Syntes: Pane et al. (2010) — Cognitive Tutor Geometry in Baltimore County: about 0.19 SD lower than the usual course

Källa: Pane, J. F., McCaffrey, D. F., Slaughter, M. E., Steele, J. L., & Ikemoto, G. S. (2010). An experiment to evaluate the efficacy of Cognitive Tutor Geometry. Journal of Research on Educational Effectiveness, 3(3), 254–281. https://doi.org/10.1080/19345741003681189

Verified numbers (full text read 2026-10-10, authors' manuscript)

• Setting: Baltimore County Public Schools (urban fringe), 8 high schools, 19 teachers, 2005–06 to 2007–08; no teacher had used the course before; 3–4 days of training.
• Design: each teacher taught one Cognitive Tutor class and one standard class; pupils scheduled in a period were individually randomised between the two teachers. 950 eligible randomised pupils; 699 in the final analysis (attrition 27.0 % vs 25.8 %); groups equivalent at baseline.
• Intervention: Cognitive Tutor Geometry course, 60 % teacher-led collaborative classroom work, 40 % individual work on the step-based tutoring software (model tracing, hints, mastery-based problem selection).
• Outcome: the district's mandatory geometry final exam (30 multiple-choice, 11 extended-response items; reviewed by the developer); pretest an ETS algebra test.
• Results (Table 3): geometry posttest −0.19 SD (SE 0.08, p = .03, 95 % CI −0.34 to −0.04), confirmed by a randomisation test; raw difference 2.2 points of 50; multiple-choice −0.23 (p < .01), open-ended −0.16 (p < .10). Attitudes −0.07 to −0.09 (ns). No moderation by school discipline, school proficiency, prior achievement or teacher experience with the course.
• Implementation: materials and software time implemented faithfully; learner-centred classroom practices low in both groups; observers often saw a third of pupils off task in group work.
• Authors: "we find a significant and substantial negative effect for the curriculum in our sample"; it "might not generalize to other student populations or to teachers who are more familiar with the curriculum".
• Note: this trial is included in Steenbergen-Hu & Cooper 2013 (their one geometry effect size, g = −0.19).

Role for C10

Limits (weaken, 2026-10-10): routine use as a whole course can lower results compared with the usual course; a single trial in one district, whole course rather than software alone.