A comparison of missing data procedures for addressing selection bias in HIV sentinel surveillance data

Table 7 Probability coverage of 95% confidence intervals for $β_{1}$ and $β_{3}$ for scenarios with differential trends in intervention and non-intervention sites (with a high level of selection bias)

True parameters	Complete-case (Fixed-effects)	Complete-case (Mixed-effects)	Model 1	Model 2	Model 3
Scenario 1
β ₁ = 0	0.183	0.924	0.049	0.039	0.715
β ₃ = –0.025	0.839	0.833	0.657	0.941	0.069
Scenario 2
β ₁ = –0.025	0.854	0.462	0.897	0.969	0.325
β ₃ = –0.025	0.757	0.349	0.541	0.897	0.088

ISSN: 1478-7954