In the first and second selection steps, respectively 20 and 166 studies were excluded.
In the third step—searching for learning outcomes as an independent variable—666 studies
were removed from the selection. In step four, 265 studies were excluded. Finally, in the fifth
selection step, another 19 studies were removed from the selection. The final selection
contains 22 articles. The results of the selection process are summarised in Table 3.1.
3.3.3 Data Analysis
The analysis of the content took place for 18 of the 22 selected articles because four
studies were classified as being of low quality. Of these 18 studies (indicated with an asterisk
* in the reference list), only nine reported that one feedback condition was significantly
superior to another.6 Two studies investigated the effects of a CBA either with or without
feedback and a paper-and-pencil test (Wang, Wang, Wang, & Huang, 2006; Wang, 2007).
Post-test results of students in a CBA that included feedback were significantly higher than
the results of students in the other two conditions. However, there were six different types of
feedback included in the CBA, which implies that no effects can be assigned to a specific type
of feedback. Only seven out of 18 studies analysed provided us with usable significant results
regarding a specific way of providing feedback. An overview of the feedback interventions in
the different studies is provided in Table 3.2. This table also shows which studies reported
significant positive effects. Subsequently, the studies with significant effects are discussed.
The feedback types distinguished by Shute (2008) as described in Section 3.1.2 are
used to present the results of the selected studies in Table 3.2. In addition, additional feedback
characteristics are presented, because the EF interventions differ from one another. The KR
interventions were mostly combined with a try-again option. Also, the timing of feedback
(Shute, 2008) was taken into consideration, as was the level of feedback (Hattie & Timperley,
2007) and the level of learning outcomes (Bloom et al., 1956; Gagné, 1985).
6 For this study, the only methods considered were those whose feedback regarding student
achievement produced significant effects. Within the selected studies, some other significant effects
were found, such as the effects of different feedback types on student motivation or test anxiety.