國中生學業情緒、情境興趣及學習涉入的交互關係

Study on Reciprocal Relations among Academic Emotions, Situational Interest, and Learning Engagement

本研究之目的在分析學業情緒、情境興趣與學習涉入之間的交互關係。為達研究目的，本研究採縱貫研究，抽取600名（男生302人）七、八年級學生為研究樣本，進行4次的測量，所蒐集的四波觀察資料以結構方程模式進行分析。研究發現如下：本研究建構之學業情緒、情境興趣與學習涉入交互效果模式受到觀察資料的支持，在愉悅模式上，學生的愉悅情緒、情境興趣與學習涉入具有正向交互效果關係，即學生第一次施測時的愉悅情緒及情境興趣可以提高其第二次施測時的學習涉入，而學生第二次施測時的學習涉入可以正向預測學生第三次施測時的愉悅情緒及情境興趣，第三次施測時的愉悅情緒及情境興趣可以提高其第四次施測的學習涉入；在無趣模式上，無趣情緒則與學習涉入具有負向交互效果關係，即學生第一次施測時的無趣興趣可以降低其第二次施測時的學習涉入，而學生第二次施測時的學習涉入可以負向預測學生第三次施測時的無趣情緒。本研究根據研究結果提出建議，以提供國中教學現場及未來研究之參考。

In educational psychology research, researchers have highlighted problems in students' learning. In fact, many researchers have suggested that students' learning processes are more crucial than their learning outcomes are (Linnenbrink & Pintrich, 2003; Schraw et al., 2001). However, little attention has been paid to students' learning activities and processes; therefore, they formed the focus of the present study. Educational psychology researchers have used learning engagement to describe behaviors in students' learning processes (Putwain et al., 2019). Learning engagement has been defined as the degree to which students participate and engage in learning activities, and it is a critical behavior in active learning. Several studies have demonstrated that learning engagement is closely related to learning motivation, situational interest, and academic emotions (Ainley & Ainley, 2011; Pekrun & Linnenbrink-Garcia, 2012). To investigate academic emotions, researchers in educational psychology have mostly used academic emotion theory (Pekrun, 2006). On the basis of the influence of academic emotion processes, Pekrun (2006) proposed the cognitive-motivational model, which posits that academic emotions influence learners' cognition, motivation, and engagement, thereby influencing their learning achievement. Moreover, in return, learners' learning achievement subsequently influenced their motivation, engagement, and academic emotions, and their motivation and engagement influenced their academic emotions. Therefore, reciprocal relations exist between academic emotions and learning engagement, yet few studies have examined them. Furthermore, a close relationship exists between situational interest and students' learning engagement (Reschly et al., 2008). Situational interest is defined as temporary interest that arises spontaneously due to environmental factors (Hidi & Harackiewicz, 2000; Renninger & Hidi, 2002). Reeve and Tseng (2011) defined student engagement as a four-dimensional concept consisting of behavioral, emotional, cognitive, and agentic aspects. Much research has found that situational interest influences students' learning engagement (Lipstein & Renninger, 2006; Renninger & Shumer, 2002; Silvia, 2006). Some researchers found that situational interest enhanced students' learning engagement, which in turn influenced their situational interest (Ainley & Ainley, 2011). However, relevant studies have only used cross-sectional analyses; limited studies have examined these reciprocal relations. Therefore, the present study proposed a reciprocal relationship between situational interest and learning engagement. According to theoretical and empirical research, the present study constructed an academic emotion, situational interest, and learning engagement reciprocal effect model, which focused on the reciprocal relations between academic emotions (enjoyment and boredom) and learning engagement, situational interest, and learning engagement. The present study employed a longitudinal design to examine the reciprocal relations between these variables. Data collected from a longitudinal study with seventh-and eighth-grade junior high school students (N = 643; 302 boys) from 6 schools (25 classes) in Tainan City, Taiwan, were examined. Data collection was conducted in four waves. Participants were asked to complete the academic emotions scale (enjoyment and boredom), situational interest scale, and learning engagement scale. All scales referred to the students' Chinese classes. Attrition analyses revealed no significant differences in T_1 enjoyment and boredom between partially missing (N = 43) and complete samples; a significant difference existed in T_1 situational interest between partially missing and complete samples, but the effect size was small (η^2 = .01). Therefore, we used 600 complete samples in the subsequent analysis. Moreover, we analyzed the sex and grade-level differences of partially missing samples. The results indicated no significant differences between boys and girls in seventh and eighth grade among the partially missing samples. Finally, the present study used a class unit for sampling (nested data), which may have entailed clustering effects. According to Meuleman and Billie (2009), a group sample size of 40 in the between-level factor structure is sufficient. Only 25 classes were involved the present study; therefore, we standardized all variables by using a class as a unit to prevent class clustering effects (Dowding & Haufe, 2018). The present study applied structural equation modeling to analyze the data by using LISREL 8.80 and SPSS for Windows 23.0. To assess the model fit, we used well-established indices such as root mean square error of approximation (RMSEA) < .10, standardized root mean residual (SRMR) < .05, Tucker-Lewis index (TLI), and comparative fit index (CFI) > .95 in addition to chi-square test statistics. Data were analyzed in two steps. After a preliminary statistical analysis, we first examined descriptive statistics, sex and grade differences, and measurement invariance. Second, we examined the model fit by using two strategies: First, the reciprocal relations models were tested competitively against two unidirectional models. If the reciprocal model exhibited a better fit with the data, then the model of the present study was supported. Second, we examined the model fit and effects of all variables. Before analyzing the model, we first analyzed sex and grade differences. No significant differences existed between boys and girls in T_2 learning engagement, T_3 enjoyment, or T_4 learning engagement. However, significant differences existed between boys and girls in T_1 enjoyment, boredom, and situational interest and in T_3 boredom and situational interest, but the effect size was small (η^2 = .01 - .02). Moreover, no significant differences existed between the seventh and eighth grades for all variables. Hence, we did not use sex as a control variable. Second, we tested measurement invariance, which included configural, metric (weak factorial), scalar (strong factorial), and covariance. The results indicated that the boredom scale had strong measurement invariance, Δχ^2 (dfs= 1-5, N = 600) = 0.40-5.30, p > .05, and the enjoyment, situational interest, and learning engagement scales had weak measurement invariance, Δχ^2 (dfs = 4-20, N = 600) = 2.21-23.62, p > .05. According to Putnick and Bornstein (2016), if measurement scales indicate invariance for all variables over time, then they are suitable for longitudinal studies. First, the results of the present study indicated that the academic emotions, situational interest, and learning engagement reciprocal effect model had an excellent fit with the data χ_s^2 (242, N = 600) = 1401.06, 1196.29, p < .05; RMSEAs = .093, .083; SRMRs = .047, .041; TLIs = .93, .97; CFIs = .98, .97. Second, the data analyses revealed reciprocal relations between enjoyment and learning engagement. Students' T_1 enjoyment enhanced their T_2 learning engagement. In turn, students' T_2 learning engagement positively predicted their T_3 enjoyment, and students' T_3 enjoyment enhanced their T_4 learning engagement. Third, reciprocal relations existed between boredom and learning engagement. Students' T_1 boredom reduced their T_2 learning engagement; in turn, students' T_2 learning engagement negatively predicted their T_3 boredom, and students' T_3 boredom reduced their T_4 learning engagement. Fourth, the data analyses revealed reciprocal relations between situational interest and learning engagement. Students' T_1 situational interest enhanced their T_2 learning engagement; in turn, students' T_2 learning engagement positively predicted their T_3 situational interest, and students' T_3 situational interest enhanced their T_4 learning engagement. To conclude, the present study demonstrated reciprocal relations between enjoyment/boredom and learning engagement as well as between situational interest and learning engagement. These findings expand the evidence base for the cognitive-motivational model and further understanding of the relationships among academic emotions, situational interest, and learning engagement.