Teacher adaptation of document-based history curricula: results of the Reading Like a Historian curriculum-use survey

1,927 teachers completed the survey (item-level n from 1,596 to 1,927); 1,320 answered the open-response item; 130 (7%) uploaded 215 modified artefacts

Self-selected users of the free online Reading Like a Historian curriculum, predominantly US history/social studies teachers; 80% White, 87% traditionally certified, 60% history majors or minors

Not an outcome study - a descriptive one-month online survey of who uses Reading Like a Historian and what they change about it. Included here because it is the only systematic evidence on what the intervention actually looks like once it leaves the trial, and the trial evidence (Reisman 2012) covers five teachers who got four days of training and twice-weekly researcher observation, while this covers roughly 1,900 teachers most of whom got nothing. Graded D: self-reports, non-random sample, no observation of classrooms, no student outcomes, and the artefact sub-sample is doubly self-selected. The authors list all four limitations themselves. Run by the curriculum's developers.

The at-scale user is untrained and the lesson survives adaptation. Fewer than 20% of respondents had ever taken any professional development on the curriculum, yet nearly a quarter used the materials weekly, over 80% at least once or twice a month, and over 65% had been using them for two to three years - so the modal real-world implementation is a free download used without training, which is exactly the condition the efficacy trial did not test. 86% had modified at least one lesson, 45% five or more, 24% every lesson they had used; 14% had never modified one. The modifications were consistently surface-level: teachers were LEAST likely to change the central historical question or the document set - the two features that constitute the document-based lesson - and MOST likely to add activities, build assessments, revise the background-information slides and rewrite discussion questions. 34% of stated reasons for modifying concerned students' reading needs (over a third of those specifically for low-skilled readers, English learners or students with special needs); time pressure accounted for under 10%, and district/state accountability pressure barely featured. The authors read this as "curricular fit": the structure is robust to teacher adaptation because teachers adapt around it rather than into it. The honest counter-reading is that the survey cannot tell whether the preserved structure was ever enacted - Reisman 2012 observed that even trained, supported teachers rarely ran the whole-class discussion the model calls for.

Not applicable - no student outcomes are estimated.

not-applicable - a one-off descriptive survey with no comparison and nothing to replicate. Its function in this cluster is to establish the gap between the evaluated version of the intervention (trained, supported, observed teachers) and the distributed one (untrained downloaders), which is where the efficacy-to-effectiveness decay would be expected to bite.