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

Table 3 Bias in ${\hat{β}}_{3} = E (\hat{β_{1}}) - β_{1}$ , $β_{3} = 0$

Level of selection bias	Complete-case (Fixed-effects)	Complete-case (Mixed-effects)	Model 1	Model 2	Model 3
High	−0.021	0.001	0.056	1.203	0.299
Moderate	−0.019	−0.001	0.013	0.415	0.191
Low	−0.008	−0.0001	0.020	0.292	0.179

ISSN: 1478-7954