Start by creating regression tables for DSB, DSF, & Trails B without list-wise deletion - that is, with unequal analysis sample n’s
Descriptives for all variables
zQuick(TeloCogR)## Dimensions: 339 18
##
## tLength LgTrailA LgTrailB DSF
## Min. :2.600 Min. :1.146 Min. :1.491 Min. : 0.000
## 1st Qu.:5.200 1st Qu.:1.380 1st Qu.:1.785 1st Qu.: 6.000
## Median :5.670 Median :1.491 Median :1.929 Median : 7.000
## Mean :5.645 Mean :1.514 Mean :2.017 Mean : 7.074
## 3rd Qu.:6.080 3rd Qu.:1.602 3rd Qu.:2.152 3rd Qu.: 9.000
## Max. :8.500 Max. :2.778 Max. :2.778 Max. :14.000
## NA's :14 NA's :14 NA's :3
## DSB TeloPov TeloRace RacePov
## Min. : 0.000 Min. : 2.600 Min. : 3.770 Min. :1.000
## 1st Qu.: 4.000 1st Qu.: 5.690 1st Qu.: 5.600 1st Qu.:1.000
## Median : 5.000 Median : 6.840 Median : 7.600 Median :2.000
## Mean : 5.507 Mean : 8.349 Mean : 8.438 Mean :2.215
## 3rd Qu.: 7.000 3rd Qu.:11.190 3rd Qu.:11.360 3rd Qu.:2.000
## Max. :13.000 Max. :16.600 Max. :17.000 Max. :4.000
## NA's :2
## TeloRacePov Educ CRPdi3 HTNDic DiabDic
## Min. : 3.770 HS+ :253 <3 :190 Min. :0.0000 No :286
## 1st Qu.: 6.825 <HS : 85 3+ :134 1st Qu.:0.0000 Yes : 52
## Median :11.200 NA's: 1 NA's: 15 Median :0.0000 NA's: 1
## Mean :12.465 Mean :0.4036
## 3rd Qu.:14.780 3rd Qu.:1.0000
## Max. :30.800 Max. :1.0000
## NA's :2
## BMI Age Sex Race PovStat
## < 30 :202 Min. :30.16 Women:167 White:171 Above:175
## >= 30:137 1st Qu.:41.07 Men :172 AfrAm:168 Below:164
## Median :48.72
## Mean :48.04
## 3rd Qu.:55.32
## Max. :64.99
## Create tables for regression models with the 3-way interaction of telo x povstat x race on DSB, DSF, & Trails B
bulkFormula= 'DSF + DSB + LgTrailB ~ Race*PovStat*tLength + tLength*Race + tLength*PovStat + Race*PovStat + tLength + Race + PovStat + Educ + CRPdi3 + HTNDic + DiabDic + BMI + Age + Sex'
zBulkReg(TeloCogR, bulkFormula)
| Effects | DSF | DSB | LgTrailB | R**2 | 0.109** | 0.160*** | 0.188*** | RaceAfrAm | -6.790* | -4.374 | 0.453 | PovStatBelow | -9.983** | -8.782** | 0.766 | tLength | -0.802 | -0.540 | 0.022 |
Educ|
-0.725*
|
-1.385***
|
0.144***
|
CRPdi33+
|
-0.327
|
0.110
|
-0.076*
|
HTNDic
|
0.780*
|
0.415
|
0.061
|
DiabDicYes
|
-0.595
|
-1.011*
|
0.020
|
BMI>= 30
|
-0.268
|
-0.233
|
0.038
|
Age
|
-0.040*
|
-0.027
|
0.004
|
SexMen
|
-0.292
|
0.019
|
0.020
|
RaceAfrAm:PovStatBelow
|
10.778*
|
8.139
|
-0.153
|
RaceAfrAm:tLength
|
1.114
|
0.613
|
-0.052
|
PovStatBelow:tLength
|
1.691**
|
1.481**
|
-0.117
|
RaceAfrAm:PovStatBelow:tLength
|
-1.984*
|
-1.460*
|
0.029
|
|
|---|
The 3-way interaction was significant for DSF & DSB, not for Trails B
Remove the three-way interaction and re-run analysis with Trails B
| Effects | LgTrailB | R**2 | 0.187*** | tLength | 0.014 | RaceAfrAm | 0.362 | PovStatBelow | 0.684* |
Educ|
0.145***
|
CRPdi33+
|
-0.077*
|
HTNDic
|
0.062
|
DiabDicYes
|
0.020
|
BMI>= 30
|
0.038
|
Age
|
0.004
|
SexMen
|
0.021
|
tLength:RaceAfrAm
|
-0.036
|
tLength:PovStatBelow
|
-0.102*
|
RaceAfrAm:PovStatBelow
|
0.010
|
|
|---|
2-way interaction of Telo x PovStat was significant
Remove the non-significant 2-ways and re-run the model
bulkFormula3= 'LgTrailB ~ tLength*PovStat + tLength + Race + PovStat + Educ + CRPdi3 + HTNDic + DiabDic + BMI + Age + Sex'
zBulkReg(TeloCogR, bulkFormula3)
| Effects | LgTrailB | R**2 | 0.186*** | tLength | -0.006 | PovStatBelow | 0.670* | RaceAfrAm | 0.161*** |
Educ|
0.145***
|
CRPdi33+
|
-0.078*
|
HTNDic
|
0.065
|
DiabDicYes
|
0.020
|
BMI>= 30
|
0.036
|
Age
|
0.004
|
SexMen
|
0.017
|
tLength:PovStatBelow
|
-0.099*
|
|
|---|
2-way interaction of Telo x Pov was significant for Trails B
Now run the analyses with equal n’s for DSF, DSB, & Trails B
TeloCogSPSS_exclude <- TeloCogSPSS[ which(TeloCogSPSS$MedHxAlzheimersDisease==0&TeloCogSPSS$MedHxDementiaAny==0&
TeloCogSPSS$MedHxCVstroke==0&TeloCogSPSS$MedHxEpilepsy==0&
TeloCogSPSS$MedHxMultipleSclerosis==0&TeloCogSPSS$MedHxParkinsonDisease==0&TeloCogSPSS$valicog==3),]
varNames = c('tLength', 'LgTrailA', 'LgTrailB', 'DSF', 'DSB', 'TeloPov', 'TeloRace', 'RacePov', 'TeloRacePov', 'Educ', 'CRPdi3', 'HTNDic', 'DiabDic', 'BMI', 'Age', 'Sex', 'Race', 'PovStat')
zNamesCheck(TeloCogSPSS_exclude, varNames)
TeloCogR=TeloCogSPSS_exclude[,varNames]
Descriptives for all variables
zQuick(TeloCogR)## Dimensions: 339 18
##
## tLength LgTrailA LgTrailB DSF
## Min. :2.600 Min. :1.146 Min. :1.491 Min. : 0.000
## 1st Qu.:5.200 1st Qu.:1.380 1st Qu.:1.785 1st Qu.: 6.000
## Median :5.670 Median :1.491 Median :1.929 Median : 7.000
## Mean :5.645 Mean :1.514 Mean :2.017 Mean : 7.074
## 3rd Qu.:6.080 3rd Qu.:1.602 3rd Qu.:2.152 3rd Qu.: 9.000
## Max. :8.500 Max. :2.778 Max. :2.778 Max. :14.000
## NA's :14 NA's :14 NA's :3
## DSB TeloPov TeloRace RacePov
## Min. : 0.000 Min. : 2.600 Min. : 3.770 Min. :1.000
## 1st Qu.: 4.000 1st Qu.: 5.690 1st Qu.: 5.600 1st Qu.:1.000
## Median : 5.000 Median : 6.840 Median : 7.600 Median :2.000
## Mean : 5.507 Mean : 8.349 Mean : 8.438 Mean :2.215
## 3rd Qu.: 7.000 3rd Qu.:11.190 3rd Qu.:11.360 3rd Qu.:2.000
## Max. :13.000 Max. :16.600 Max. :17.000 Max. :4.000
## NA's :2
## TeloRacePov Educ CRPdi3 HTNDic DiabDic
## Min. : 3.770 HS+ :253 <3 :190 Min. :0.0000 No :286
## 1st Qu.: 6.825 <HS : 85 3+ :134 1st Qu.:0.0000 Yes : 52
## Median :11.200 NA's: 1 NA's: 15 Median :0.0000 NA's: 1
## Mean :12.465 Mean :0.4036
## 3rd Qu.:14.780 3rd Qu.:1.0000
## Max. :30.800 Max. :1.0000
## NA's :2
## BMI Age Sex Race PovStat
## < 30 :202 Min. :30.16 Women:167 White:171 Above:175
## >= 30:137 1st Qu.:41.07 Men :172 AfrAm:168 Below:164
## Median :48.72
## Mean :48.04
## 3rd Qu.:55.32
## Max. :64.99
## Create tables for regression models with the 3-way interaction of telo x povstat x race on DSB, DSF, & Trails B
| Effects | DSF | DSB | LgTrailB | R**2 | 0.109** | 0.160*** | 0.188*** | RaceAfrAm | -6.790* | -4.374 | 0.453 | PovStatBelow | -9.983** | -8.782** | 0.766 | tLength | -0.802 | -0.540 | 0.022 |
Educ|
-0.725*
|
-1.385***
|
0.144***
|
CRPdi33+
|
-0.327
|
0.110
|
-0.076*
|
HTNDic
|
0.780*
|
0.415
|
0.061
|
DiabDicYes
|
-0.595
|
-1.011*
|
0.020
|
BMI>= 30
|
-0.268
|
-0.233
|
0.038
|
Age
|
-0.040*
|
-0.027
|
0.004
|
SexMen
|
-0.292
|
0.019
|
0.020
|
RaceAfrAm:PovStatBelow
|
10.778*
|
8.139
|
-0.153
|
RaceAfrAm:tLength
|
1.114
|
0.613
|
-0.052
|
PovStatBelow:tLength
|
1.691**
|
1.481**
|
-0.117
|
RaceAfrAm:PovStatBelow:tLength
|
-1.984*
|
-1.460*
|
0.029
|
|
|---|
The 3-way interaction was significant for DSF, but not DSB or Trails B
Remove the three-way interaction and re-run analysis with DSB & Trails B
| Effects | DSB | LgTrailB | R**2 | 0.149*** | 0.187*** | tLength | -0.071 | 0.014 | RaceAfrAm | -0.041 | 0.362 | PovStatBelow | -4.345* | 0.684* |
Educ|
-1.416***
|
0.145***
|
CRPdi33+
|
0.128
|
-0.077*
|
HTNDic
|
0.391
|
0.062
|
DiabDicYes
|
-1.040**
|
0.020
|
BMI>= 30
|
-0.225
|
0.038
|
Age
|
-0.028
|
0.004
|
SexMen
|
-0.003
|
0.021
|
tLength:RaceAfrAm
|
-0.149
|
-0.036
|
tLength:PovStatBelow
|
0.700
|
-0.102*
|
RaceAfrAm:PovStatBelow
|
-0.103
|
0.010
|
|
|---|
The 2-way interaction of Telo x PovStat was significant for both DSB & Trails B
Remove the nonsignificant 2-way interactions and re-run analysis with DSB & Trails B
| Effects | DSB | LgTrailB | R**2 | 0.149*** | 0.187*** | tLength | -0.071 | 0.014 | RaceAfrAm | -0.041 | 0.362 | PovStatBelow | -4.345* | 0.684* |
Educ|
-1.416***
|
0.145***
|
CRPdi33+
|
0.128
|
-0.077*
|
HTNDic
|
0.391
|
0.062
|
DiabDicYes
|
-1.040**
|
0.020
|
BMI>= 30
|
-0.225
|
0.038
|
Age
|
-0.028
|
0.004
|
SexMen
|
-0.003
|
0.021
|
tLength:RaceAfrAm
|
-0.149
|
-0.036
|
tLength:PovStatBelow
|
0.700
|
-0.102*
|
RaceAfrAm:PovStatBelow
|
-0.103
|
0.010
|
|
|---|