Figure 5.

Evaluations of regression coefficient and regression p-value accuracy. Methods are listed across the x-axis. The differences between the regression coefficients (or p-values) computed on the complete data and the imputed data are shown on the y-axis. The top row demonstrates differences in regression coefficients evaluated in the baboon plasma samples for metabolites with <10%, 10–20%, and 20–30% missingness. The top row demonstrates differences in regression p-values evaluated in the baboon plasma samples for metabolites with <10%, 10–20%, and 20–30% missingness. The center line represents the median. The lower and upper box limits represent the 25% and 75% quantiles, respectively. The whiskers extend to the largest observation within the box limit ± 1.5 × interquartile range. Black dots represent single iterations of evaluating differences that are outliers. The p-values are based on pairwise testing with the Wilcoxon Rank Sum test (* corresponds to p ≤ 0.05).