Trail Making Test Part A - Psychological Aggression
Model 1
options(scipen = 999)
TMTAlog1 <- glm(PsychAggress ~ TrailsA + WRATtotal, data=Allvars,family = "binomial")
summary(TMTAlog1)
##
## Call:
## glm(formula = PsychAggress ~ TrailsA + WRATtotal, family = "binomial",
## data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.1303 0.4823 0.5085 0.5449 1.0981
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.335665 0.655826 2.037 0.0417 *
## TrailsA -0.006319 0.002737 -2.309 0.0209 *
## WRATtotal 0.017840 0.014644 1.218 0.2231
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 505.40 on 640 degrees of freedom
## Residual deviance: 498.57 on 638 degrees of freedom
## AIC: 504.57
##
## Number of Fisher Scoring iterations: 4
round(exp(cbind(OR = coef(TMTAlog1), confint(TMTAlog1,level = 0.95))),digits=3)
## OR 2.5 % 97.5 %
## (Intercept) 3.803 1.084 14.311
## TrailsA 0.994 0.988 0.999
## WRATtotal 1.018 0.989 1.047
plot(predictorEffect("TrailsA",TMTAlog1))
########Compare to null model
#Difference in Deviance
with(TMTAlog1,null.deviance - deviance)
## [1] 6.831695
#Degrees of freedom for the difference between two models
with(TMTAlog1,df.null - df.residual)
## [1] 2
#p-value
with(TMTAlog1,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.03284855
#Pseudo R-Squared
nagelkerke(TMTAlog1)
## $Models
##
## Model: "glm, PsychAggress ~ TrailsA + WRATtotal, binomial, Allvars"
## Null: "glm, PsychAggress ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.0135174
## Cox and Snell (ML) 0.0106013
## Nagelkerke (Cragg and Uhler) 0.0194357
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -2 -3.4158 6.8317 0.032849
##
## $Number.of.observations
##
## Model: 641
## Null: 641
##
## $Messages
## [1] "Note: For models fit with REML, these statistics are based on refitting with ML"
##
## $Warnings
## [1] "None"
Model 3
options(scipen = 999)
TMTAlog3 <- glm(PsychAggress ~ (TrailsA + Sex + PovStat)^3 + Age + WRATtotal, data = Allvars, family = "binomial")
summary(TMTAlog3)
##
## Call:
## glm(formula = PsychAggress ~ (TrailsA + Sex + PovStat)^3 + Age +
## WRATtotal, family = "binomial", data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.3050 0.4145 0.4876 0.5662 0.9371
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.3232837 0.9521720 2.440 0.0147 *
## TrailsA -0.0046246 0.0044733 -1.034 0.3012
## SexMen -0.5352512 0.3757272 -1.425 0.1543
## PovStatBelow -0.3211181 0.9165205 -0.350 0.7261
## Age -0.0211628 0.0133563 -1.584 0.1131
## WRATtotal 0.0213502 0.0147965 1.443 0.1490
## TrailsA:SexMen 0.0002853 0.0061339 0.047 0.9629
## TrailsA:PovStatBelow 0.0121948 0.0262325 0.465 0.6420
## SexMen:PovStatBelow 1.1973680 1.0734939 1.115 0.2647
## TrailsA:SexMen:PovStatBelow -0.0212556 0.0279137 -0.761 0.4464
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 505.40 on 640 degrees of freedom
## Residual deviance: 489.18 on 631 degrees of freedom
## AIC: 509.18
##
## Number of Fisher Scoring iterations: 5
round(exp(cbind(OR = coef(TMTAlog3), confint(TMTAlog3,level = 0.95))),digits=3)
## OR 2.5 % 97.5 %
## (Intercept) 10.209 1.618 68.160
## TrailsA 0.995 0.987 1.006
## SexMen 0.586 0.277 1.222
## PovStatBelow 0.725 0.104 3.906
## Age 0.979 0.954 1.005
## WRATtotal 1.022 0.992 1.051
## TrailsA:SexMen 1.000 0.988 1.013
## TrailsA:PovStatBelow 1.012 0.970 1.077
## SexMen:PovStatBelow 3.311 0.451 31.397
## TrailsA:SexMen:PovStatBelow 0.979 0.918 1.026
########Compare to null model
#Difference in Deviance
with(TMTAlog3,null.deviance - deviance)
## [1] 16.21973
#Degrees of freedom for the difference between two models
with(TMTAlog3,df.null - df.residual)
## [1] 9
#p-value
with(TMTAlog3,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.06243246
#Pseudo R-Squared
nagelkerke(TMTAlog3)
## $Models
##
## Model: "glm, PsychAggress ~ (TrailsA + Sex + PovStat)^3 + Age + WRATtotal, binomial, Allvars"
## Null: "glm, PsychAggress ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.0320928
## Cox and Snell (ML) 0.0249863
## Nagelkerke (Cragg and Uhler) 0.0458083
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -9 -8.1099 16.22 0.062432
##
## $Number.of.observations
##
## Model: 641
## Null: 641
##
## $Messages
## [1] "Note: For models fit with REML, these statistics are based on refitting with ML"
##
## $Warnings
## [1] "None"
Compare Models 1 & 3
anova(TMTAlog1,TMTAlog3,test = "LR")
## Analysis of Deviance Table
##
## Model 1: PsychAggress ~ TrailsA + WRATtotal
## Model 2: PsychAggress ~ (TrailsA + Sex + PovStat)^3 + Age + WRATtotal
## Resid. Df Resid. Dev Df Deviance Pr(>Chi)
## 1 638 498.57
## 2 631 489.18 7 9.388 0.226
Trail Making Test Part A - Physical Assault
Model 1
options(scipen = 999)
TMTAlog1 <- glm(PhysAssault ~ TrailsA + WRATtotal, data=Allvars,family = "binomial")
summary(TMTAlog1)
##
## Call:
## glm(formula = PhysAssault ~ TrailsA + WRATtotal, family = "binomial",
## data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.5836 -0.5414 -0.5166 -0.4765 2.2735
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.982492 0.861673 -2.301 0.0214 *
## TrailsA -0.009439 0.008155 -1.157 0.2471
## WRATtotal 0.007887 0.017005 0.464 0.6428
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 482.53 on 640 degrees of freedom
## Residual deviance: 479.65 on 638 degrees of freedom
## AIC: 485.65
##
## Number of Fisher Scoring iterations: 6
round(exp(cbind(OR = coef(TMTAlog1), confint(TMTAlog1,level = 0.95))),digits=3)
## OR 2.5 % 97.5 %
## (Intercept) 0.138 0.026 0.764
## TrailsA 0.991 0.972 1.002
## WRATtotal 1.008 0.975 1.043
########Compare to null model
#Difference in Deviance
with(TMTAlog1,null.deviance - deviance)
## [1] 2.881826
#Degrees of freedom for the difference between two models
with(TMTAlog1,df.null - df.residual)
## [1] 2
#p-value
with(TMTAlog1,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.2367116
#Pseudo R-Squared
nagelkerke(TMTAlog1)
## $Models
##
## Model: "glm, PhysAssault ~ TrailsA + WRATtotal, binomial, Allvars"
## Null: "glm, PhysAssault ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.00597229
## Cox and Snell (ML) 0.00448574
## Nagelkerke (Cragg and Uhler) 0.00848053
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -2 -1.4409 2.8818 0.23671
##
## $Number.of.observations
##
## Model: 641
## Null: 641
##
## $Messages
## [1] "Note: For models fit with REML, these statistics are based on refitting with ML"
##
## $Warnings
## [1] "None"
Model 3
options(scipen = 999)
TMTAlog3 <- glm(PhysAssault ~ (TrailsA + Sex + PovStat)^3 + Age + WRATtotal, data = Allvars, family = "binomial")
summary(TMTAlog3)
##
## Call:
## glm(formula = PhysAssault ~ (TrailsA + Sex + PovStat)^3 + Age +
## WRATtotal, family = "binomial", data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.9896 -0.5444 -0.4687 -0.3914 2.4456
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.36348112 1.05751855 -1.289 0.1973
## TrailsA 0.00007282 0.00727697 0.010 0.9920
## SexMen 0.07629996 0.57331921 0.133 0.8941
## PovStatBelow 1.66166783 0.89445973 1.858 0.0632 .
## Age -0.02951309 0.01411907 -2.090 0.0366 *
## WRATtotal 0.01521473 0.01744790 0.872 0.3832
## TrailsA:SexMen -0.00802885 0.01525606 -0.526 0.5987
## TrailsA:PovStatBelow -0.04104160 0.02947504 -1.392 0.1638
## SexMen:PovStatBelow -1.19151471 1.15924755 -1.028 0.3040
## TrailsA:SexMen:PovStatBelow 0.04371608 0.03509486 1.246 0.2129
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 482.53 on 640 degrees of freedom
## Residual deviance: 466.00 on 631 degrees of freedom
## AIC: 486
##
## Number of Fisher Scoring iterations: 6
round(exp(cbind(OR = coef(TMTAlog3), confint(TMTAlog3,level = 0.95))),digits=3)
## Waiting for profiling to be done...
## OR 2.5 % 97.5 %
## (Intercept) 0.256 0.031 2.011
## TrailsA 1.000 0.978 1.011
## SexMen 1.079 0.378 4.055
## PovStatBelow 5.268 1.006 33.547
## Age 0.971 0.944 0.998
## WRATtotal 1.015 0.982 1.052
## TrailsA:SexMen 0.992 0.953 1.020
## TrailsA:PovStatBelow 0.960 0.900 1.010
## SexMen:PovStatBelow 0.304 0.029 3.170
## TrailsA:SexMen:PovStatBelow 1.045 0.975 1.127
########Compare to null model
#Difference in Deviance
with(TMTAlog3,null.deviance - deviance)
## [1] 16.52871
#Degrees of freedom for the difference between two models
with(TMTAlog3,df.null - df.residual)
## [1] 9
#p-value
with(TMTAlog3,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.05662862
#Pseudo R-Squared
nagelkerke(TMTAlog3)
## $Models
##
## Model: "glm, PhysAssault ~ (TrailsA + Sex + PovStat)^3 + Age + WRATtotal, binomial, Allvars"
## Null: "glm, PhysAssault ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.0342541
## Cox and Snell (ML) 0.0254562
## Nagelkerke (Cragg and Uhler) 0.0481263
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -9 -8.2644 16.529 0.056629
##
## $Number.of.observations
##
## Model: 641
## Null: 641
##
## $Messages
## [1] "Note: For models fit with REML, these statistics are based on refitting with ML"
##
## $Warnings
## [1] "None"
Compare Models 1 & 3
anova(TMTAlog1,TMTAlog3,test = "LR")
## Analysis of Deviance Table
##
## Model 1: PhysAssault ~ TrailsA + WRATtotal
## Model 2: PhysAssault ~ (TrailsA + Sex + PovStat)^3 + Age + WRATtotal
## Resid. Df Resid. Dev Df Deviance Pr(>Chi)
## 1 638 479.65
## 2 631 466.00 7 13.647 0.05783 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1