Fig. 2From: Managing missing items in the Fagerström Test for Nicotine Dependence: a simulation studyPlots of simulation results. For the large sample size (nobs = 788) with probability of subject-level missingness psub = 0.50, plots of estimates of the (A) percent sample size was reduced and (B-E) performance measures (mean with 95% Monte Carlo confidence interval) for each method against the probability of item-level missingness (pitem) for both missingness mechanisms (MAR and MNAR). CCA = Complete Case Analysis; FTND = Fagerström Test for Nicotine Dependence; HR = Half-Rule; MAR = missing at random; MNAR = missing not at random; regr. Coeff. = regression coefficient; SE = standard errorBack to article page