Skip to main content
. Author manuscript; available in PMC: 2016 Feb 5.
Published in final edited form as: Psychol Aging. 2015 Sep 21;30(4):911–929. doi: 10.1037/pag0000046

Table 5.

MSE Ratios, Simulation A

Listwise
Log
CART
RF
MI
30% 50% 30% 50% 30% 50% 30% 50% 30% 50%
BX2Y1
 Linear   .869   .907   .898 1.101 1.007 1.104 1.027 1.500   .880   .922
 One split   .695   .837 1.198 1.011 1.006   .968 1.945 1.738   .682   .863
 Two splits   .803   .781   .958 1.132 1.087 1.071 1.660 2.203   .762   .832
 Three splits   .752   .777   .831   .816 1.081 1.021 1.124 1.574   .754   .810
BY2X1
 Linear   .999   .944   .975   .979 1.015   .998   .95 1.228 1.012   .860
 One split 1.054 1.043 1.021   .953   .994 1.034 1.162 1.250   .897   .939
 Two splits 1.201 1.232 1.249 1.234 1.063 1.023 1.120 1.236   .936   .932
 Three splits 1.117   .992 1.009   .986 1.039 1.002 1.096 1.372   .985   .892
BX2X1
 Linear 1.004   .971 1.006   .996 1.013 1.019 1.004 1.276 1.020   .985
 One split 1.004   .977 1.036   .967 1.012 1.008 1.287 1.100   .933 1.003
 Two splits 1.108 1.065 1.129 1.077 1.062 1.028   .989 1.266 1.007   .973
 Three splits 1.039   .972   .986   .968   .993   .981 1.012 1.070   .956   .994
BY2Y1
 Linear 1.009 1.022   .994 1.013 1.002 1.008 1.001 1.096   .984   .957
 One split 1.076   .987 1.035   .972   .995 1.020 1.070 1.116   .938   .947
 Two splits 1.151 1.202 1.156 1.198   .996 1.020   .969 1.243   .969   .966
 Three splits 1.090 1.014 1.018   .967 1.018   .999 1.017 1.225   .989   .971
rY2X2
 Linear   .983   .963   .997 1.042   .993 1.059 1.017 1.340   .961   .987
 One split   .861   .984   .951 1.056 1.039 1.007 1.209 1.295   .899   .995
 Two splits   .904   .869   .926 1.002 1.042   .986 1.138 1.266   .937   .924
 Three splits   .926   .898   .977   .943 1.013   .991 1.084 1.185   .936   .902
Resid (X2)
 Linear   .902   .901   .917 1.047 1.025 1.043 1.016 1.538   .944   .843
 One split   .733   .921   .930 1.092 1.060 1.019 1.847 1.470   .697   .855
 Two splits   .850   .697   .946   .961 1.062 1.034 1.332 2.128   .807   .684
 Three splits   .840   .754   .890   .751 1.051 1.068 1.131 1.367   .788   .746
Resid (Y2)
 Linear 1.004   .998 1.006 1.009 1.001 1.010 1.014 1.005 1.018 1.059
 One split   .962   .979   .965 1.018   .999 1.001 1.008   .963   .994 1.035
 Two splits 1.018   .959   .996   .963 1.005   .996   .975 1.044 1.054 1.030
 Three splits   .985   .964   .990   .970   .99   .999 1.000 1.030 1.004 1.022

Note. αX = αY = .9, N = 500. MSE ratio is computed as MSEMeasureX/MSEPrune. MSE = mean squared error; Log = logistic regression; CART = classification and regression trees; RF = random forests; MI = multiple imputation.