Affective Teacher—Student Relationships and Students' Externalizing Behavior Problems: A Meta-Analysis

This meta-analysis of 57 primary studies with 73,933 students shows strong links between affective teacher—student relationships (TSRs) and students' externalizing behavior problems (EBPs). Moreover, students' culture, age, gender, and the report types of EBPs moderated these effects. The negative correlation between positive indicators of affective TSRs and students' EBPs was stronger (a) among Western students than Eastern ones, (b) for students in the lower grades of primary school than for other students, (c) when rated by teachers or parents than by students or peers, and (d) among females than among males. In contrast, the positive correlation between negative indicators of affective TSRs and students' EBPs was stronger (a) among Eastern students than Western ones, (b) for students in the higher grades of primary school than for other students, and (c) when rated by students or peers than by teachers or parents.


INTRODUCTION
Behavior problems occur when an individual violates social norms or rules of behavior (social maladjustment), leading to adverse effects and possibly behaviors that are harmful to himself or herself, others or even society (Zhang, 1999).
Over the past decade, researched on behavior problems have attracted the attention of an increasing number of psychology, education, sociology, and even psychopathology experts. Many researchers have explored the influence of school climate, parenting style, child-parent relationships, and family function on students' behavior problems (Haynes et al., 1997;Aunola and Nurmi, 2005;Pettit and Arsiwalla, 2008;Thornton et al., 2008). In particular, researchers have examined the links between teacher-student relationships (TSRs) and students' behavior problems (Vick, 2008;Ewing, 2009;Leflot et al., 2011;Spilt et al., 2012c;De Laet et al., 2014). Many theoretical and empirical studies have yield varied conclusions (Gest et al., 2005;Palermo et al., 2007;Doumen et al., 2009a). Nevertheless, the scope of problem behaviors includes many factors with different orientations and natures. This has led researchers to neglect communication with each other and avoid comparisons of their results.
Some researchers maintain that behavior problems should be classified as externalizing behavior problems (EBPs) vs. internalizing behavior problems (IBPs; Achenbach and Dumenci, 2001). EBPs are individual reflections regarding an external environment with negative external behaviors (Liu, 2004). Researchers have adopted different standards for classifying EBPs. For example, consider two dimensions: openness-concealment and destructive-nondestructive. Loeber et al. (2000) believe that EBPs should be divided into aggression, agonistic behavior, property damage, and reputation infringement. Based on individual behaviors, Cai and Zhou (2006) argued that EBPs should be divided into hyperactivity, aggression, and conduct problems. In contrast, IBPs are negative moods and emotions that lead to emotional disorder, including depression, anxiety, withdrawal, and guilt (Zanh-Waxler et al., 2000).
Highlighting these definitions of behavior problems clarifies various concepts' theoretical boundaries that determine the nature, direction, and veracity of research inquiries. According to these definitions, EBPs tend to be explicit and more destructive than IBPs. More importantly, several studies found that the correlation between affective TSRs and EBPs was stronger than that the correlation between affective TSRs and IBPs (Zhang and Sun, 2011;Gyllborg, 2013). Therefore, this study used a metaanalysis to explore the link between affective TSRs and students' EBPs, and excluded IBPs.
Researchers used different indicators of EBPs. For example, Achenbach (1991) developed the Achenbach child-behavior checklist (CBCL), according to which EBPs included delinquent behavior and aggressive behavior; Reynolds (2004) developed the Behavior Assessment System for Children-Teacher Rating Scales for Children (BASC-TRS), in which EBPs include hyperactivity, aggression, and conduct problems. Thus, in accordance with these previous studies, this study will consider delinquency, aggression, hyperactivity, and conduct problems as indicators of EBPs.
TSRs are an important component of interpersonal communication ability and social adaptability. This study focuses on a specific subset of TSRs, namely affective TSRs; this choice was inspired by Cornelius-White's findings that the affective variables "empathy" and "warmth" are strongly associated with student outcomes (Cornelius- White, 2007). Roorda et al. (2011) considered both positive and negative indicators of affective TSRs. Specifically, positive indicators of affective TSRs comprise closeness, support, liking, warmth, and trust. In contrast, negative indicators comprise conflict, anger, and dislike. Although some argue that dependency is a component of affective TSRs (Pianta, 2001;Fraire et al., 2013;Settanni et al., 2015), other studies using multiple methods to examine relationship quality questioned the validity of dependency as a measure of dyadic relationship quality (Doumen et al., 2009b;Roorda et al., 2011); thus, dependency was excluded from this study.
According to stage-environment fit theory, individual development requires an interpersonal relationship that has trust, support, caring, self-expression, self-choice, and selfdetermination; in cases where. A teacher who did not provide these interpersonal relationships and opportunities created an environmental mismatch with individual development, thus leading to students showing EBPs (Wang, 2009;van Lier et al., 2012;Loukas et al., 2013). Moreover, many empirical studies have found that positive indicators of TSRs were negatively correlated with students' EBPs (Gest et al., 2005;Spilt et al., 2012a;Thijs et al., 2012) while negative indicators of affective TSRs were positively correlated with students' EBPs (Doumen et al., 2009a;Spilt et al., 2012a). However, correlations varied across studies. To resolve this issue, several researchers have summarized research results with reviews (Baker et al., 2008;Nurmi, 2012), but these studies only partly verified the phenomena. Their limitations include convenience samples, various sample sizes, or ignoring moderators, which led to inconsistencies and low reliability. Therefore, a meta-analysis is needed to determine the link between affective TSRs and students' EBPs.
Our review of past empirical studies showed that many effect sizes were heterogeneous, suggesting that moderating factors might account for different links between affective TSRs and students' EBPs. Thus, we hypothesized that one or more variables may moderate the effect sizes of the correlation between affective TSRs and students' EBPs, such as differences in students' cultures, ages, genders, and the report types of EBPs.
First, we examine whether students' culture (as a latent variable) moderates the link between affective TSRs and students' EBPs (Chang et al., 2007;Roorda et al., 2014). Several studies suggest that culture influences the link between affective TSRs and students' EBPs (e.g., closeness and EBP, and conflict and EBPs). Baker (2006) found a moderate correlation between closeness and students' EBPs among Western students; however, Ly (2013), whose sample included Eastern students, found a weak correlation between the two factors. Many studies found a strong correlation between conflict and students' EBPs among Western students (Doumen et al., 2008(Doumen et al., , 2009aLy, 2013); however, Fu (2014), whose sample included Eastern students, found moderate correlation between the two factors. Thus, in accordance with these findings, this study tests whether the correlation between positive indicators of affective TSRs and students' EBPs for Western students is stronger than that for Eastern students, and whether the correlation between positive indicators of affective TSRs and students' EBPs for Western students is weaker than for Eastern students.
Second, as the level of affective TSRs and students' EBPs might differ as a function of students' age (Zhang et al., 2008), we test whether students' age moderates the link between affective TSRs and students' EBPs. Differences in age have been found in the correlations between affective TSRs and students' EBPs. For example, previous studies indicated that positive indicators for affective TSRs and students' EBPs varied among students in kindergarten lower primary grades (LPG), and higher primary grades (Silver et al., 2005;HPG, Kuhns, 2008;Stewart and Suldo, 2011). In contrast negative indicators of affective TSRs and students' EBPs among kindergarten, LPG, and HPG students all showed similar phenomenon (Ezzell et al., 2000;Pianta and Stuhlman, 2004;Vick, 2008;Troop-Gordon and Kopp, 2011;Rudasill et al., 2013). Based on these findings, we expect age to moderate the link between affective TSRs and students' EBPs.
Third, we examine whether the report type of EBPs (as a latent variable) moderates the link between affective TSRs and students' EBPs. Raters with different ages, standpoints, values, and degrees of understanding a student might rate his or her EBPs inconsistently (Van Lier et al., 2005;Ladd, 2006). Moreover, several studies have found that different raters might account for the lack of coherence in research on the link between affective TSRs and students' EBPs. For example, some previous studies have relied on EBPs rated by students, which were only weakly related to positive indicators of affective TSRs (Troop-Gordon and Kopp, 2011;Li et al., 2012) while other studies found that student EBPs rated by teachers were moderately related to both positive indicators of affective TSRs (Colwell and Lindsey, 2003;Shin and Kim, 2008) and negative indicators of affective TSRs (White and Renk, 2012;Ly, 2013;Skalická et al., 2015). In contrast, other researchers found that student EBPs rated by teachers were strongly related to negative indicators of affective TSRs (Palermo et al., 2007;Fowler et al., 2008;Stipek and Miles, 2008). Thus, in accordance with these findings, we test whether Report type of EBPs moderate the link between affective TSRs and students' EBPs.
Fourth, we test whether gender (as a latent variable) moderates the link between affective TSRs and students' EBPs. Female students tend to have more affective TSRs than male students do (Spilt et al., 2012b), and male students tend to develop more EBPs than female students do (Hill et al., 2006). As a result, gender might influence the correlation between positive or negative indicators of affective TSRs and students' EBPs. Several empirical studies have showed gender differences in the link between indicators of affective TSRs and students' EBPs, such as closeness, support, and warmth (Ostrov and Crick, 2007;Spilt et al., 2012a;Thijs et al., 2012). Hence, these findings suggest that gender moderates the link between affective TSRs and students' EBPs.

Study Purpose
The current study models the link between affective TSRs and students' EBPs using meta-analysis. Specifically, this study (a) estimates the effect sizes of correlations between affective TSRs and students' EBPs and (b) tests whether the links between affective TSRs and students' EBPs are moderated by culture, age, report type of EBP, or gender.

Literature Search
To identify studies on affective TSRs and students' EBPs, we systematically searched the literature from January 2000 to March 2016 in electronic databases, including ProQuest Dissertations, Web of Science, Google Scholar, Springer, Taylor & Francis, EBSCO, PsycINFO, and Elsevier SDOL. Indexed keywords primarily included terms reflecting affective TSRs (relationship(s), closeness, warmth, support, empathy, trust, sensitivity, conflict, negativity, and anger) and students' EBPs (behavior problems, externalizing, aggression, conduct problem, hyperactivity, and oppositional). When articles could not be found online, we obtained full-text versions of articles from libraries. All articles obtained were written in English. We used inclusion and exclusion criteria to analyze and filter the collected studies.

Literature Exclusion Criteria
We included articles based on the following criteria: (a) tested the relation between affective TSRs and students' EBPs; (b) measured affective TSRs, including closeness, warmth, support, empathy, trust, sensitivity, conflict, negativity, or anger; (c) measured EBPs, including behavior problems, externalizing, aggression, conduct problem, hyperactivity, oppositional, or other indicators, (d) included an explicit sample size, and (e) explicitly reported the Pearson product-moment correlation coefficient (or a t or F-value that could be transformed into r). Table 1 summarizes the studies included in the Meta-Analysis.

Coding Study
To facilitate meta-analysis, feature coding was conducted on 57 articles. We considered the following variables: author(s) and publication year, proportion of male students, age, indicators of affective TSRs, indicators of EBPs, number of students, and r. The following criteria guided the coding procedure: (a) effect sizes of each independent sample were encoded based on an independent sample, and effect sizes were separately encoded if a study had several independent samples; (b) if a study reported a correlation between affective TSRs and EBPs many times, the mean value was instead of effect sizes; (c) if an independent sample provided effect sizes (expressed as r) for sample characteristics such as gender, the results for the two genders were coded separately; (d) if a study reported not only a correlation between a total of EBPs and affective TSRs but also a correlation between the dimensions of EBPs and affective TSRs, we only coded the former; (e) if a study reported a correlation between different indicators of affective TSRs and EBPs, we coded these separately; and (f) if a study reported a correlation between indicators of affective TSRs and different indicators of EBPs, we coded these separately.
When coding was complete, based on principles of metaanalysis (Lipsey and Wilson, 2001), effect sizes between affective TSRs and students' EBPs were calculated for each sample. Then, we test whether the links between affective TSRs and students' EBPs were moderated by (a) culture; (b) age; (d) report types of EBPs; or (e) gender.
Culture was coded as "Eastern, " "Western, " and "other"; "Eastern" referred to students from Asian countries such as China (mainland China, Hong Kong, Taiwan), South Korea, Philippines, Singapore, and so on. "Western" referred to students from European and North American countries such as Germany, the United States of America, and so on. Age was coded as "Kindergarten (3-6 years), " "lower grades of primary school (6-9 years), " "higher grades of primary school (9-12 years), " "Middle school (12-15 years), " "High school (15-18 years), " and "Mixed." "Mixed" indicated that students included at least two of the above categories. Report type of EBP was coded as "students rated, " "teacher rated, " "peer rated, " "parent rated." Gender was coded as the proportion of male students.

Data Analysis
All data were analyzed using Comprehensive Meta-Analysis software (CMA Version 2.0). A fixed effects model calculated the homogeneity test and mean effect. Averaged weighted (withinand between-inverse variance weights) correlation coefficients   of independent samples were used to compute mean effect sizes. Moderators were decided by the homogeneity test, which revealed variance in effect sizes between different samples' characteristics. When the homogeneity test was significant (Q Bet > 0.05), post-hoc contrasts were implemented to test whether the groups were statistically different. This study used meta-analysis to test whether each moderator accounted for the variation in the effect sizes.

Correlation between Affective TSRs and Students' EBPs
After filtering the literature, we used 57 independent samples and calculated 149 effect sizes (78 effect sizes between positive indicators of affective TSRs and EBPs and 71 effect sizes between negative indicators of affective TSRs and EBPs). In these reviewed studies, 73,933 students participated, and the sample sizes ranged from 20 to 2335. We calculated sample sizes (k), weighted effect sizes (r), and 95% confidence intervals (see Table 2). Furthermore, a fixed effects model was used to homogenize the analysis. The results showed significant negative correlations between positive indicators of affective TSRs and EBPs (r = −0.263 [z = −52.031, p < 0.001, k = 78, 95% CI = −0.272, −0.253]) and significant positive correlations between negative indicators of affective TSRs and EBPs (r = 0.554 [z = 118.588, p < 0.001, k = 71, 95% CI = 0.547, 0.561]). These effect sizes were suitable for moderator analysis.

Moderator Analysis
We conducted two total homogeneity tests across 78 (PI) and 71 (NI) independent samples. The results showed a significant homogeneity coefficient between affective TSRs and students' EBPs [Q T(77)PAE = 879.022, p < 0.001; Q T(70)NAE = 2431.398, p < 0.001]. These results indicate that culture, age, report types of EBPs and gender might moderate the links between affective TSRs and students' EBPs. Therefore, we used meta-analysis of variance to examine whether culture, age, and report types of EBPs moderated the correlations between affective TSRs and students' EBPs, and we used meta-regression analyses to examine whether gender influenced the relation between affective TSRs and students' EBPs. Table 3, a homogeneity test showed a significant homogeneity coefficient between positive indicators of affective TSRs and EBPs across Eastern culture students and across Western culture students (Q BET = 8.816, df = 1, p < 0.001). In particular, Table 3 shows that the Western students (r = −0.267, 95% CI = −0.277, −0.258) indicated a stronger correlation between positive indicators of affective TSRs and EBPs than the Eastern students (r = −0.207, 95% CI = −0.246, −0.167). Likewise, the homogeneity test found significant differences in the correlation between negative indicators of affective TSRs and EBPs across the two cultures (Q BET = 25.307, df = 1, p < 0.001). Table 3 also shows a stronger correlation between negative indicators of affective TSRs and EBPs among Eastern students (r = 0.675, 95% CI = 0.631, 0.714) than Western students (r = 0.551, 95% CI = 0.544, 0.558).

Report type of EBPs
The results of the homogeneity test (Q BET = 101.736, df = 3, p < 0.001) suggested that age influenced the link between affective TSRs and EBPs. Positive indicators of affective TSRs were negatively correlated with EBPs when rated by teachers ( The correlation between positive indicators of affective TSRs and EBPs were stronger when rated by teachers or parents than by others. As shown in Table 3, the homogeneity test (Q BET = 349.373, df = 4, p < 0.001) suggested that age moderated the link between negative indicators of affective TSRs and EBPs. Negative indicators of affective TSRs were positively correlated to EBPs when rated by teachers (r = 0.602, 95% CI = 0.594, 0.610), students (r = 0.314, 95% CI = 0.262, 0.364), peers (r = 0.404, 95% CI = 0.351, 0.455), or parents (r = 0.429, 95% CI = 0.410, 0.448). These results indicate that the correlation between negative indicators of affective TSRs and EBPs were lower when student rated than when rated by others.

Gender
To examine whether gender moderated the links between affective TSRs and students' EBPs, r was meta-regressed onto the percentage of male students in each sample. In Table 4, meta-regression analysis (Q Model [1, k = 74] NI = 4.106, p < 0.05) showed that gender moderated the link between positive indicators of affective TSRs and students' EBPs; as the proportion of female students increased, the link was stronger. The correlations between positive indicators of affective TSRs and EBPs for an all-female sample (r = −0.315) were stronger than those for an all-male sample (r = −0.249). In contrast, metaregression analysis (Q Model [1, k = 66] PAE = 1.666, p > 0.05) showed that gender did not moderate the link between negative indicators of affective TSRs and students' EBPs.

Publication Bias
To examine whether the results were biased due to effect sizes from various sources, we drew a funnel plot (see Figure 1). It showed that the 149 effect sizes were symmetrically distributed on both sides of the average effect size, and an Egger's regression (Egger et al., 1997) revealed no significant bias [t (147) = 0.010, p = 0.991 > 0.05]. Egger's regression is an effective method for examining publication bias (Teng et al., 2015). In addition, we conducted Egger's regression analysis on both positive and negative indicators of affective TSRs and EBPs. The results also showed no publication bias [t PI (76) = 0.767, p = 0.445 > 0.05; t NI (69) = 0.568, p = 0.572 > 0.05]. Together, these results indicated stability in the overall correlation between affective TSRs and students' EBPs in this study.

DISCUSSION
In the current meta-analysis 57 recent studies, with 149 effect sizes and 73,933 students are reviewed. We examined the effect sizes of correlations between affective TSRs and students' EBPs, revealing culture, age, report type of EBPs and gender as moderators influencing the links. The results showed that negative affective TSRs was negatively correlated with students' EBPs and negative affective TSRs was positively correlated with students' EBPs. The correlation coefficients for these results were both medium. In addition, these results showed that students' cultures, ages, genders, and report type of EBPs moderated the link between affective TSRs and students' EBPs.

The Significant Correlation between Affective TSRs and Students' EBPs
The meta-analysis results indicated a significant negative correlation between positive indicators of affective TSRs and EBPs and a significant positive correlation between negative indicators of affective TSRs and EBPs. These results suggested that affective TSRs help students reduce EBPs. As indicated by Masten and Garmezy (1985), TSRs are an important support system for students' behavioral development and many studies focusing on improving students' behavior problems with TSRs. Moreover, students with closer TSRs had fewer antisocial behaviors (Birch and Ladd, 1998), and high levels of TSR closeness outperformed students' early problem behaviors when  predicting their later behavior problems (Pianta and Nimetz, 1991). In addition, this study found that, compared with the positive indicators of affective TSRs, negative indicators of affective TSRs showed more strong correlation with students' EBPs, suggesting that negative affect TSRs are more influential than positive affect TSRs on students' EBPs. These results suggest that reducing negative affective TSRs or increasing positive affective TSRs might reduce individuals' EBPs. Therefore, teachers might explore using diverse communication strategies to help students build positive affective TSRs and reduce negative affective TSRs. In addition, results suggest that targeted interventions might help students develop affective TSRs when they show EBPs.
This study's results support the direct effect model but did not test the indirect effect model. Future studies can test the indirect effect model of affective TSRs and students' EBPs.

Moderating Effects
Moderation analysis showed that students' cultures, ages, genders, and the report type of EBPs moderated many of the links between affective TSRs and students' EBPs. Gender did not moderate the link between the negative indicators of affective TSRs and EBPs.

Moderating Role of Culture
We hypothesized that students' culture might moderate the link between affective TSRs and students' EBPs. The results of this meta-analysis support this hypothesis. In particular, the correlation between positive indicators of affective TSRs and students' EBPs was stronger among Western students than Eastern ones. In contrast, the correlation between negative indicators of affective TSRs and students' EBPs was stronger among Eastern students than Western ones. These results suggest that positive affective TSRs might reduce EBPs more for Western students than for Eastern ones. In contrast, negative affective TSRs might increase EBPs more for Eastern students than Western ones. These results are consistent with previous studies (Fowler et al., 2008;Solheim et al., 2011;Zhang and Sun, 2011). The higher expectations and stricter TSRs in collectivist Eastern cultures compared to the lower expectations of relaxed TSRs in individualistic Western cultures might account for these differences; positive, relaxed TSRs might cultivate good behaviors and limit behavior problems while negative, strict TSRs might yield behavior problems more easily. These results suggest that differences in students' cultures must be considered when developing affective TSRs to reduce students' EBPs. Together, they suggest that training and interventions based on the specific culture of the student might be beneficial.

Moderating Role of Age
This meta-analysis found that age moderates the link between affective TSRs and students' EBP, consistent with past studies (Hughes and Cavell, 1999;Denham et al., 2000). Furthermore, additional analysis found that LPG students showed a stronger correlation between positive indicators of affective TSRs and EBPs than those in kindergarten and HPG students. Students' developing emotions at these ages and their interest in talking and building relationships with their teachers might explain these differences. LPG students might be exploring emotional relationships with their teacher and hence might be more willing to listen to their teachers' suggestions about correcting their EBPs. Additional analysis showed that the correlation between negative indicators of affective TSRs and EBPs are stronger among HPG students than kindergarteners or LPG students, possibly because as students get older, the proportion of positive affective TSRs decreases while that of negative affective TSRs increases (Wang and Wang, 2002). LPG students might be more likely than younger students to use disruptive behaviors to attract teacher attention; these behaviors can reduce the positive affect and increase the negative affect of their TSR, fostering a vicious cycle between affective TSRs and students' EBPs. These results suggest that we pay closer attention to the age and development of students when developing affective TSRs to reduce students' EBPs.

Moderating Role of Report Type of EBPs
This study showed that the report type of EBPs moderates the link between affective TSRs and students' EBPs. Specifically, the correlation between positive indicators of affective TSRs and EBPs were stronger when rated by teachers or parents than by others. Also, the correlation between negative indicators of affective TSRs and EBPs was lower when rated by students or peers than otherwise. These results are supported by many other studies (Loeber et al., 1990;Deater-Deckard et al., 1998) and suggest that a link between affective TSRs and students' EBPs is more visible when rated by teachers or parents than otherwise. Teachers and parents might exaggerate the degree of the link between affective TSRs and students' EBPs, if students and peers understand their own EBPs better than their teachers and parents do (Achenbach, 1991). A possible alternate explanation is that students and peers downplay this link, as they pay less attention than teachers or parents to the teacher's role in students' behavior development.

Moderating Role of Gender
This study showed that gender moderates the links between positive affective TSRs and students' EBPs. As predicted, the allfemale group showed a stronger correlation between positive indicators of affective TSRs and students' EBPs than the allmale group. However, gender did not moderate the links between negative indicators of affective TSRs and students' EBPs. These results suggest that positive affective TSRs reduce female students' EBPs more easily than they do those of male students, possibly because female students care more about their relationships with their teachers, seek more positive emotions from them (Hu et al., 2015), and are more easily influenced by them, resulting in fewer EBPs compared to male students (Deater-Deckard and Dodge, 1997). In addition, this result suggest that we might need to attend more to developing TSRs with male students than with female students to reduce their EBPs.

Limitations and Implications
This meta-analysis has several limitations. First, only closeness, warmth, support, empathy, trust, sensitivity, conflict, negativity, and anger were selected as indicators of affective TSRs; other indicators, such as concern, caring, were not found. Furthermore, the selected indicators may overlap. Second, this study selected several familiar indicators of EBPs; others indicators, such destructive behavior, were excluded. Third, all the studies reviewed examined only direct effects; however, other studies have found that affective TSRs affects students' EBPs across other variables as well (Stanger and Lewis, 1993;Yoon, 2002). Therefore, future studies should test the indirect effects of affective TSRs on students' EBPs. Fourth, this study only considers whether students' culture, age, gender, and report type of EBPs moderated the link between affective TSRs and students' EBPs. Other variables, notably other indicators of affective TSRs and students' EBPs, should be examined in future studies as they may influence the links between affective TSRs and students' EBPs. Fifth, this study included only English articles, which may have narrowed its scope and neglected some cultures. Sixth, this meta-analysis was based on cross-sectional studies and correlational data; hence a causal relationship cannot be inferred.

CONCLUSION
Through reviewing 57 studies, 149 effect sizes, and 73,933 student participants, meta-analysis results revealed that positive and negative affective TSRs were significant correlated with students' EBPs. Furthermore, these correlations were moderated by students' culture, age, report type of EBPs, and gender. In particular, negative affective TSRs were more strongly linked than positive affective TSRs to students' EBPs. Also, the negative correlation between positive indicators of affective TSRs and EBPs was stronger among Western students than Eastern students. In contrast, the positive correlation between negative indicators of affective TSRs and EBPs was stronger among Eastern students than Western students. The negative correlation between positive indicators of affective TSRs and EBPs was stronger among LPG students than among other students (except the mixed group). Also, the positive correlation between negative indicators of affective TSRs and students' EBPs was stronger among HPG students than other students. The negative correlation between positive indicators of affective TSRs and EBPs was stronger when rated by teachers or parents than by students or peers. However, the positive correlation between negative indicators of affective TSRs and EBPs was stronger when rated by students or peers. The negative correlation between positive indicators of affective TSRs and students' EBPs was stronger among girls than among boys. However, gender did not moderate the link between negative indicators of affective TSRs and students' EBPs. This meta-analysis estimated effect sizes for students' EBPs during the past 17 years and suggests that differences in students' cultures, age, and gender can inform future research and practices.