Table 1.

Study characteristics and findings of randomized controlled trials evaluating the Hawthorne effect

Characteristic	Granberg and Holmberg [27]	Van Rooyen et al. [28]	Feil et al. [29]	O'Sullivan et al. [30]	Kypri et al. [31]	Kypri et al. [32]	McCambridge and Day [33]	Evans et al. [34]
Population	Swedish general population	Academic peer reviewers	Adolescent dental patients	Colorectal cancer screening population	University students	University students	University students	Male prostate cancer-testing population
Setting	Community	Correspondence	Dental clinic	Community	Student health service	Student health service	Student union	General practice initiated internet
Operationalization of HE	Being interviewed before an election	Awareness of study participation	Participation in experimental arm of clinical trial	Being given a questionnaire with screening kit	Completing a questionnaire	Completing a questionnaire	Completing a questionnaire on alcohol	Completing a questionnaire
Comparison group	Being interviewed after an election	No awareness of study participation	Usual care	No questionnaire with screening kit	No questionnaire	No questionnaire	No questionnaire on alcohol	No questionnaire
Participant blinding	Not clear if aware of outcome assessment	Yes. Control group blinded to study conduct	Yes. HE group blinded to study purpose. Control group blinded to all aspects of study participation.	Not clear if aware of outcome assessment	Yes. Both groups blinded to conduct of trial and study purpose	Yes. Both groups blinded to conduct of trial and study purpose.	Yes. Both groups blinded to study purpose and focus on drinking, HE group capable of inferring the latter	Not clear if aware of outcome assessment
Outcome measure	Objectively ascertained voting records	Quality of reviews produced.	Objectively ascertained plaque scores	Uptake of screening ascertained in records	4 Self-reported health behaviors	Self-reported drinking and related problems	Self-reported drinking and related problems	Uptake of prostate cancer test in medical records
Follow-up intervals	Not reported	Not reported	3 and 6 mo	6 wk and 6 mo	6 wk	6 and 12 mo	2-3 mo	6 mo
Sample size	Preelection interview: 4,720, postelection interview: 4,999	316 unaware, 149 aware	40, 20 per group	1,944 sent a questionnaire, 10,413 not sent one	74 completed a questionnaire, 72 did not	147 completed a questionnaire, 146 did not	217 completed a questionnaire, 204 did not	150 per group
Attrition	None	None	Two lost to follow-up (one in each group)	None	86%, Not differential by group	84% and 86% not differential by group	77%, Not differential by group	83%, Not differential by group
Summary of reported findings	People interviewed before the election were more likely to vote (95% vs. 93%), and this effect was stronger for those with low political interest (93% vs. 90%)	No evidence of any difference	Large between-group differences in plaque score at both 3 (54 ± 13.79 vs. 78 ± 12.18) and 6 mo (52 ± 13.04 vs. 79 ± 10.76)	Small statistically significant differences in uptake at 6 wk (54.4% vs. 51.9%), no longer significant at 6 mo (64.7% vs. 62.9%)	No differences detected	No differences detected at 6 mo, 3 of 7 statistically significant differences in outcomes at 12 mo	Small statistically significant differences in 4 of 9 outcomes including primary outcome (0.23 standard deviations)	Those completing questionnaire more likely to undergo test (11 of 123 vs. 2 of 126)
Reviewer comments including on principal risks of bias	Completed interviews analyzed, not ITT. Various other sample refinements. Not a formal research report, details of study design and methods unclear.	Validated outcome measure completed by author. Potential for information bias unclear.	Family of three randomized ad hoc as cluster. Small study.	Sequence generation for "every sixth person" not described. Effects of reminder letters not reported.	Self-report.	Self-report. Reasons for lack of effect at 6 mo unclear.	Self-report. Attrition.	Two of four arms in online trial evaluating a decision support aid described. Limited information in report. Small numbers.

Abbreviation: HE, Hawthorne effect.