Effects of a cognitive strategies intervention on secondary students' argument writing in history

867 students in grades 6-11; 12 randomly selected treatment teachers, 15 control teachers, 15 self-selected returning teachers

Grade 6-11 students in three districts in the southwestern United States

A yearlong, PD-delivered version of exactly what the disciplinary-literacy movement recommends: cognitive, sociocognitive and disciplinary strategies for source-based argument writing in history. The comparison structure is unusually informative because it separates two things that most studies of this kind conflate: teachers were RANDOMLY SELECTED into treatment (n = 12) or control (n = 15), while a third arm consisted of teachers who had already done a pilot year and CHOSE to continue (n = 15). Models controlled for student demographics, prior achievement and English-language status. Outcome is a scored argument-writing task - researcher-designed, and aligned to what the intervention teaches, which is the maximally favourable measure for this kind of programme. Read at abstract depth; the full article was not accessible.

No significant difference in student writing quality between the randomly assigned treatment and control teachers - a null on the arm with the credible comparison, on the intervention's own aligned outcome. The only significant effect (Cohen d = 0.29) was for students of the SELF-SELECTED teachers in their second year of the intervention, which is a comparison contaminated by both teacher self-selection and prior experience and cannot be read causally. Fidelity predicted writing improvement across the treatment and returning arms, which is the standard dose-response pattern and is likewise a post-randomisation contrast. The authors read the second-year result as evidence that the approach needs time to bed in; the archive's reading is that the one arm with random selection produced nothing, and the positive result sits exactly where selection effects would put it.

Low for the main contrast (teachers randomly selected, students controlled on prior achievement); HIGH for the second-year result, which compares students of teachers who volunteered to continue against students of teachers who did not - a selection contrast on both teacher and, plausibly, school characteristics.

Unreplicated. It is, however, the largest and best-identified test of PD-delivered historical argument writing located, and it converges with De La Paz et al. (2014), whose comparable intervention produced growth without a credible control contrast.