Digit Span Backward - Psychological Aggression
Model 1
DSBlog1 <- glm(PsychAggress ~ DigitSpanBck + WRATtotal, data=Allvars,family = "binomial")
summary(DSBlog1)
##
## Call:
## glm(formula = PsychAggress ~ DigitSpanBck + WRATtotal, family = "binomial",
## data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.1328 0.4853 0.5220 0.5539 0.7214
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.93837 0.63316 1.482 0.138
## DigitSpanBck 0.03600 0.06181 0.582 0.560
## WRATtotal 0.01668 0.01691 0.987 0.324
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 505.40 on 640 degrees of freedom
## Residual deviance: 502.89 on 638 degrees of freedom
## AIC: 508.89
##
## Number of Fisher Scoring iterations: 4
exp(cbind(OR = coef(DSBlog1), confint(DSBlog1)))
## OR 2.5 % 97.5 %
## (Intercept) 2.555822 0.7637836 9.219778
## DigitSpanBck 1.036652 0.9202872 1.173106
## WRATtotal 1.016825 0.9831461 1.050716
########Compare to null model
#Difference in Deviance
with(DSBlog1,null.deviance - deviance)
## [1] 2.512546
#Degrees of freedom for the difference between two models
with(DSBlog1,df.null - df.residual)
## [1] 2
#p-value
with(DSBlog1,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.2847132
#Pseudo R-Squared
nagelkerke(DSBlog1)
## $Models
##
## Model: "glm, PsychAggress ~ DigitSpanBck + WRATtotal, binomial, Allvars"
## Null: "glm, PsychAggress ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.00497139
## Cox and Snell (ML) 0.00391206
## Nagelkerke (Cragg and Uhler) 0.00717210
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -2 -1.2563 2.5125 0.28471
##
## $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
DSBlog3 <- glm(PsychAggress ~ (DigitSpanBck + Sex + PovStat)^3 + Age + WRATtotal, data = Allvars, family = "binomial")
summary(DSBlog3)
##
## Call:
## glm(formula = PsychAggress ~ (DigitSpanBck + Sex + PovStat)^3 +
## Age + WRATtotal, family = "binomial", data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.3576 0.3941 0.4862 0.5675 0.9101
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.86140 1.04549 1.780 0.0750 .
## DigitSpanBck 0.09323 0.11541 0.808 0.4192
## SexMen -0.62912 0.78860 -0.798 0.4250
## PovStatBelow 1.65113 1.19085 1.387 0.1656
## Age -0.02285 0.01323 -1.728 0.0841 .
## WRATtotal 0.01775 0.01714 1.036 0.3003
## DigitSpanBck:SexMen 0.01376 0.13382 0.103 0.9181
## DigitSpanBck:PovStatBelow -0.27451 0.19329 -1.420 0.1555
## SexMen:PovStatBelow 0.18041 1.54837 0.117 0.9072
## DigitSpanBck:SexMen:PovStatBelow 0.03102 0.24940 0.124 0.9010
## ---
## 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: 489.29 on 631 degrees of freedom
## AIC: 509.29
##
## Number of Fisher Scoring iterations: 5
exp(cbind(OR = coef(DSBlog3), confint(DSBlog3)))
## OR 2.5 % 97.5 %
## (Intercept) 6.4327376 0.85524267 52.076797
## DigitSpanBck 1.0977113 0.88332354 1.389694
## SexMen 0.5330596 0.11227394 2.487986
## PovStatBelow 5.2128734 0.53055769 58.665242
## Age 0.9774056 0.95217978 1.002976
## WRATtotal 1.0179052 0.98383788 1.052385
## DigitSpanBck:SexMen 1.0138573 0.77627240 1.314001
## DigitSpanBck:PovStatBelow 0.7599464 0.51955503 1.116229
## SexMen:PovStatBelow 1.1977140 0.05586013 24.916160
## DigitSpanBck:SexMen:PovStatBelow 1.0315050 0.63172517 1.688070
########Compare to null model
#Difference in Deviance
with(DSBlog3,null.deviance - deviance)
## [1] 16.11477
#Degrees of freedom for the difference between two models
with(DSBlog3,df.null - df.residual)
## [1] 9
#p-value
with(DSBlog3,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.06452314
#Pseudo R-Squared
nagelkerke(DSBlog3)
## $Models
##
## Model: "glm, PsychAggress ~ (DigitSpanBck + Sex + PovStat)^3 + Age + WRATtotal, binomial, Allvars"
## Null: "glm, PsychAggress ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.0318851
## Cox and Snell (ML) 0.0248267
## Nagelkerke (Cragg and Uhler) 0.0455156
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -9 -8.0574 16.115 0.064523
##
## $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(DSBlog1,DSBlog3,test = "LR")
## Analysis of Deviance Table
##
## Model 1: PsychAggress ~ DigitSpanBck + WRATtotal
## Model 2: PsychAggress ~ (DigitSpanBck + Sex + PovStat)^3 + Age + WRATtotal
## Resid. Df Resid. Dev Df Deviance Pr(>Chi)
## 1 638 502.89
## 2 631 489.29 7 13.602 0.05873 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Digit Span Backward - Physical Assault
Model 1
DSBlog1 <- glm(PhysAssault ~ DigitSpanBck + WRATtotal, data=Allvars,family = "binomial")
summary(DSBlog1)
##
## Call:
## glm(formula = PhysAssault ~ DigitSpanBck + WRATtotal, family = "binomial",
## data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.6479 -0.5320 -0.5020 -0.4776 2.2345
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.480502 0.722952 -3.431 0.000601 ***
## DigitSpanBck 0.064338 0.059616 1.079 0.280490
## WRATtotal 0.003327 0.018664 0.178 0.858527
## ---
## 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: 480.70 on 638 degrees of freedom
## AIC: 486.7
##
## Number of Fisher Scoring iterations: 4
exp(cbind(OR = coef(DSBlog1), confint(DSBlog1)))
## OR 2.5 % 97.5 %
## (Intercept) 0.08370116 0.01899563 0.3256606
## DigitSpanBck 1.06645295 0.94728347 1.1973200
## WRATtotal 1.00333240 0.96802001 1.0416671
########Compare to null model
#Difference in Deviance
with(DSBlog1,null.deviance - deviance)
## [1] 1.829701
#Degrees of freedom for the difference between two models
with(DSBlog1,df.null - df.residual)
## [1] 2
#p-value
with(DSBlog1,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.4005766
#Pseudo R-Squared
nagelkerke(DSBlog1)
## $Models
##
## Model: "glm, PhysAssault ~ DigitSpanBck + WRATtotal, binomial, Allvars"
## Null: "glm, PhysAssault ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.00379187
## Cox and Snell (ML) 0.00285038
## Nagelkerke (Cragg and Uhler) 0.00538879
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -2 -0.91485 1.8297 0.40058
##
## $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
DSBlog3 <- glm(PhysAssault ~ (DigitSpanBck + Sex + PovStat)^3 + Age + WRATtotal, data = Allvars, family = "binomial")
summary(DSBlog3)
##
## Call:
## glm(formula = PhysAssault ~ (DigitSpanBck + Sex + PovStat)^3 +
## Age + WRATtotal, family = "binomial", data = Allvars)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8330 -0.5503 -0.4685 -0.3884 2.4253
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.255613 1.120611 -1.120 0.2625
## DigitSpanBck 0.044520 0.107315 0.415 0.6782
## SexMen -0.455390 0.908308 -0.501 0.6161
## PovStatBelow 0.408942 1.027509 0.398 0.6906
## Age -0.031862 0.013778 -2.313 0.0207 *
## WRATtotal 0.009189 0.019285 0.476 0.6337
## DigitSpanBck:SexMen 0.038519 0.133091 0.289 0.7723
## DigitSpanBck:PovStatBelow 0.008996 0.165178 0.054 0.9566
## SexMen:PovStatBelow 0.440727 1.437951 0.306 0.7592
## DigitSpanBck:SexMen:PovStatBelow -0.056578 0.225189 -0.251 0.8016
## ---
## 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: 468.32 on 631 degrees of freedom
## AIC: 488.32
##
## Number of Fisher Scoring iterations: 5
exp(cbind(OR = coef(DSBlog3), confint(DSBlog3)))
## OR 2.5 % 97.5 %
## (Intercept) 0.2849011 0.03017875 2.4638141
## DigitSpanBck 1.0455263 0.84108528 1.2839719
## SexMen 0.6342003 0.10483198 3.7433863
## PovStatBelow 1.5052248 0.19592363 11.2177285
## Age 0.9686402 0.94241171 0.9948367
## WRATtotal 1.0092314 0.97257573 1.0491430
## DigitSpanBck:SexMen 1.0392708 0.80147234 1.3534598
## DigitSpanBck:PovStatBelow 1.0090362 0.72802901 1.3967021
## SexMen:PovStatBelow 1.5538360 0.09235157 26.3418136
## DigitSpanBck:SexMen:PovStatBelow 0.9449929 0.60540337 1.4687422
########Compare to null model
#Difference in Deviance
with(DSBlog3,null.deviance - deviance)
## [1] 14.20855
#Degrees of freedom for the difference between two models
with(DSBlog3,df.null - df.residual)
## [1] 9
#p-value
with(DSBlog3,pchisq(null.deviance-deviance,df.null-df.residual,lower.tail=FALSE))
## [1] 0.1150977
#Pseudo R-Squared
nagelkerke(DSBlog3)
## $Models
##
## Model: "glm, PhysAssault ~ (DigitSpanBck + Sex + PovStat)^3 + Age + WRATtotal, binomial, Allvars"
## Null: "glm, PhysAssault ~ 1, binomial, Allvars"
##
## $Pseudo.R.squared.for.model.vs.null
## Pseudo.R.squared
## McFadden 0.0294458
## Cox and Snell (ML) 0.0219224
## Nagelkerke (Cragg and Uhler) 0.0414454
##
## $Likelihood.ratio.test
## Df.diff LogLik.diff Chisq p.value
## -9 -7.1043 14.209 0.1151
##
## $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(DSBlog1,DSBlog3,test = "LR")
## Analysis of Deviance Table
##
## Model 1: PhysAssault ~ DigitSpanBck + WRATtotal
## Model 2: PhysAssault ~ (DigitSpanBck + Sex + PovStat)^3 + Age + WRATtotal
## Resid. Df Resid. Dev Df Deviance Pr(>Chi)
## 1 638 480.70
## 2 631 468.32 7 12.379 0.08877 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1