g., Huang et al., 2015; Kim et al., 2015; Wang &
Bai, 2017; Wang et al., 2012; Wang et al., 2013). Although EFL students generally had moderate levels of self-efficacy beliefs
(Wang et al., 2012, 2013), it is still an attribute conducive to English language learning (Huang et al., 2015; Kim et al., 2015). For
example, a recent study examined social support, self-efficacy, and English language achievement among 1092 EFL Chinese
students and found that self-efficacy was positively correlated with English learning outcomes (Bai et al., 2019). Self-efficacy
was also correlated with learning strategies (Kim et al., 2015; Wang et al., 2012). A study with college students in Korea noted
that students with high levels of self-efficacy had more frequent use of learning strategies in comparison to students with low
levels of self-efficacy and that the strategies used by efficacious students were more diverse (Kim et al., 2015).
Self-efficacy was found to be predictive of reading (Amogne, 2008; Shang, 2010), writing (Amogne, 2008; Sun & Wang,
2020; Woodrow, 2011; Zhang & Guo, 2012), listening (Rahimi & Abedini, 2009), and speaking (Sarshar & Oroji, 2016;
Zhang et al., 2020), respectively. General English learning competence refers to an individual’s overall English proficiency,
which can be measured by English skills in specific domains, including listening, speaking, reading, and writing. A study with
80 EFL college students in Iran noted a significant and strong relationship between self-efficacy beliefs and students’ reading
comprehension scores (Naseri, 2012). The predictive effect of self-efficacy on English achievement was also replicated in
writing (Amogne, 2008; Woodrow, 2006; Zhang & Guo, 2012). A study with Chinese EFL college students contended that self-
efficacy not only played predictive roles in writing outcomes, but also played mediational roles in the relationship between
writing performance and anxiety (Woodrow, 2011).
The contention of the positive relationship between self-efficacy and language learning was, nevertheless, rejected by
some studies under certain circumstances. For example, Corkett et al. (2011) noted non-significant relationships between
sixth graders’ literacy ability (i.e., reading and writing abilities) and their self-efficacy. The researchers indicated that this non-
significant result might be due to the inaccurate perceptions of self-efficacy rated by young students because “students use
their personal accomplishments to gauge their ability … this positive relationship increases in accuracy as student mature”
(Corkett et al., 2011, p. 67). Since students’ ages tend to affect the relationship between self-efficacy and their language ability,
we can assume this can be a factor moderating the effect sizes.

2.5. Potential moderators

Literature suggests that effect size tends to be moderated by study characteristics and sample characteristics. Publication
type was found to have moderating effects on effect size estimates regardless of research topics (Borenstein et al., 2009). For
example, Polanin et al. (2016) conducted a meta-analysis and found published studies tended to report larger effect sizes than
unpublished studies. Educational setting was found to be another moderator. Results from another meta-analysis revealed a
large effect size of the relationship between self-efficacy beliefs and academic outcomes with college students (Multon et al.,
1991). Bandura (1986) contended that self-efficacy is task- and domain-specific. It is also reasonable to speculate variations of
self-efficacy and the relationship between self-efficacy and language proficiency across languages and domains of language
learning.
Effect size tends to vary across different cultural contexts as well. Studies examining self-efficacy across nations indicated
that Asian students reported lower self-efficacy beliefs despite outperforming their counterparts in academic achievement
(Earley, 1999; Eaton & Dembo, 1997; Luszczynska et al., 2005; Scholz et al., 2002; Yan & Gaier, 1994). Researchers have
postulated that attributions may account for differences in perceived self-efficacy because effort and hard work are more
highly valued than ability in collectivistic cultures such as in China, Korea, and Japan; as a result, self-efficacy tends to be

3
C. Wang and T. Sun System 95 (2020) 102366

reported lower in collective cultures than in individualistic cultures (Eaton & Dembo, 1997). Literature also suggests that
cultural context moderates the effect size of the relationship between self-efficacy and language proficiency with effect sizes
in studies conducted in Asian countries being larger than those in the Western countries (Wang et al., 2013). Wang et al.
(2013) reported an effect size of 0.41 for the relationship between self-efficacy and English proficiency among Chinese
students and an effect size of .20 among German students.
Although most studies found significant relationships between self-efficacy and language proficiency, the strength of their
relationships was not consistent (e.g., r2 ¼.063 in Williams & Takaku, 2011; r2 ¼.163 in Bruning et al., 2013). Literature suggests
that this variation is moderated by study characteristics and students’ characteristics (e.g., Multon et al., 1991; Polanin et al.,
2016). No meta-analytic studies, however, have been conducted to investigate the magnitude of the relationship or to
quantify the heterogeneity. Therefore, it is warranted to conduct a meta-analysis to estimate the true effect size and to
examine factors moderating the effect size. The research questions that guided this study are:

1. What is the average effect size of the relationship between self-efficacy beliefs and language proficiency?
2. To what extent is the effect size of the relationship between language proficiency and self-efficacy moderated by language,
language domain, analytical procedure, publication type, publication period, learner level, and cultural context?

3. Method

3.1. Literature Search and screening

Both electronic searches and manual searches were employed to identify studies related to self-efficacy and language
proficiency since 1995. Electronic articles were located through computerized journal databases (e.g., PsychINFO, Google
Scholar, ERIC, ProQuest, & JSTOR) and Web of Science databases, using combinations of the following keywords: self-efficacy,
language acquisition, English language learners, English proficiency, and language proficiency. Our inclusion criteria in the
searching process included: (1) peer-reviewed journal articles or dissertations/theses; (2) published in the time frame of
1995e2019; and (3) being written in English. We limited our search to peer-viewed journal articles or dissertations/theses to
ensure the quality of primary studies. Only studies written in English were searched, which was constrained by the language
competence of members in our research team. In manual searches, the following journals were targeted: American Educa-
tional Research Journal, Journal of Education Psychology, Educational Psychology Review, Contemporary Educational Psychology,
TESOL Quarterly, Studies in Second Language Acquisition, Language Learning, Journal of Second Language Writing, Second Lan-
guage Research, and System. The two searches produced 195 documents for screening.
PRISMA 2009 (Moher et al., 2009) checklist was followed in the screening process, and four inclusion criteria were applied:
(1) relevancy; (2) empirical quantitative studies; (3) being in a formal educational setting; and (4) reporting effect sizes or had
sufficient statistics to compute effect sizes. Relevancy means that the document discussed one of the following topics: the
relationships between self-efficacy and language proficiency, the effects of self-efficacy on language proficiency, self-efficacy
differences as a function of language proficiency levels, and language proficiency differences as a function of self-efficacy
profiles. There were 117 documents remaining after applying the first criterion. The second criterion required the docu-
ment to be an empirical and quantitative study, so conceptual studies and qualitative studies were excluded, the application
of which resulted in 89 documents. The third criterion was used to determine if the document was in a formal educational
setting, so studies employing workshops or language trainings were excluded. Applying this criterion yielded 79 documents
that reported relevant statistics. The fourth criterion was to determine whether the document reported an effect size or had
adequate statistics to compute an effect size related to the relationships between self-efficacy and language proficiency.
Finally, we ended up with 74 documents and 493 effect sizes. The screening process was presented in Fig. 1.

3.2. Data coding

Two coding forms were developed by the lead author, addressing both the sample characteristics and the statistical
characteristics of all the documents. The first coding form included Author, Independent Variable, Dependent Variable, r
Correlation (r), Unstandardized Beta (B), Beta Standardized (b), Standard Error (SE), Sample Size, Sample Description, Domain,
Method, Dependent Variable Mean (M), Dependent Variable Standard Deviation (SD), Independent Variable Mean (M), In-
dependent Variable Standard Deviation (SD), Effect Size, and Type of Effect Size. The second coding form is the Article-Level
Coding Form, including Author, Publication Type, Publication Date, Grade Level, Language, and Country. The variables
included in our final model were coded as follows.
Domain included the domain of overall language learning and the domains of listening, speaking, reading, and writing,
respectively. As a U.S.ebased collaborative group of 40 member states, territories, and federal agencies, WIDA provides
language development resources to those who support the academic success of multilingual learners. According to the WIDA
Standard Framework, language learning can be classified into the categories of receptive (listening and reading) and pro-
ductive (speaking and writing) learning (WIDA Consortium, 2014). Therefore, listening and reading were merged into one
category (receptive skills) and speaking and writing were merged into the other category (productive skills). As a result,
domains of language learning outcomes were classified into three categories: overall, receptive, and productive skills. Overall

4
C. Wang and T. Sun System 95 (2020) 102366

Fig. 1. Prisma flow diagram of literature search and screening.

refers to the overall language proficiency without examining a specific domain. Two discrete variables (i.e., receptive and
productive) were dummy coded to represent receptive and productive skills, respectively, with 0 for overall and 1 for the
respective variable.
Methods refers to the statistical methods used in the documents, which were classified into three categories: mean dif-
ferences, regression, and correlation. Mean differences included the methods such as t-test, analysis of variance (ANOVA),
multivariate analysis of variance (MANOVA), and analysis of covariance (ANCOVA). Regression consisted of simple linear
regression, hierarchical linear regression, path analysis, and structural equation models. Correlation included Pearson cor-
relation, Spearman correlation, and canonical correlation. Two variables were dummy-coded to represent the methods of
mean differences and regression with 0 for correlation and 1 for mean differences and regression, respectively.
Language was coded as 0 for English and 1 for any other languages. Publication Type was coded as 0 for journal article and 1
for dissertation/thesis. Publication Date was dummy coded as a new variable Publication Period with 0 representing the period
of the first decade (1995e2006) and 1 representing the second decade (2007e2019). Grade was coded as 0 for K-12 level and 1
for college level. Culture was classified into three categories based on Marsh et al.’ study (2015): Western Culture, East Asian
Culture, and Middle East Culture. Two discrete variables were used to represent East Asia and Middle East with 0 for Western
and 1 for East Asia and Middle East, respectively.
Of these 74 publications, 59 (79.7%) were journal articles and 15 (20.3%) were dissertations/theses (Table 1). There were 12
(16.2%) documents published or written during the period of 1995e2006, and 62 (83.8%) documents published or written
during the period of 2007e2019. The majority of documents (82.4%) studied the language of English, and the rest studied
languages including Arabic, French, German, Greek, Norwegian, Spanish, and Turkish. Both English as a first language and EFL

5
C. Wang and T. Sun System 95 (2020) 102366

Table 1
Frequency distribution of documents by language, publication type, publication period, grade, and culture.

Variables Frequency Percentage

Language English 61 82.4
Other 13 17.6
Publication Type Journal Article 59 79.7
Dissertation/Thesis 15 20.3
Publication Period 1995e2006 12 16.2
2007e2019 62 83.8
Grade K-12 20 27.0
College 54 73.0
Culture Western 34 45.9
East Asian 24 32.4
Middle East 16 21.6

contexts were included. There were 20 (27.0%) studies working with K-12 students and 54 (73.0%) studies with college
students. In addition, there were 34 (45.9%) studies conducted in the Western countries, 24 (32.4%) studies in East Asian
countries, and 16 (21.6%) in Middle Eastern countries.
Frequency distribution of effect size by the domain of language learning and data analysis methods were presented in
Table 2. There were 110 (22.3%) effect size measures related to overall language learning, 105 (21.3%) related to listening and
reading, and 278 (56.4%) related to speaking and writing. The majority of effect size measures (312; 63.3%) were calculated by
using correlation, followed by 89 (18.1%) by using mean differences, and 92 (18.7%) by using regression.
Effect sizes used in this study were all coefficient of determinations (r2). They were either directly reported or converted
from Pearson correlation coefficients, path analysis loadings, or structural equation model loadings.
Since various data analysis methods were employed in these documents, the effect size measures were reported or
computed in different forms, which included: Cohen’s d (d), Cohen’s f (f), eta-squared (h2 Þ, r-squared (r 2 ), R-squared (R2 Þ, rho-
squared ðr2 Þ, andw2 . Borenstein et al. (2009) provided formulas to convert among effect sizes, which makes it possible to
synthesize effect sizes on different metrics. The different effect size measures were all transformed into r 2 using the following
transformation formulas.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h2
f¼
1  h2

d ¼ 2*f

d2
r2 ¼
d2 þ4

3.3. Publication bias

Since published studies tend to report statistically significant results and are more likely to be included in meta-analytic
studies, publication bias always occurs. Therefore, the resultant documents may not be representative of the population of
completed documents, which may skew the results and mislead the interpretation of meta-analysis (Rothstein et al., 2005).
We addressed this problem by searching for and including both published (i.e., journal articles) and unpublished documents
(i.e., theses and dissertations). Furthermore, we conducted a test called Orwin’ Fail-Safe N (1983) to estimate the publication
bias, with the formula being presented as

N0 ðdo  dc Þ
Nfs ¼ ;
dc  dfs

where N0 is the number of effect sizes, do represents the mean of Cohen’s d from this meta-analysis, dc is the criterion level of
a small effect size (0.02), dfs is the fail-safe lower boundary (typically 0). The Orwin’s fail-safe N suggested a need for 575
additional studies to yield an insignificant result, suggesting no publication bias in the present study (Borenstein et al., 2009).

Table 2
Frequency distribution of effect size by domain and methods.

Variables Domain Methods

General Receptive Productive Correlation Differences Regression

Frequency 110 105 278 312 89 92
Percentage 22.3 21.3 56.4 63.3 18.1 18.7

6
C. Wang and T. Sun System 95 (2020) 102366

3.4. Data analysis procedure

Descriptive statistics were used to report the means and standard deviations of the effect sizes by language, domain of
language learning, statistical method, publication type, publication period, grade, and culture. Since effect sizes were nested
within studies, a multilevel data analysis procedure was employed to examine factors moderating the effect sizes (Research
Question Two) while distinguishing the variance between and within studies (Raudenbush & Bryk, 2002). An unconditional
model without entering any predictors was run first with the purpose of calculating the intra-class correlation coefficients (ICC)
and estimating the overall average effect size (Research Question One). ICC tells how much variance can be accounted for by the
between-study variance, which will provide or deny the legitimacy of conducting a two-level regression model. Second, a
random intercept model with all the predictors entered was run to check factors associated with the effect sizes. We chose a
random intercept model because we examined the level-2 predictors on effect size estimates (the intercept) and the associ-
ations between second-level factors and the slopes in the first level were assumed to be fixed. The HLM was represented as:
Level 1:
         
Effect Sizeij ¼ b0j þ b1j * Receptiveij þ b2j * Productiveij þ b3j * Sizeij þ b4j * Differenceij þ b5j * Regressionij þ rij

Level 2:
     
b0j ¼ g00 þ g01 * Languagej þ g02 * PublicationTypej þ g03 * ðPublicationPeriodÞ þ g04 * Collegej
   
þ g05 * Asianj þ g06 * MiddleEastj þ u0j

bij ¼ gi0 ; i ¼ 1; 2; 3; 4 5:

Level 1 variables represent characteristics of each effect size reported, whereas Level 2 variables represent characteristics
of each study. All variables at Level 1 were group-mean centered so that the intercept represents the average effect size and all
variables at Level 2 were grand-mean centered to adjust for the proportions of published works by language, publication type,
publication period, education level, and culture. Results with robust standard errors were chosen because these results were
robust against violations of assumptions. Pseudo R Squared value was used as the effect size of the hierarchical linear model
by considering the ratio between the reduction of level-1 variance and the original level-1 variance.

4. Results

4.1. Average effect size

Descriptive statistics about the effect sizes from all the studies included in this meta-analysis were reported in Table 3. The
effect sizes were reported in coefficient of determination (r2), so the mean values represented the percentage of variance of
English proficiency explained. In Table 3, the percentage of variance ranged from 0.10 to 0.18, which means 10% of the variance
of language proficiency was explained by self-efficacy beliefs when regression method was used and 18% of the variance was
explained when East Asian students were used as the sample. Specific interpretations of the results in Table 3 are presented in
the next paragraph.
In language domain, self-efficacy explained 15% of general language proficiency, 13% of receptive language (listening and
reading) proficiency, and 12% of productive language (speaking and writing) proficiency. As for the statistical method used in
the studies reviewed, mean comparisons reported 17%, regression reported 10%, and Pearson correlation reported 13% of the
variance in language proficiency explained by self-efficacy. Furthermore, studies with the English language, on average, re-
ported 12%, whereas studies with other languages reported 15% of the variance in language proficiency explained by self-
efficacy beliefs. Studies published in journals reported 13%, while studies published in dissertations/thesis reported 11% of
the same variance. Moreover, studies with K-12 students reported 11%, whereas studies with college students reported 15% of
the variance. Finally, studies with East Asian students reported 18%, studies with Middle Eastern students reported 13%, and
studies with Western Culture students reported 11% of the variance in language proficiency explained by self-efficacy beliefs.

4.2. Factors moderating the effect size

Intra-class correlation coefficient calculated from the unconditional model was 0.77 which means 77% of the variance in
effect sizes reported was between publications. Results from the conditional model with all the predictors of effect sizes were
presented in Table 4. Although sample size is not expected to have an influence on the effect size (Coladarci et al., 2014), this
study showed a statistically significant and positive relationship between sample size and effect size, t(411) ¼ 2.10, p < .05.
The use of statistical approaches was a statistically significant predictor for the estimated effect sizes of English general
language proficiency of U.S. K-12 students in publications of journal articles while sample size was controlled. The use of
linear regression, path analysis, and structural equation models statistically and significantly reduced the reported effect sizes
in comparison to the use of Pearson correlation method, t(411) ¼ 4.59, p < .001. The effect size reported by Pearson

7
C. Wang and T. Sun System 95 (2020) 102366

Table 3
Descriptive statistics for effect sizes by language, domain, method, publication type, publication period, grade, and culture.

Variables Categories M SD n
Language General 0.15 0.15 110
Domain Receptive 0.13 0.16 105
Productive 0.12 0.12 278
Method Mean Difference 0.17 0.16 89
Regression 0.10 0.11 92
Correlation 0.13 0.13 312
Language English 0.12 0.14 402
Other 0.15 0.13 91
Publication Type Journal Article 0.13 0.15 345
Dissertation/Thesis 0.11 0.11 148
Publication Period 1995e2006 0.13 0.13 141
2007e2019 0.13 0.14 352
Grade K-12 0.11 0.12 269
College 0.15 0.16 224
Culture East Asian 0.18 0.15 101
Middle East 0.13 0.20 63
Western Culture 0.11 0.11 329

correlation method was 0.07 higher than that reported by linear regressions or structural equation models although the effect
sizes reported by Pearson correlation method and the mean difference method were also statistically significantly different,
t(411) ¼ 2.08, p < .05. Moreover, the difference in effect sizes reported did not differ statistically with respect to the domains
of language learning outcomes: overall, receptive, and productive.
At the study level, the language in which the article was written, publication type (journal article, book chapter, disser-
tations), publication period, and the age of the participants (college versus secondary schools) were not statistically signif-
icantly related to the effect sizes reported (ps > .05). However, the participants’ culture did have a statistically significantly
impact on the effect sizes reported. Specifically, the effect size reported from the participants in East Asia and the Middle East
for the relationship between self-efficacy beliefs and English language proficiency was statistically and significantly higher
than that reported from the participants in the West (e.g., Europe and the United States). The variables included in the
conditional hierarchical linear model reduced the variance of the effect sizes by 18.80%.

5. Discussion

The small to medium effect sizes reported in Table 3 confirmed findings from previous studies about the positive rela-
tionship between self-efficacy beliefs and language proficiency (Prat-Sala & Redford, 2012; Wang et al., 2012). The Orwin’s
fail-safe N suggested a need of 575 additional studies to yield an insignificant result. Since Orwin’s Fail-Safe test is more
stringent, the results from the current study suggest no publication bias.
Intra-class correlation coefficient reported in this study was 0.77, which means 77% of the variance existed between
studies for the effect size of the relationship between self-efficacy beliefs and language proficiencies. This result justifies the
use of hierarchical linear models with the data and calls for more meta-analyses in this field. As discussed in the literature
(e.g., Bong, 2002; Klassen, 2004), the relationship between self-efficacy beliefs and language proficiency varies between
studies by the context of the study and the co-variates used in the statistical models. For example, results from this study
seemed to show discrepancies in the parameter estimates from Hierarchical Linear Models in Table 4. The estimated

Table 4
Parameter estimates for effect size.

Parameter Coefficient t df p
Effect Size
Intercept 0.13895 6.60 61 <.001
Receptive 0.01018 0.39 411 .695
Productive 0.00692 0.27 411 .786
Size 0.00009 2.10 411 .036
Difference 0.05615 2.08 411 .039
Regression 0.06686 4.59 411 <.001
Article Level
Language 0.00713 0.15 68 .882
Publication Type 0.01454 0.33 64 .746
Publication Period 0.08492 1.68 63 .099
College 0.03521 0.80 64 .425
East Asian 0.09311 2.05 67 .045
Middle East 0.13665 2.55 68 .013

Note. Publication Type is a dummy variable (1 ¼ journal; 0 ¼ thesis); Asian is a dummy variable (1 ¼ Asian, 0 ¼ Western); Middle East is a dummy variable
(1 ¼ Middle East, 0 ¼ Western).

8
C. Wang and T. Sun System 95 (2020) 102366

parameter for the slope of regression in HLM was 0.07, which means that the difference between the use of Pearson cor-
relation method and the use of regression method predicted a difference of 0.07 in the estimated effect size. As the outcome
variable in the HLM is effect size in the form of r-square, this difference means that using Pearson correlation method
overestimated the effect size (percentage of variance explained) by six percent. Multi-level models not only considered the
nested nature of individual effect sizes within multiple studies, but also controls the contextual factors such as grade-level,
country, language, publication type, receptive versus productive language, and sample size. The difference between
regression and Pearson correlation in the estimation of effect sizes is expected. Since Pearson correlation coefficients were bi-
variate relationships whereas the regression slopes or structural equation loadings represent the relationship with all the co-
variates controlled, Pearson correlation coefficients inflated the effect size by ignoring the contribution of other factors to
language proficiency. Therefore, researchers are recommended to rely more on the regression slopes or loadings from
structural equation models. Future researchers should be cautious that the magnitude of the effects of self-efficacy on English
language proficiency varies across contexts (Bong, 2002; Klassen, 2004) so factors in the cultural and societal context should
be included in the model when estimating the effect size of self-efficacy beliefs.
Although previous studies suggest that East Asian students tend to underestimate their self-efficacy beliefs in comparison
to students from the Western culture (e.g., Klassen, 2004; Wang et al., 2013), this study showed that the estimated effect size
for the relationship between language proficiency and self-efficacy beliefs with East Asian and Middle Eastern students were
larger than that with students from the Western culture. This finding suggests that self-efficacy plays a more prominent role
in Asian and Middle Eastern cultures than in Western culture. Since students in Asian cultures consistently reported lower
efficacy beliefs than their Western counterparts (Luszczynska et al., 2005; Scholz et al., 2002), it is more urgent for language
teachers in Asian countries to improve students’ self-efficacy beliefs.

6. Significance and conclusion

Although the use of hierarchical linear models solved the issue of non-independence of multiple results from the same
study in a meta-analysis, this study is still limited in the use of secondary data and the complexity and diversity of all studies
included in the meta-analysis. For example, various measures were used to measure self-efficacy in different studies included
in the meta-analysis and we had no way to account for the measurement errors. Another limitation of this study is that
sources of self-efficacy were not examined. Qualitative studies are needed to have a better understanding and further
development of the sources of self-efficacy in the context of teaching or learning English as a foreign language. Future re-
searchers should also pay more attention to how to foster self-efficacy beliefs in the classrooms across various cultures with
well-designed quasi-experimental studies and longitudinal studies in order to understand causal relationships with objective
measures of learning outcomes.
Nevertheless, the study has implications for both researchers and practitioners in the field of language learning and
teaching. For researchers, the results from this study confirmed a positive relationship between self-efficacy beliefs and
language learning outcomes. This relationship is consistent across various domains (overall, receptive, and productive lan-
guage skills), education levels (K-12 and higher education), publication type (journal articles and dissertation/thesis), and
culture (East and West). Therefore, the next step for future researchers is to explore what factors mediate the relationship
between self-efficacy beliefs and language learning outcomes. For example, future researchers can investigate whether
motivation to learn or willingness to communicate mediates the relationship between self-efficacy and learning outcomes
and if student background information such as family socioeconomic status moderates the relationship. Researchers should
also pay more attention to the four sources of self-efficacy (Bandura, 1997) in order to find out how to effectively promote
students’ self-efficacy beliefs in the classroom. What is more important in the findings is that the magnitude of effect for the
relationship between self-efficacy and learning outcomes varies across studies, especially by the statistical method used and
cultural contexts. The effect sizes reported by Pearson correlation coefficients failed to consider the contribution of other
factors and therefore inflated the effect size. As a result, researchers should be cautious when reporting or citing effect sizes in
previous studies because the effect size could vary due to the statistical model used, e.g., linear regression with various
covariates or structural equation models with multiple constructs. A unique contribution of this study is the inclusion of
research with Middle Eastern students who are under-researched in the literature.
This study also has implications for educational researchers in language studies. We found that the primary studies
included in the current meta-analysis were not evenly distributed by grade levels or domains of language learning. Fewer
studies examined self-efficacy in listening and speaking as compared to other domains, which may be accounted for by the
fact that schools give priority to students’ literacy development, whereas insufficient attention has been paid to listening and
speaking (Ferris & Tagg, 1996). There were also few studies in the K-12 setting compared to the college setting. Therefore,
more studies are recommended to investigate the relationship between self-efficacy and listening or speaking proficiency
with K-12 students.
This study reminds practitioners in language teaching to make decisions about their pedagogy based on multiple re-
sources rather than a single study. This meta-analysis found that the relationship between self-efficacy and language pro-
ficiency was stronger in East Asian and Middle Eastern cultures than in Western culture. This claim was based on the evidence
of a statistically significantly larger effect size for the relationship between self-efficacy beliefs and English language profi-
ciency for East Asian and Middle Eastern students. This conclusion was also supported by social cognitive theories and
previous empirical studies (e.g., Bandura, 1997; Pajares, 1996; Usher & Pajares, 2008; Wang & Bai, 2017). Implications of this

9
C. Wang and T. Sun System 95 (2020) 102366

finding in the practice include helping students in various cultures develop more accurate self-efficacy of themselves. For
example, English language teachers in East Asian and Middle Eastern classrooms may help students foster a higher level of
self-efficacy beliefs by recognizing their improvements/achievements and giving them more positive feedback whereas
teachers in European and American classrooms may help students evaluate the difficulty of the tasks and their own
competence. Teachers in all classrooms should provide more constructive feedback with details about how to improve the
students’ works rather than simply giving them verbal compliments. Besides, classroom teachers should take the re-
sponsibility of creating a more supportive environment for students to gain more successful experience and observe their
peers for modeling. Finally, caring about the students’ emotional and psychological well-being is also important to facilitate
the fostering of the students’ self-efficacy beliefs (Bandura, 1997).

Author statement

I am writing to submit our manuscript entitled “Relationship between Self-efficacy and Language Proficiency: A Meta-
Analyses” to System. The project was approved by the IRB, and the results of the study was not published elsewhere or written
in any manuscript under review by other journals. All American Educational Research Association ethical standards were
followed. APA 7th edition was also followed. There is no conflict of authorship.

Acknowledgment

Support for this research was provided in part by the University of Macau Research and Development Grant for Chair
Professor (CPG2019-00031-FED). The opinions expressed do not necessarily reflect the position or policy of the University of
Macau, and no official endorsement should be inferred. Correspondence concerning this article should be addressed to
Chuang Wang ([email protected]).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.system.2020.102366.

References

Amogne, D. (2008). An investigation of the correlation among efficacy sources, students’ self-efficacy and performance in reading and writing skills: Bahir Dar
University in focus. Addis Ababa, Ethiopia: Addis Ababa University (Unpublished master’s thesis).
Bai, B. (2018). Understanding primary school students’ use of self-regulated writing strategies through think-aloud protocols. System, 78, 15e26. https://doi.
org/10.1016/j.system.2018.07.003
Bai, B., Chao, G. C., & Wang, C. (2019). The relationship between social support, self-efficacy, and English language learning achievement in Hong Kong. Tesol
Quarterly, 53, 208e221. https://doi.org/10.1002/tesq.439
Bai, B., & Wang, J. (2020). Growth mindset, self-efficacy and intrinsic value: How self-regulated learning mediates the role of motivational beliefs in English
language learning achievements in an Asian context. Language Teaching Research. https://doi.org/10.1177/1362168820933190
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191e215.
Bandura, A. (1986). Social foundations of thought and action. Prentice- Hall.
Bandura, A. (1997). Self-efficacy: The exercise of control. Freeman.
Bong, M. (2002). Predictive utility of subject-, task-, and problem-specific self-efficacy judgments for immediate and delayed academic performances. The
Journal of Experimental Education, 70, 133e162. https://doi.org/10.2307/20152673
Bong, M., & Skaalvic, E. M. (2003). Academic self-concept and self-efficacy: How different are they really? Educational Psychology Review, 15, 1e40. https://
doi.org/10.1023/A:1021302408382
Borenstein, M., Hedges, L. V., Higgins, J. P., & Rothstein, H. R. (2009). Introduction to meta-analysis. Chichester, UK: John Wiley & Sons.
Bruning, R., Dempsey, M., Kauffman, D. F., Zumbruun, S., & McKim, C. (2013). Examining dimensions of self-efficacy for writing. Journal of Educational
Psychology, 105(1), 25e38. https://doi.org/10.1037/a0029692
Chea, S. (2012). The relationships among writing self-efficacy, writing goal orientation, learning strategies, and writing achievement (Master’s thesis). Retrieved
form ProQuest Dissertations and Theses. (Accession or Order No. 1513414).
Coladarci, T., Cobb, C. D., Minium, E. W., & Clarke, R. B. (2014). Fundamentals of statistical reasoning in education (4th ed.). Hoboken, NJ: John Wiley & Sons, Inc.
Corkett, J., Hatt, B., & Benevides, T. (2011). Students and teacher self-efficacy and the connection to reading and writing. Canadian Journal of Education, 34(1),
65e98.
Diseth, A. (2011). Self-efficacy, goal orientations and learning strategies as mediators between preceding and subsequent academic achievement. Learning
and Individual Differences, 21, 191e195. https://doi.org/10.1016/j.lindif.2011.01.003
Duffin, E. (2019). The most spoken languages worldwide (native speakers in millions). In Statista. Retrieved from https://www.statista.com/statistics/
266808/the-most-spoken-languages-worldwide/.
Earley, P. C. (1999). Playing follow the leader: Status-determining traits in relation to collective efficacy across cultures. Organizational Behavior and Human
Decision Processes, 80, 192e212. https://doi.org/10.1006/obhd.1999.2863
Eaton, M. J., & Dembo, M. H. (1997). Differences in the motivation beliefs of Asian American and non-Asian students. Journal of Educational Psychology, 89,
433e440. https://doi.org/10.1037/0022-0663.89.3.433
Ferris, D., & Tagg, T. (1996). Academic listening/speaking tasks for ESL students: Problems, suggestions, and implications. Tesol Quarterly, 30(2), 297e320.
Galla, B. M., Wood, J. J., Tsukayma, E., Har, K., Chui, A. W., & Langer, D. A. (2014). A longitudinal multilevel model analysis of the within-person and between-
person effect of effortful engagement and academic self-efficacy on academic performance. Journal of School Psychology, 52, 295e308. https://doi.org/10.
1016/j.jsp.2014.04.001
Grigg, S., Perera, H. N., McIlveen, P., & Svetleff, Z. (2018). Relations among meth self-efficacy, interest, intentions, and achievement: A social cognitive
perspective. Contemporary Educational Psychology, 53, 73e86. https://doi.org/10.1016/j.cedpsych.2018.01.007
Huang, P., Wang, C., & Xie, L. (2015). Self-efficacy beliefs and self-regulated learning strategies of Chinese students learning English as a second language and
as a foreign language. In C. Wang, W. Ma, & C. L. Martin (Eds.), Chinese education from the perspective of American educators (pp. 179e197). Information
Age Publishing.

10
C. Wang and T. Sun System 95 (2020) 102366

Jalaluddin, I., Yunus, M. M., Yamat, H., & Jusoff, H. K. (2010). A case study on teacher’s assistance in writing classroom: A look at the effects on rural learners’
writing self-efficacy. The International Journal of Learning, 17(9), 27e44. https://doi.org/10.18848/1447-9494/CGP/v17i09/47229
Kim, D., Wang, C., Ahn, H. S., & Bong, M. (2015). English language learners’ self-efficacy profiles and relationship with self-regulated learning strategies.
Leaning and Individual Differences, 38, 136e142. https://doi.org/10.1016/j.lindif.2015.01.016
Klassen, R. M. (2004). Optimism and realism: A review of self-efficacy from a cross-cultural perspective. International Journal of Psychology, 39, 205e230.
https://doi.org/10.1080/00207590344000330
Komarraju, M., & Nadler, D. (2013). Self-efficacy and academic achievement: Why do implicit beliefs, goals, and effort regulation matter? Learning and
Individual Differences, 25, 67e72. https://doi.org/10.1016/j.lindif.2013.01.005
Liem, A. D., Lau, S., & Nie, Y. (2008). The role of self-efficacy, task value, and achievement goals in predicting learning strategies, task disengagement, peer
relationship, and achievement outcome. Contemporary Educational Psychology, 33, 486e512. https://doi.org/10.1016/j.cedpsych.2007.08.001
Luszczynska, A., Scholz, U., & Schwarzer, R. (2005). The general self-efficacy scale: Multicultural validation studies. Journal of Psychology, 139, 439e457.
https://doi.org/10.3200/JRLP.139.5.439-457
Magogwe, J. M., & Oliver, R. (2007). The relationship between language learning strategies, proficiency, age and self-efficacy beliefs: A study of language
learning in Botswana. System, 35, 338e352. https://doi.org/10.1016/j.system.2007.01.003
Marsh, H. W., Abduljabbar, A. S., Morin, A. J., Parker, P., Abdelfattah, F., Nagengast, B., & Abu-Hilal, M. M. (2015). The big-fish-little-pond effect: General-
izability of social comparison process over two age cohorts from Western, Asian, and Middle Eastern Islamic countries. Journal of Educational Psychology,
107(1), 258e271. https://doi.org/10.1037/a0037485
McKay, S. L. (2002). Teaching English as an international language: Rethinking goals and perspectives. Oxford University Press.
Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & The Prisma Group. (2009). Preferred reporting items for systematic reviews and meta-analyses: The
PRISMA statement. PLoS Medicine, 6(7), Article e1000097. https://doi.org/10.1371/journal.pmed1000097
Multon, K., Brown, S. D., & Lent, R. W. (1991). Relation of self-efficacy beliefs to academic outcomes: A meta-analytic investigation. Journal of Counseling
Psychology, 38(1), 30e38. https://doi.org/10.1037/0022-0167.38.1.30
Naseri, M. (2012). The relationship between reading self-efficacy beliefs, reading strategy use and reading comprehension level of Iranian EFL learners.
World Journal of Education, 2(2), 64e75.
Nevill, M. A. (2008). The impact of reading self-efficacy and the regulation of cognition on the reading achievement of an intermediate elementary sample
(Doctoral dissertation). Retrieved from ProQuest Dissertations and Theses. (Accession or Order No. 3303552).
Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of Educational Research, 66, 543e578. https://doi.org/10.3102/00346543066004543
Pajares, F., & Graham, L. (1999). Self-efficacy, motivation constructs, and mathematics performance of entering middle school students. Contemporary
Educational Psychology, 24, 124e139. https://doi.org/10.1006/ceps.1998.0991
Pajares, F., & Valiante, G. (2001). Gender differences in writing motivation and achievement of middle school students: A function of gender orientation?
Contemporary Educational Psychology, 26, 366e381. https://doi.org/10.1006/ceps.2000.1069
Polanin, J. R., Tanner-Smith, E. E., & Hennessy, E. A. (2016). Estimating the difference between published and unpublished effect sizes: A meta-review. Review
of Educational Research, 86, 207e236. https://doi.org/10.3102/0034654315582067
Prat-Sala, M., & Redford, P. (2012). Writing essays: Does self-efficacy matter? The relationship between self-efficacy in reading and in writing and un-
dergraduate students’ performance in essay writing. Educational Psychology: International Journal of Energy Economics and Policy, 32(1), 9e20. https://
doi.org/10.1080/01443410.2011.621411
Rahimi, A., & Abedini, A. (2009). The interface between EFL learners’ self-efficacy concerning listening comprehension and listening proficiency. Novitas
Royal Research on Youth and Language, 3(1), 14e28.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Sage Publications.
Rothstein, H. R., Sutton, A. J., & Borenstein, M. (2005). Publication bias in meta-analysis: Prevention, assessment and adjustments. John Wiley & Sons.
Sanders-Reio, J., Alexander, P. A., Reio, T. G., Jr., & Newman, I. (2014). Do students’ beliefs about writing relate to their writing self-efficacy, apprehension, and
performance? Learning and Instruction, 33, 1e11. https://doi.org/10.1016/j.learninstruc.2014.02.001
Sarshar, N., & Oroji, M. R. (2016). The relationship between intermediate EFL students’ oral reproduction and their willingness to communicate and self-
efficacy. International Journal of Emerging Trends in Science and Technology, 3, 4149e4156. https://doi.org/10.18535/ijetst/v3i06.18
Scholz, U., Gutierrez-Dona, B., Sud, S., & Schwarzer, R. (2002). Is general self-efficacy a universal construct? Psychometric findings from 25 countries.
European Journal of Psychological Assessment, 18, 242e251. https://doi.org/10.1027//1015-5759.18.3.242
Shang, H. (2010). Reading strategy use, self-efficacy and EFL reading comprehension. The Asian EFL Journal Quarterly, 12(2), 18e42.
Sun, T., & Wang, C. (2020). College students’ writing self-efficacy and writing self-regulated learning strategies in learning English as a foreign language.
System, 90, 1e17. https://doi.org/10.1016/j.system.2020.102221
Swanson, H. L., Kudo, M., & Guzman-Orth, D. (2016). Cognition and literacy in English language learners at risk for reading disabilities: A latent transition
analysis. Journal of Educational Psychology, 108, 830e856. https://doi.org/10.1037/edu0000102
Truong, T. N. N., & Wang, C. (2019). Understanding Vietnamese college students’ self-efficacy beliefs in learning English as a foreign language. System, 84,
123e132. https://doi.org/10.1016/j.system.2019.06.007
Usher, E. L., & Pajares, F. (2008). Sources of self-efficacy in school: Critical review of the literature and future directions. Review of Educational Research, 78,
751e796. https://doi.org/10.3102/0034654308321456
Wang, C., & Bai, B. (2017). Validating the instruments to measure ESL/EFL learners’ self-efficacy beliefs and self-regulated learning strategies. Tesol Quarterly,
51(4), 931e947. https://doi.org/10.1002/tesq.355
Wang, C., Hu, J., Zhang, G., Chang, Y., & Xu, Y. (2012). Chinese college students’ self-regulated learning strategies and self-efficacy beliefs in learning English
as a foreign language. Journal of Research in Education, 22(2), 103e135. Retrieved from https://eric.ed.gov/?id¼EJ1098422.
Wang, C., & Pape, S. (2007). A probe into three Chinese boys’ self-efficacy beliefs learning English as a second language. Journal of Research in Childhood
Education, 21(4), 364e377. https://doi.org/10.1080/02568540709594601
Wang, C., Schwab, G., Fenn, P., & Chang, M. (2013). Self-efficacy and self-regulated learning strategies for English language learners: Comparison between
Chinese and German college students. Journal of Educational and Developmental Psychology, 3(1), 173e191. https://doi.org/10.5539/jedp.v3n1p173
Wida Consortium. (2014). WIDA’s 2012 Amplification of English language development standards, kindergarten-grade 12. Retrieved from https://wida.wisc.edu/
sites/default/files/resource/2012-ELD-Standards.pdf.
Williams, J. D., & Takaku, S. (2011). Help seeking, self-efficacy, and writing performance among college students. Journal of Writing Research, 3(1), 1e18.
Wilson, K., & Narayan, A. (2016). Relationships among individual task self-efficacy, self-regulated learning strategy use and academic performance in a
computer-supported collaborative learning environment. Educational Psychology, 36, 236e253. https://doi.org/10.1080/01443410.2014.926312
Woodrow, L. (2011). College English writing affect: Self-efficacy and anxiety. System, 39, 510e522. https://doi.org/10.1016/j.system.2011.10.017
Yan, W., & Gaier, E. L. (1994). Causal attributions for college success and failure: An Asian-American comparison. Journal of Cross-Cultural Psychology, 25,
146e158. https://doi.org/10.1177/0022022194251009
Zhang, X., Ardasheva, Y., & Austin, B. W. (2020). Self-efficacy and English public speaking performance: A mixed-method approach. English for Specific
Purposes, 59, 1e16.
Zhang, Y., & Guo, H. (2012). A study of English writing and domain-specific motivation and self-efficacy of Chines EFL learners. Pan-Pacific Association of
Applied Linguistics, 16(2), 101e121.
Zimmerman, B. J., & Kitsantas, A. (2005). Homework practices and academic achievement: The mediating role of self-efficacy and perceived responsibility
belief. Contemporary Educational Psychology, 30, 397e417. https://doi.org/10.1016/j.cedpsych.2005.05.003

11