Sources of bias in outcome assessment in randomised controlled trials: a case study

Abstract

Randomised controlled trials (RCTs) can be at risk of bias. Using data from a RCT, we considered the impact of post-randomisation bias. We compared the trial primary outcome, which was administered blindly, with the secondary outcome, which was not administered blindly. From 44 schools, 522 children were randomised to receive a one-to-one maths tuition programme that was assessed using 2 outcome measures. The primary outcome measure was assessed blindly whilst the secondary outcome was delivered by the classroom teacher, and therefore this was unblinded. The effect sizes for primary and secondary outcomes were substantially different (0.33 and 1.11, respectively). Test questions that were similar between the 2 tests did not explain the difference. There was greater heterogeneity between schools for the primary outcome, compared with the secondary outcome. We conclude that, in this trial, the difference between the primary and secondary outcomes was likely to have been due to lack of blinding of testers.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 11 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.