2,009
Views
3
CrossRef citations to date
0
Altmetric
Research Article

Improved methods for estimating fraction of missing information in multiple imputationFootnote1

& | (Reviewing editor)
Article: 1551504 | Received 26 Jul 2018, Accepted 18 Nov 2018, Published online: 13 Dec 2018

Figures & data

Table 1. Changes of E(γm), Dγ, and R with the increase of m at different γ0 levels, where Dγ = 100(E(γm)−γ0)/γ0 and R = 100(γmγm+1)/γm+1

Figure 1. The mE(γm) relationship curve at γ0 = 0.2 and 0.15 as determined by Equation (14).

Figure 1. The m–E(γm) relationship curve at γ0 = 0.2 and 0.15 as determined by Equation (14).

Figure 2. Effects of γ0 levels on Dγ as defined by Equation (15) and R as defined by Equation (16): a. Dγ at m = 5; b. Dγ at m = 2.

Figure 2. Effects of γ0 levels on Dγ as defined by Equation (15) and RDγ as defined by Equation (16): a. Dγ at m = 5; b. Dγ at m = 2.

Figure 3. Effects of m on γm at δ = 29% for analytic model = Anal-2: a. MI model = MI-1; b. MI model = MI-2.

Figure 3. Effects of m on γm at δ = 29% for analytic model = Anal-2: a. MI model = MI-1; b. MI model = MI-2.

Table 2. Coefficient of variations (%) of Bm, Um, and γm for SIZE100

Table 3. Comparison of different γ0 estimation methods for SIZE20 with imputation model = MI-2 and analytic model = Anal-2 in the PWS12 MI trials. The best γ̂0 was calculated by Equation (19) using (100)/(100) as the estimate of B0/U0, where 100 and 100 were the mean of the 30 replicates of B100 and U100, respectively