Skip to main content
. 2022 Mar 17;13:856971. doi: 10.3389/fpsyg.2022.856971

Table 1.

Comparison of different linear mixed effects models.

Dependent variable/tested model df AIC logLik Cond.R2 Against p(χ2)
%correct
#0: null 9 35,483 −17,732 0.128
#1: + Mask 10 35,316 −17,648 0.168 #0 <0.0001
#2: + EI + SREIS 12 35,320 −17,648 0.168 #1 0.9081 n.s.
#3a: + FamiliarityOthers 11 30,243 −15,111 #1 <0.0001
#3b: + FamiliarityOwn 11 35,318 −17,648 0.168 #1 0.8360 n.s.
#4: + attitudeMasks 11 35,318 −17,648 0.168 #1 0.5975 n.s.
#5: + exprEmo:Mask 15 35,177 −17,527 0.224 #1 <0.0001
%confidence
#0: null 9 37,253 −15,317 0.240
#1: + Mask 10 30,246 −15,113 0.324 #0 <0.0001
#2: + EI + SREIS 12 30,324 −15,113 0.324 #1 0.9468 n.s.
#3a: + FamiliarityOthers 11 30,243 −15,111 0.324 #1 0.0325
#3a: + FamiliarityOwn 11 30,245 −15,111 0.324 #1 0.0838 n.s.
#4: + attitudeMasks 11 30,247 −15,113 0.324 #1 0.5115 n.s.

The table shows the results of linear mixed effects analysis of different models in comparison with less complex models, separated by the two tested dependent variables % correct (percentage of correct emotion classifications) and % confidence (for correct emotion classifications). FS, fixed slopes (fixed factors); RS, random slopes (random factors); df, degrees of freedom; R2, conditional coefficient of determination, based on the likelihood-ratio test; and “against” indicates the model against which the current model was tested, p(χ2) provides the probability of accepting a significant effect despite a nonexistent difference regarding the more complex model versus the model specified in the “against” column.