How teacher-student collaborative assessment relates to L2 writing self-efficacy: insights from Chinese EFL learners

Teacher-student collaborative assessment (TSCA) has attracted increasing attention as a participatory assessment approach in L2 writing, yet its relationship with writing self-efficacy remains underexplored, particularly in secondary-level EFL contexts. This study aimed to examine whether TSCA was associated with changes in students' L2 writing self-efficacy and whether different patterns emerged across gender and internally defined pretest performance subgroups.

A quasi-experimental pretest-posttest design was implemented with two intact Grade 10 classes in a Chinese senior high school. One class received TSCA-based writing instruction, whereas the other followed a traditional teacher-centered assessment approach. Writing self-efficacy was measured using an adapted questionnaire, and supplementary qualitative data were collected through semi-structured interviews. Quantitative data were analyzed using descriptive statistics and t-tests, while interview data were examined through thematic analysis.

The experimental group showed stronger improvement patterns in overall writing self-efficacy than the control group, including improvement in both task self-efficacy and skill self-efficacy. Within the experimental group, no statistically significant gender difference was observed in the subgroup comparisons, whereas differences were observed across internally defined pretest performance subgroups, with the higher pretest-performing subgroup showing a stronger improvement pattern than the lower pretest-performing subgroup. The interview findings provided supplementary insight into how students perceived the TSCA process and how these perceptions differed across subgroups.

The findings suggest that, within the present instructional context, students in the TSCA group showed more favorable patterns of self-efficacy development than those in the comparison group. This pattern was observed in overall writing self-efficacy and in both subdimensions, although subgroup findings were exploratory and should be interpreted cautiously. Given the quasi-experimental design, the single-school sample, and the limited subgroup analyses, the results should be understood as context-bound rather than as strong causal evidence.