Thinking, Doing, Talking Science: Evaluation report and executive summary

42 schools recruited, 41 completed (21 intervention, 20 control); 655 Year 5 pupils analysed

Year 5 (age 9-10) pupils in English primary schools, 2013/14 academic year.

EFFICACY trial — the first leg of a three-trial chain and the one everybody quotes. Schools were matched into pairs on pre-test score, FSM, EAL and school size, then one of each pair was randomised to TDTS. The intervention is teacher CPD, not a curriculum: two teachers per school attended FIVE days of professional development delivered BY THE PROGRAMME DEVELOPERS (Science Oxford and Oxford Brookes), plus two funded in-school planning days, plus a preceding pilot year in the same schools. The pedagogy is guided, discussion-heavy inquiry — "Big Questions" ("How do you know that the earth is a sphere?"), Odd One Out, practical work, more talk and less writing-up — applied by teachers to their normal science lessons; it is not a scripted scheme and it is not open discovery. The outcome was a bespoke science test assembled from Russell and McGuigan's Science Assessment Series items, ADMINISTERED BY THE PARTICIPATING TEACHERS THEMSELVES (they did not see it before the day). EEF security: MODERATE, with the report explicitly noting the absence of a nationally recognised science assessment as the reason a bespoke test was used. Cost about GBP 26 per pupil. Note the trial was slightly underpowered as analysed: MDES 0.28 against an observed effect of 0.22.

Overall effect on the bespoke science test +0.22 (three months' additional progress). Girls +0.32 (four months), boys +0.12 (two months, chance not excluded). Low prior attainers +0.38 reported as four months, high prior attainers three months. FSM pupils +0.38 (five months), flagged as lower security because of small FSM numbers. Attitudes to science improved on most measures. The far-transfer test is the one usually omitted: when the trial cohort's Key Stage 2 results arrived a year later, there was NO effect on KS2 reading (ES -0.01, 95% CI -0.12 to 0.10) or KS2 maths (ES -0.06, -0.16 to 0.05); the evaluators write "the impact of TDTS on science failed to transfer into English and Maths". This +0.22, produced by developers training 42 self-selected schools and measured on a teacher-administered bespoke test, is the number that then collapsed twice under independent scale-up (see the 2018 and 2025 effectiveness trials).

Low. Pair-matched school randomisation with only one school lost; the exposure is outcome measurement, not selection.

FAILED at scale, twice. Kitmitto et al. 2018 (205 schools, 7,806 pupils, independent evaluators AIR, same science assessment) found +0.01. Robinson-Smith et al. 2025 (180 schools, 6,107 pupils, York Trials Unit, improved train-the-trainer model) found +0.02. The EEF's own scale-up review (Demack et al. 2021) lists TDTS as one of seven interventions whose mean primary-outcome effect fell from 0.25 (efficacy) to 0.01 (effectiveness).