Table 2. Estimated overall proportion of clustering for observed, incomplete-observed and imputed-complete datasets under MNAR.

Deletion weight	Clus, Obs	Clus, In-Obs	Clus, Impute-Com	Cov, In-Obs	Cov, Impute-Com
Proportional	0.043	0.018	0.022	8%	42%
Inversely proportional		0.043	0.053	80%	93%

Clustering assessed at the 0.10 threshold, with m = 100 deletions. Clus stands for clustering; Obs for observed data; In-Obs for ncomplete-observed data; Impute-Com for imputed-complete data; Cov stands for coverage.