The feedback given by teachers to students is one of the most important and frequent instructional practices in the classroom, which would affect the individual’s learning motivation, emotions, strategies, behavior and performance. The concept of feedback (i.e., evaluation) is conceived as information about a students’ performance on a leaning task or a test (Pekrun et al., 2014), which conveying different types of messages, such as performance improvement vs. relative performance (Butler, 1987; Steele-Johnson et al., 2008), success vs. failure feedback (Senko & Harackiewicz, 2005), temporal evaluation vs. normative evaluation (Butler, 2006), and self-referential feedback vs. normative feedback (Pekrun et al., 2014). So far, a number of studies have addressed which types of feedback students received after task engagement influences on their achievement goals and achievement-relevant outcomes (e.g., Butler, 1987, 2006; Pekrun et al. 2014; Senko & Harackiewicz, 2005). However, a few studies have examined the impact of multiple feedback on students’ learning motivations and outcomes except for the synthesize-referenced feedback (combining effects of normative and self-referential feedback) proposed by Ju and Lin (2011); in addition, the impact of feedback on both students’ expectancy and value belief has been neglected. Accordingly, based on the view of multiple goal on achievement goal theory (Pintrich, 2000), the expectancy-value theory (Eccles et al., 1983), and work by Ju and Lin, we proposed a framework linking different kinds of feedback to learning motivations and performance. More specifically, we addresses self-referential feedback based on student’s improvement of performance over time, normative feedback based on comparing a students’ performance with the performance of other students, and multiple-referential feedback based on the combined characteristics of the above two feedbacks. It posited that these kinds of feedback influence students’ expectancy-values beliefs and academic performance. These hypotheses were examined in a teaching experimental study with junior high school students. 

The aims of this study was to test the impacts of different kind of feedback instructions on Taiwanese junior high school seven-grade students’ expectancy-value beliefs and performance in the subject of civil ethics. Correspondingly, two research questions guided the study: (1) do different kinds of feedbacks given by teachers to students influence on their expectancy and value belief? (2) and academic performance? Moreover, which one have the most promotion effect on these beliefs and performance? For research question 1, we hypothesized that students who are given experimental manipulations will report holding higher level of learning motivation (both in terms of expectancy and value beliefs) compared to students who will not receive any feedback (i.e., control group). Among them, it is further posited that the group of multiple-referential feedback has the most effect on these belief; in addition, the expectancy and value belief held by control group will decline over time. For research question 2, we hypothesized that students who receive experimental manipulations will have better performance in the second and the third regular examinations respectively compared to control group after controlling for their performance of the first regular examination. Similarly, it is further posited that the group of multiple-referential feedback has the best effect on these performance. 

For expectancy-value beliefs, 4 (group) × 3 (measurement wave) two-way mixed experimental design is conducted. The first independent variable is group including four levels: self-referential feedback, normative feedback, multiple-referential feedback, and control group (no feedback) as the between-subject factor; the second independent variable is measurement waves including three measurements as the within-subject factor. Dependent variables are students’ expectancy-value beliefs including self-efficacy, interest, utility, and attainment. For performance, nonequivalent pretest-posttest-delay posttest design is conducted. The independent variables include group (including three manipulation groups and control group) and measurement wave (including three measurements); and dependent variables are students’ performances of the second and the third regular examinations. 

A convenience sample was involved. A total of 170 seven-grades Taiwanese junior high school students (60 females and 60 males at ages 13-14) from eight classes with 1 school in Tainan City voluntarily participated (upon written parental consent) in this study after the study was approved by the University Ethics Committee. The period of the teaching experiment executed was the first semester of the 107 academic year (from September 2018 to January 2019) including six sessions: firstly, all participants completed the first measurement instruments (including perceived teacher feedback and expectancy-value belief questionnaires) one week before the first regular examination; secondly, each experimental group received four manipulations via quizzes during the first to the second examinations; thirdly, all participants completed the second measurement instruments one week before the second regular examination; fourthly, each experimental group received one manipulations via the second regular examination; fifthly, each experimental group received four manipulations via quizzes during the second to the third examinations; finally, all participants completed the third measurement instruments one week before the third regular examination. The data of students’ expectancy-value beliefs and academic performance were collected respectively before and after the implementation of the three regular examinations. 

We used SPSS Statistics Version 17.0 for Windows in all the analyses. Firstly, we conducted a 4 × 3 two-way mixed ANOVA analysis for the first research question; and one-way ANCOVA for the second research question. 

On the whole, the results showed that the various types of feedback exerted different impacts on students’ expectancy-value belief and academic performance. For the first research question, results revealed that there were significant two-way interaction effects between groups and measurement waves on self-efficacy and attainment-and-utility. More specifically, the normative and multiple-referential feedback groups had a greater effect on students’ self-efficacy, and the scores of the two experimental groups in the third measurement were significantly higher than those in the first measurement and the control group; however, as hypothesized, students’ self-efficacy of control group in the third measurement was lower than that in the first measurement, implying that students’ self-efficacy of control group gradually declined over time. Moreover, the multiple-referential feedback had a positive influence on attainment-and-utility value. That is, students’ attainment-and-utility value were enhanced by multiple-referential feedback in the second measurement and its positive effect remained in the third measurement as the number of manipulations increased. However, as our expectation, students’ attainment-and-utility value among the other three groups gradually decreased over time. For the second research question, this study found that after excluding the students’ performance of the first regular examination, their performance of the second regular examination of self-referential feedback group was significantly higher than those of multiple-referential feedback group and the control group, but there was no difference from the normative feedback group. Nevertheless, with the increase of feedback manipulation time, students’ performance of the third regular examination among all the three experimental groups was significantly higher than that of control group, but differences were not found among the three experimental groups. 

What kind of feedback promotes students’ expectancy-value beliefs and academic performance? Results disclosed that different types of feedbacks exerts distinct impacts on students’ learning motivations and academic performance. Specifically, normative and multiple-referential feedbacks had the most effects on promoting students’ self-efficacy; multiple-referential feedback exerted the best promotion on students’ attainment-and-utility value; and all the three experimental groups had positive impacts on students’ academic performance. Results have theoretical and applied implications for understanding and improving feedback practices and students’ motivations and performance.