The Research on Effective Writing Instruction
Unlike reading, where decades of cognitive science research have established clear evidence-based practices, writing instruction has historically relied more on tradition and intuition. However, recent meta-analyses and randomised controlled trials have identified what actually improves student writing.
Graham & Perin's landmark 2007 meta-analysis Writing Next examined over 400 experimental studies and identified 11 evidence-based instructional practices. Strategy instruction (explicitly teaching the steps of the writing process) produces an effect size of 0.82, making it one of the strongest effects in education research. Other highly effective practices include summarisation (0.82), peer collaboration (0.75), setting product goals (0.70), and sentence combining (0.50).
Writing is cognitively the most demanding academic task students face. Unlike reading (which can proceed somewhat automatically once skills are established), writing requires simultaneous attention to content generation, organization, transcription, audience awareness, and genre conventions. This cognitive load explains why even articulate students who speak eloquently produce disorganized, error-filled writing—their working memory is overwhelmed.
At RootsTutorServices, we teach explicit, step-by-step strategies for each genre and stage of the writing process. For example, TREE (a Self-Regulated Strategy Development framework for opinion/argument writing): Topic sentence, Reasons, Explain, Ending. Students learn the strategy, practice it with support, and gradually internalize it until it becomes automatic.
We also teach that revision is re-vision—seeing the piece again with fresh eyes. Students learn to revise in multiple passes, focusing on different aspects each time: content revision, organization revision, style revision, and editing (last). This prevents cognitive overload and produces significantly better writing than single-pass editing.