Comparing Parameter Estimates

Updated Exclusion Criteria:

• 5-responses required
• No requirement for number of exams

Updated Sample Sizes:

• Exploratory: N = 75 (previous: N = 44)
• Confirmatory: N = 81 (previous: N = 51)

New Gammas Fit to Full Exploratory Data Frame:

• PA: 0.85
• NA: 0.91

Parameter Estimates: Exploratory & Confirmatory

Affect ~ outcome + PE + ( 1 + outcome + PE | cohort / id / exam_num)

With REML, PE is trending in exploratory dataset, and significant in confirmatory

Dataset	Affect	predictor	Estimate	SE	df	t_stat	p_Val
Exploratory	Positive	outcome	0.3102355	0.2160938	2.3624013	1.435652	0.2692764
Exploratory	Positive	PE	3.5752270	0.6013576	1.3909792	5.945259	0.0590905
Exploratory	Negative	outcome	-0.6449324	0.3796406	1.0432242	-1.698798	0.3309908
Exploratory	Negative	PE	-2.6717013	0.7686514	1.6284042	-3.475830	0.0980165
Confirmatory	Positive	outcome	0.4384121	0.2576754	0.9817883	1.701412	0.3416694
Confirmatory	Positive	PE	3.4700851	0.5053408	72.9347299	6.866822	0.0000000
Confirmatory	Negative	outcome	-0.5384790	0.2885021	1.1152864	-1.866465	0.2928893
Confirmatory	Negative	PE	-2.1758658	0.6399473	1.4356868	-3.400070	0.1190673

Testing for differences between PE and outcome parameter estimates

• Esticon
• Paired Sample T-test of estimates from single subject fixed effects models (w/ DF-fit gammas)

Esticon

Exploratory PA model

est_CI <- esticon(exp.PA.lmer, L = c(0,-1,1), conf.in = NULL) # PE - Outcome
est_CI

##               beta0       Estimate      Std.Error       X2.value
## [1,]  0.00000000000  3.26499156946  0.66216238160 24.31280797526
##                  DF     Pr(>|X^2|)          Lower  Upper
## [1,]  1.00000000000  0.00000081893  1.96717714961 4.5628

p-value: 0.0000008

Exploratory NA model

est_CI <- esticon(exp.NA.lmer, L = c(0,-1,1), conf.in = NULL) # PE - Outcome
est_CI

##          beta0  Estimate Std.Error  X2.value        DF Pr(>|X^2|)
## [1,]  0.000000 -2.026769  0.990306  4.188605  1.000000   0.040697
##          Lower   Upper
## [1,] -3.967733 -0.0858

p-value: 0.0406966

Confirmatory PA model

est_CI <- esticon(con.PA.lmer, L = c(0,-1,1), conf.in = NULL) # PE - Outcome
est_CI

##               beta0       Estimate      Std.Error       X2.value
## [1,]  0.00000000000  3.03167301243  0.56214669110 29.08475421775
##                  DF     Pr(>|X^2|)          Lower  Upper
## [1,]  1.00000000000  0.00000006928  1.92988574385 4.1335

p-value: 0.0000001

Confirmatory NA model

est_CI <- esticon(con.NA.lmer, L = c(0,-1,1), conf.in = NULL) # PE - Outcome
est_CI

##           beta0   Estimate  Std.Error   X2.value         DF Pr(>|X^2|)
## [1,]  0.0000000 -1.6373868  0.6219449  6.9310445  1.0000000  0.0084712
##           Lower   Upper
## [1,] -2.8563765 -0.4184

p-value: 0.0084712

T-Tests

Exploratory PA model

est_exp.PA.t <- t.test(est_exp$PA_outcome_B, est_exp$PA_PE_B, paired = TRUE, alternative = "two.sided")
est_exp.PA.t

## 
##  Paired t-test
## 
## data:  est_exp$PA_outcome_B and est_exp$PA_PE_B
## t = -2.1847, df = 74, p-value = 0.03208
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -5.5613786 -0.2558294
## sample estimates:
## mean of the differences 
##               -2.908604

p-value: 0.0320779

Exploratory NA model

est_exp.NA.t <- t.test(est_exp$NA_outcome_B, est_exp$NA_PE_B, paired = TRUE, alternative = "two.sided")
est_exp.NA.t

## 
##  Paired t-test
## 
## data:  est_exp$NA_outcome_B and est_exp$NA_PE_B
## t = 0.58541, df = 74, p-value = 0.5601
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -1.844530  3.379289
## sample estimates:
## mean of the differences 
##               0.7673793

p-value: 0.5600534

Confimratory PA model

est_con.PA.t <- t.test(est_con$PA_outcome_B, est_con$PA_PE_B, paired = TRUE, alternative = "two.sided")
est_con.PA.t

## 
##  Paired t-test
## 
## data:  est_con$PA_outcome_B and est_con$PA_PE_B
## t = 0.66193, df = 80, p-value = 0.5099
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -7.278904 14.534404
## sample estimates:
## mean of the differences 
##                 3.62775

p-value: 0.5099177

Confimratory NA model

est_con.NA.t <- t.test(est_con$NA_outcome_B, est_con$NA_PE_B, paired = TRUE, alternative = "two.sided")
est_con.NA.t

## 
##  Paired t-test
## 
## data:  est_con$NA_outcome_B and est_con$NA_PE_B
## t = -1.2188, df = 80, p-value = 0.2265
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -23.498162   5.647518
## sample estimates:
## mean of the differences 
##               -8.925322

p-value: 0.2264869