= Inverse-observation-weighted estimating equations for longitudinal dropout
{title2=$\sum_{i,j}w_{ij}\frac{R_{ij}}{\rho_{ij}}(Y_{ij}-m_{ij})=0$}
For monotone <missing data>, let $R_{ij}$ indicate observation and let $\rho_{ij}$ be the product of the conditional retention probabilities through visit $j$. Under sequential <missing at random>, conditioning on the complete response history makes these retention probabilities depend only on the history already observed at each decision. Therefore $\mathbb E(R_{ij}/\rho_{ij}\mid Y_i,X_i)=1$. Weighting a working-independence <generalized estimating equation> by $R_{ij}/\rho_{ij}$ restores its mean-zero property. This needs correct retention probabilities, positivity and regularity; uncertainty must account for estimating the weights.
Back to article page
