Data set Option Effects on Grand Mean (GM) Effects on approximated Confidence Interval (CI)
Fig. 5 – MNAR 1) Complete-case analysis Increased in volatility with the percentage of missing data. Deviation increased non-linearly with the percentage of missing SDs for Hedges’ d. Increased non-linearly with the percentage of missing data.
2) Unweighted analysis Unbiased. Unbiased, except smaller CI for Fisher’s z.
3) SS-weighted analysis Unbiased. Unbiased.
4 -14) Imputations in general Random sample imputation Bayes predictive mean matching Unbiased for log response ratio. Slightly volatile for Fisher’s z. Deviation increased non-linearly with the percentage of missing SDs for Hedges’ d. - - Unbiased for log response ratio. Decreased with the percentage of missing data for Fisher’s z. Increased non-linearly from a threshold of 50-60% of missing data for Hedges’ d. Unbiased, except smaller for Hedges’ d. Increased non-linearly with the percentage of missing data.