Research Proposal Example: A Full APA 7 Proposal, Annotated, With a Template

Structured Peer Feedback and Revision Quality in High School Writing Classrooms: A Proposal

Introduction and Problem Statement

High school writing teachers routinely assign peer feedback during the revision stage, on the assumption that reading a classmate's draft and commenting on it improves the reviewer's own writing as much as the writer's. That assumption rests on a body of feedback research built mostly around teacher-to-student and expert-to-novice feedback, not peer-to-peer exchange inside a single class period. Hattie and Timperley (2007) established that feedback's effect on learning depends heavily on whether it targets the task, the process, or the student's self-concept, and Kluger and DeNisi's (1996) meta-analysis found that a substantial share of feedback interventions measurably lower performance when they redirect attention toward the self rather than the task. Peer feedback, delivered by a novice reviewer under time pressure, is exactly the kind of intervention where that risk is highest and least studied. The research reviewed below does not isolate the effect of a structured peer feedback protocol, one that constrains reviewers to task- and process-level, criterion-referenced comments, on the revision quality of the writer receiving it, inside an ordinary high school English classroom rather than a controlled lab setting.

The question matters for classroom practice as much as for research. Peer review takes class periods that could go to direct instruction, so if unstructured peer feedback does little for revision, that time is poorly spent; if a one-page structured form improves revisions, any English teacher can adopt it after one short training session.

Purpose and Research Questions

This study examines whether a structured peer feedback protocol improves revision quality more than the unstructured peer feedback routine already common in the participating classrooms.

RQ1: Do students who receive structured, criterion-referenced peer feedback produce higher-quality revisions, as measured by a standardized revision rubric, than students who receive unstructured peer feedback?

RQ2: Does the effect of structured peer feedback differ between higher- and lower-proficiency writers, as identified by their pre-unit diagnostic writing score?

Theoretical Framework

The study is grounded in Hattie and Timperley's (2007) feedback model, which distinguishes feedback operating at the task level (is the work correct), the process level (what strategy produced it), the self-regulation level (how the learner monitors their own progress), and the self level (praise or criticism of the person). The model predicts that feedback aimed at the task and process levels produces stronger learning gains than feedback aimed at the self, and it supplies the basis for the study's structured protocol, which trains peer reviewers to write comments at the task and process levels only. Nicol and Macfarlane-Dick's (2006) seven principles of good feedback practice, developed from a model of self-regulated learning, list peer dialogue among the seven and support the argument that evaluating a classmate's work against an explicit criterion develops the reviewer's own internal standard — a mechanism this study treats as a reason structured peer feedback might improve outcomes for the reviewer as well as the writer.

Literature Review

Black and Wiliam's (1998) review of classroom-based formative-assessment studies found consistent learning gains when feedback told students not just whether an answer was correct but what to do next, with larger gains for lower-achieving students. That finding was later formalized in Hattie and Timperley's (2007) three-question model (where am I going, how am I going, where to next) and their four-level framework separating feedback about the task, the process, self-regulation, and the self.

Kluger and DeNisi's (1996) meta-analysis complicates any assumption that more feedback simply helps: across 607 effect sizes, more than a third of feedback interventions measurably decreased performance, an effect the authors traced to feedback that redirected attention toward the self rather than the task. Read alongside Hattie and Timperley (2007), the implication for peer feedback is direct: an untrained peer reviewer, without a structure forcing task-level comments, is plausibly more likely to produce exactly the kind of self-directed, evaluative comment ("this is confusing," "good job") that Kluger and DeNisi's data associates with weaker, not stronger, outcomes.

Nicol and Macfarlane-Dick (2006) supply the mechanism connecting structured feedback to self-regulated learning specifically: reviewing a peer's work against an explicit, shared criterion requires the reviewer to internalize that criterion, which their model treats as a driver of the reviewer's own later self-assessment. Their seven principles argue for feedback that is criterion-referenced and that positions the student as an active evaluator, not only a passive recipient — a description that fits a structured peer-review protocol more naturally than an unstructured one.

What these four sources do not resolve, because none tests it directly, is whether structuring peer feedback at the task and process levels measurably improves the writer's revision quality in an ordinary secondary classroom, whether it primarily benefits the reviewer's own later writing, or both. That comparison — structured against unstructured peer feedback, measured on the writer's revision rather than only on attitudes toward the process — is the gap this study addresses.

Methodology

Design. This study uses a quasi-experimental, pretest-posttest comparison design across four intact sections of tenth-grade English at a single comprehensive high school. Two teachers each teach two of the sections, and a coin flip before the unit decides which of each teacher's sections follows the structured peer feedback protocol and which keeps the school's existing unstructured peer feedback routine. Because each teacher delivers both conditions, a difference between conditions cannot simply reflect a difference between teachers, and the sections' prior-unit grade distributions are compared before the unit to check that the groups start from similar places.

Participants. Participants are approximately 96 tenth-grade students (four sections of roughly 24), all enrolled in the same required English course and completing the same persuasive-essay unit. Consent follows the district's standard parental-consent and student-assent procedure for classroom-embedded research; no student is excluded from the instructional activity itself, only from having their revision scored as study data.

Instruments. Revision quality is scored using an adapted six-trait rubric (organization, evidence, clarity, mechanics, voice, and use of feedback), scored 1-4 per trait by two raters blind to condition, with inter-rater reliability established on a 20% double-scored sample before full scoring begins. The pre-unit diagnostic is the department's existing timed persuasive-writing task, scored by the same raters on the first five traits, since use of feedback does not apply to a draft written without feedback. Peer comments from both conditions are also scored, using a feedback-quality rubric adapted from Nicol and Macfarlane-Dick's (2006) principles, and each comment is coded by Hattie and Timperley's (2007) four levels, to confirm the manipulation actually changed what reviewers wrote.

Procedure. Both conditions complete the same first-draft persuasive essay, then exchange drafts with one assigned peer. The unstructured condition receives the department's standard instruction ("read your partner's draft and leave helpful comments"). The structured condition uses a criterion-referenced peer-review form requiring one task-level comment and one process-level comment for each of the five traits that apply to a first draft, with no open-ended praise or criticism field. Both conditions get the same class period for the exchange and the same two days to revise, so only the form of the feedback differs. All students then revise individually before the final draft is collected and scored. The researcher observes one peer-review session in each section with a short checklist (time given, instructions read as scripted, forms completed) to confirm that each condition ran as designed.

Analysis Plan. RQ1 is tested with an ANCOVA comparing final-draft revision-rubric scores between conditions, controlling for pre-unit diagnostic writing score. RQ2 is tested by adding a condition-by-proficiency interaction term to the same model, using a median split on the diagnostic score to define higher- and lower-proficiency groups. Inter-rater reliability is reported as a weighted kappa before any hypothesis test is run. Results are reported as adjusted mean differences with 95% confidence intervals, not only as p-values, and teacher enters both models as a covariate. Students who miss either the first draft or the revision are excluded from the analysis, and the number excluded is reported by condition.

Limitations and Delimitations. The study is delimited to one grade, one school, one genre and one essay unit, so its results describe revision in that setting only. Its main limitation is scale: with four intact sections, the design cannot fully separate a condition effect from a section effect, so a positive result is evidence for a larger multi-school trial rather than a final answer. Because every student both gives and receives feedback, the design also cannot separate the benefit of receiving structured comments from the benefit of writing them. The median split in RQ2 makes the result easy to read but discards part of the information in the diagnostic score.

Ethics. The study is submitted to the district's research-review process and the researcher's institutional review board before data collection begins. Because both conditions are forms of normal classroom instruction, the board may classify the study as exempt research on normal educational practices or review it on an expedited basis; that classification is the board's decision, not the researcher's. Drafts are stored under study codes, and raters never see student names.

Timeline

Scroll the table sideways to see every column.

References

Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education: Principles, Policy & Practice, 5(1), 7-74. https://doi.org/10.1080/0969595980050102

Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81-112. https://doi.org/10.3102/003465430298487

Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119(2), 254-284. https://doi.org/10.1037/0033-2909.119.2.254

Nicol, D. J., & Macfarlane-Dick, D. (2006). Formative assessment and self-regulated learning: A model and seven principles of good feedback practice. Studies in Higher Education, 31(2), 199-218. https://doi.org/10.1080/03075070600572090

Margin noteNote 1 (framework tied to design): The framework isn't just cited, it's operationalized: "task and process level comments only" becomes the actual rule the peer-review form enforces later, in the Procedure section.

Margin noteNote 2 (synthesis across sources): The second paragraph reads Kluger and DeNisi's meta-analysis against Hattie and Timperley's model and draws both into a single argument about risk, rather than summarizing each finding in its own paragraph; the third ties Nicol and Macfarlane-Dick to the structured protocol that answers that risk. That merge is what synthesis means.

Margin noteNote 3 (gap statement): The gap is one specific missing comparison, not a general "more research is needed" line.