The roles of attitudes towards learning and opposite sex as a predictor of school engagement: mixed or single gender education?

The study investigates how students’ attitudes toward learning and toward the opposite sex, as well as school type (single-sex vs. coeducational), parental attitudes, and parents’ education levels relate to and predict high school students’ school engagement. It also examines the relationship between school engagement and self-perceived academic achievement. The purpose is to clarify which factors most strongly predict school engagement to inform educational policies and practices. Research questions: (1) Are there relationships between school engagement and attitudes toward learning, opposite-sex attitudes, school type, parental attitudes, and parents’ education? (2) Is school engagement related to achievement level? (3) To what extent do school type, achievement level, parental attitude, and parents’ educational levels predict school engagement?

School engagement encompasses cognitive, behavioral, and emotional components and is linked to attendance, lower dropout, and academic success (Finn, 1993; Fredricks et al., 2004; Connell et al., 1994; Klem & Connell, 2004). Prior studies highlight effects of peer relations and teacher support (Erdoğdu, 2016; Shin et al., 2007; Thaliah & Hashim, 2008) and mixed findings regarding socioeconomic and parental factors (Arastaman, 2009; Conchas, 2001). Attitudes toward learning correlate with academic performance and motivation (Bråten & Strømsø, 2006; Duarte, 2007; Aktürk, 2012; Erdoğdu, 2017). Opposite-sex attitudes and friendships evolve developmentally and may influence adolescent priorities and school-related outcomes (Sullivan, 1953; Liem & Martin, 2011; Lacey et al., 2004; Underwood et al., 2009). The efficacy of single-sex versus coeducation remains debated, with studies reporting advantages for girls in single-sex settings, advantages for coeducation, or no differences (Kohlhaas et al., 2010; McFarland et al., 2011; Mulholland et al., 2004; Marsh & Rowe, 1996; Smithers & Robinson, 2006). Turkish context lacks research on coed versus single-sex schooling and school engagement, motivating this study.

Design: Relational survey model to examine relationships and predictive effects among variables related to school engagement. Sample: N=843 high school students in Istanbul (525 girls, 318 boys). School types: girls’ vocational high school (n=316), boys’ high school (n=154), coeducational high schools public/private (n=373 total; private 100F/65M=165; public 109F/99M=208). Self-perceived achievement: successful (n=264), moderately successful (n=531), unsuccessful (n=48). Reported parental attitudes: authoritarian (n=93), democratic (n=163), over-protective (n=366), over-demanding (n=136), other (n=79). Mothers’ education: illiterate (n=35), primary (n=306), high school (n=351), university (n=147). Fathers’ education: illiterate (n=10), primary (n=243), high school (n=377), university (n=208). Measures: - School Engagement Questionnaire (SEQ; Arastaman, 2006): 9 items, 5 subscales (not used separately here), 5-point Likert; current study Cronbach’s alpha=0.95. Descriptives: mean 100.87, SD 17.72, min 29, max 135. - Scale of Attitudes Towards Learning (SATL; Kara, 2010): 40 items, 4 factors (Nature of Learning, Expectation, Openness, Anxiety); prior alphas 0.72–0.78; test-retest 0.87; current alpha=0.85. Descriptives: mean 146.40, SD 16.78, min 70, max 187. - Opposite-Sex Attitude Scale (OSAS; Erdoğdu, 2018): 26 items, single-factor (variance 53.13%), prior alpha=0.95; current alpha=0.96. Descriptives: mean 93.35, SD 21.20, min 26, max 130. Procedure: Consents obtained; instruments administered in classrooms. Analysis: Pearson product-moment correlations among variables. Categorical predictors converted to dummy variables (K-1 coding). Multiple regression analyses assessed predictive power of each categorical set (school type, achievement, parental attitude, mother’s education, father’s education) on school engagement. Stepwise multiple regression identified the most parsimonious set of predictors and their incremental variance contributions (R² change).

Descriptive statistics: - SEQ: mean 100.87 (SD 17.72) - OSAS: mean 93.35 (SD 21.20) - SATL: mean 146.40 (SD 16.78) Correlations (Table 3): - SEQ with SATL: r=0.486, p<0.01 (moderate positive) - SEQ with OSAS: r=-0.028, ns (reported as non-significant) - SEQ with achievement level: r=0.236, p<0.01 (positive) - SEQ with type of school (coded composite): r=-0.335, p<0.01 (negative) - SEQ with parental attitude: r=-0.056 (reported as non-significant) - SEQ with mother’s education: r=-0.154, p<0.01 (low negative) - SEQ with father’s education: r=-0.185, p<0.01 (low negative) Multiple regressions with categorical blocks (Table 4): - School type (girls’, boys’, coed) predicted SEQ: R=0.342, R²=0.117, p<0.001 (≈12% variance). Girls’ high school positive predictor (β=0.283, p<0.001); boys’ high school negative predictor (β=-0.236, p<0.001); coed not significant in this block (p=0.095). - Achievement level (successful, moderate, unsuccessful) predicted SEQ: R=0.261, R²=0.068, p<0.001 (≈7%). Successful positive (β=0.178, p<0.001); unsuccessful negative (β=-0.201, p=0.036); moderate not significant. - Perceived parental attitude predicted SEQ: R=0.145, R²=0.021, p<0.001 (≈2%). Democratic (β=0.195, p<0.001) and over-protective (β=0.190, p<0.001) significant; authoritarian and over-demanding not significant. - Mother’s education predicted SEQ: R=0.165, R²=0.027, p<0.001; individual levels not significant. - Father’s education predicted SEQ: R=0.228, R²=0.052, p<0.001 (≈5%). Primary (β=0.245, p<0.001) and secondary (β=0.239, p<0.001) significant; illiterate, university not significant. Stepwise regression (Table 5; final R≈0.614, R²≈0.377): cumulative explained variance ≈38%. - Step 1: Attitudes toward learning (SATL) entered first; β≈0.486; R²=0.236 (24% variance). - Step 2: Type of school (coed) added; B≈-13.44; β≈-0.265; R² increase ≈0.070 (to 0.306). Negative association for coed vs reference. - Step 3: Unsuccessful achievement group added; B≈-12.16; β≈-0.153; ΔR²≈0.023 (to 0.329). - Step 4: Boys’ school added; B≈-5.87; β≈-0.165; ΔR²≈0.023 (to 0.352). - Step 5: Successful achievement group added; B≈0.685; β≈0.143; ΔR²≈0.018 (to 0.371). - Step 6: Opposite-sex attitude added; B≈-0.047; β≈-0.057; ΔR²≈0.004 (to 0.375) (small negative effect). - Step 7: Father’s education (primary) added; B≈2.054; β≈0.058; ΔR²≈0.003 (to 0.377). Group differences summarized: - Higher school engagement among girls in single-sex schools relative to boys’ single-sex and coed schools. - Coed and boys’ school attendance associated with lower engagement in multivariate models. - Higher self-perceived achievement associated with higher engagement; being unsuccessful associated with lower engagement. - Democratic and over-protective parental attitudes associated with higher engagement. - Lower parental education (particularly father primary/secondary) showed small positive prediction of engagement in some models.

Findings indicate that positive attitudes toward learning are the strongest predictor of school engagement, aligning with prior work linking learning motivation and strategies to engagement and academic outcomes. School type also matters: attending coeducational and boys’ schools is associated with lower engagement compared to girls’ single-sex schools. The authors suggest that girls may demonstrate higher engagement due to faster maturation, stronger peer interactions, and sociocultural factors limiting out-of-school opportunities, increasing the value of school participation. Achievement relates positively to engagement; students who feel successful show more belonging and involvement, reflecting reciprocal influences between engagement and achievement. Democratic and over-protective parental attitudes are linked to higher engagement, possibly through supportive involvement and respect for the child’s autonomy. Opposite-sex attitudes show a small negative association with engagement in multivariate analysis, interpreted as adolescents’ increased focus on opposite-sex relationships potentially diverting energy from school. Lower paternal education predicting slightly higher engagement may reflect perceived educational importance as a pathway to improved socioeconomic status in families with fewer resources. These results collectively highlight cognitive-motivational (attitudes toward learning), contextual (school type), and familial (parental attitudes, education) influences on engagement, with implications for tailoring school environments and family–school partnerships.

This study contributes evidence that attitudes toward learning are the most influential predictor of school engagement among Turkish high school students. School type is also important: girls’ single-sex schools exhibit higher engagement, whereas coeducational and boys’ single-sex contexts are associated with lower engagement. Engagement is positively related to self-perceived academic success, and supportive parental attitudes (democratic/over-protective) are associated with higher engagement. Opposite-sex attitudes have a small negative relation with engagement. Practically, interventions to strengthen positive learning attitudes, enhance school climates (with attention to gendered needs), and foster constructive parental involvement may improve engagement and achievement. The authors recommend exploring single-sex class pilots, designing curricula sensitive to gender differences, and further investigating why higher parental education may relate to lower engagement in this context. Future research should replicate in varied regions, school types, and with longitudinal designs to clarify causal pathways and generalizability.

The authors note that findings should be interpreted in light of the specific context (Istanbul high schools) and characteristics of the participating adolescent students. They recommend conducting similar studies in different contexts and with different samples to assess generalizability. Additionally, the use of self-reported measures and cross-sectional design limits causal inference.