The nature of feedback: how different types of peer feedback affect writing performance

Although providing feedback is commonly practiced in education, there is no general agreement regarding what type of feedback is most helpful and why it is helpful. This study examined the relationship between various types of feedback, potential internal mediators, and the likelihood of implementing feedback. Five main predictions were developed from the feedback literature in writing, specifically regarding feedback features (summarization, identifying problems, providing solutions, localization, explanations, scope, praise, and mitigating language) as they relate to potential causal mediators of problem or solution understanding and problem or solution agreement, leading to the final outcome of feedback implementation. To empirically test the proposed feedback model, 1,073 feedback segments from writing assessed by peers was analyzed. Feedback was collected using SWoRD, an online peer review system. Each segment was coded for each of the feedback features, implementation, agreement, and understanding. The correlations between the feedback features, levels of mediating variables, and implementation rates revealed several significant relationships. Understanding was the only significant mediator of implementation. Several feedback features were associated with understanding: including solutions, a summary of the performance, and the location of the problem were associated with increased understanding; and explanations of problems were associated with decreased understanding. Implications of these results are discussed.