Skip to main content
. 2021 Nov 14;23(2):e13369. doi: 10.1111/obr.13369

TABLE 2.

Odds for detecting a nominally statistically significant p value (p < 0.05 and p < 0.005) in a larger scale trial from the presence of a risk of generalizability bias

Risk of generalizability bias Odds ratio for p < 0.05 in larger scale trial Odds ratio for p < 0.005 in larger scale trial
OR (95% CI) % Effects OR (95% CI) % Effects
Delivery agent Not present Reference 34% Reference 23%
Pilot only 0.62 (0.43, 0.90) 24% 0.60 (0.39, 0.93) 15%
Both 1.39 (0.97, 2.00) 42% 0.96 (0.63, 1.47) 22%
Implementation support Not present Reference 34% Reference 22%
Pilot only 0.81 (0.42, 1.56) 30% 0.54 (0.22, 1.30) 13%
Both 0.74 (0.51, 1.08) 30% 1.03 (0.67, 1.57) 22%
Target audience Not present Reference 32% Reference 20%
Pilot only 0.72 (0.47, 1.11) 26% 1.11 (0.70, 1.75) 21%
Both 2.05 (1.35, 3.12) 49% 2.51 (1.58, 3.98) 34%
Setting Not present Reference 33% Reference 21%
Pilot only 0.17 (0.06, 0.47) 8% 0.22 (0.07, 0.70) 6%
Both 2.01 (1.29, 3.14) 48% 2.49 (1.54, 4.03) 35%
Intervention duration Not present Reference 34% Reference 23%
Pilot only 0.83 (0.61, 1.12) 32% 0.60 (0.42, 0.84) 17%
Intervention intensity Not present Reference 34% Reference 22%
Pilot only 0.06 (0.01, 0.46) 3% 0.12 (0.02, 0.90) 3%
Measurement Not present Reference 35% Reference 22%
Pilot only 0.86 (0.52, 1.41) 15% 0.93 (0.52, 1.65) 20%
Directional conclusions Not present Reference 34% Reference 23%
Pilot only 0.27 (0.13, 0.52) 32% 0.25 (0.10, 0.60) 8%

Note: Bolded values 95% confidence intervals (CIs) do not cross 1.00. Reference group is pilot/feasibility study and larger scale trial pairs without a risk of generalizability bias present.