Apurv Varshney
library(lme4)
library(effectsize)
library(lmerTest)
library(ggplot2)
library(svglite)
library(ggpubr)
library(car)
library(dplyr)
library(lmtest)
library(ordinal)
library(DAAG)
library(robustlmm)
library(moments)
library(gghalves)
library(apaTables)
library(MuMIn)
library(insight)data <- read.csv("data_wayfinding.csv")
data$trialType.f <- factor(data$trialType)
data$subject.f <- factor(data$subject)
data$trialID.f <- factor(data$trialID)
data$stress_level.s <- as.numeric(scale(data$stress_level))
data$success.f <- factor(data$success)
data$Efficiency.s <- as.numeric(scale(data$Efficiency))
data$speed.s <- as.numeric(scale(data$speed))
data$trial_type.int <- dplyr::recode(data$trialType.f, A=1, B=2, C=3)
data$trial_type.p <- dplyr::recode(data$trialType.f, A='control', B='time pressure non-blocked', C='time pressure blocked')
data_subj_success <- data[data$success==1, ] %>%
group_by(subject.f) %>%
summarise(Efficiency = mean(Efficiency))data_si <- read.csv("SI_trial.csv")
data_si$trialType.f <- factor(data_si$trial_type)
data_si$subject.f <- factor(data_si$subject)
data_si$shortcut.f <- factor(data_si$shortcut)
data_with_si <- merge(data[data$success==1, ], data_si, by=c("subject.f","trialID"))
data_shortcut_subj <- data_with_si %>%
group_by(subject.f) %>%
summarise(shortcut = mean(shortcut))data_other <- read.csv("data_questionnaire.csv")
data_other$subject.f <- factor(data_other$ID)
data_per_subj <- data %>% group_by(subject.f) %>%
summarise(stress_level = mean(stress_level),
speed = mean(speed),
success = mean(success))
data_per_subj <- merge(data_per_subj, data_subj_success, by=c("subject.f"))
data_per_subj <- merge(data_per_subj, data_shortcut_subj, by=c("subject.f"))
data_merge_per_subj <- merge(data_per_subj, data_other, by=c("subject.f"))cbPalette <- c("red", "darkgreen", "darkblue")
data_b <- within(data, trialType.f <- relevel(trialType.f, ref = "B"))
data_subj <- data %>%
group_by(subject.f, trialType.f, trial_type.p) %>%
summarise(stress_level = mean(stress_level),
speed = mean(speed),
success = mean(success))model <- lmer(stress_level.s ~ trialType.f + trialNumber + (1 | subject.f), data = data)
summary(model, ddf = "Satterthwaite")## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: stress_level.s ~ trialType.f + trialNumber + (1 | subject.f)
## Data: data
##
## REML criterion at convergence: 2026.4
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.6912 -0.6564 -0.0181 0.5580 4.0117
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 0.3403 0.5833
## Residual 0.3063 0.5534
## Number of obs: 1118, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) -3.478e-01 9.333e-02 6.568e+01 -3.727 0.000406 ***
## trialType.fB 8.110e-01 3.995e-02 1.067e+03 20.304 < 2e-16 ***
## trialType.fC 1.427e+00 4.178e-02 1.067e+03 34.153 < 2e-16 ***
## trialNumber -1.369e-02 2.396e-03 1.067e+03 -5.714 1.43e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trlT.B trlT.C
## trialTyp.fB -0.158
## trialTyp.fC -0.142 0.320
## trialNumber -0.354 0.043 0.016
r.squaredGLMM(model)## R2m R2c
## [1,] 0.3575919 0.6956869
model <- lmer(stress_level.s ~ trialType.f + trialNumber + (1 | subject.f), data = data_b)
summary(model, ddf = "Satterthwaite")## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: stress_level.s ~ trialType.f + trialNumber + (1 | subject.f)
## Data: data_b
##
## REML criterion at convergence: 2026.4
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.6912 -0.6564 -0.0181 0.5580 4.0117
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 0.3403 0.5833
## Residual 0.3063 0.5534
## Number of obs: 1118, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 4.632e-01 9.553e-02 7.203e+01 4.849 6.95e-06 ***
## trialType.fA -8.110e-01 3.995e-02 1.067e+03 -20.304 < 2e-16 ***
## trialType.fC 6.159e-01 4.768e-02 1.067e+03 12.917 < 2e-16 ***
## trialNumber -1.369e-02 2.396e-03 1.067e+03 -5.714 1.43e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trlT.A trlT.C
## trialTyp.fA -0.264
## trialTyp.fC -0.226 0.557
## trialNumber -0.327 -0.043 -0.022
stress <- ggplot(data = data_subj, aes(x=trial_type.p, y=stress_level, shape=trial_type.p)) + geom_half_boxplot(errorbar.draw = TRUE, outlier.color = NA, data=data_subj, aes(x=trial_type.p, y=stress_level, color=trial_type.p)) + geom_half_violin(side = "r", data=data_subj, aes(x=trial_type.p, y=stress_level, color=trial_type.p)) + geom_half_point(side = "l", alpha = .5, data=data_subj, aes(x=trial_type.p, y=stress_level, color=trial_type.p)) + geom_signif(comparisons = list(c("control","time pressure non-blocked"),c("control","time pressure blocked"),c("time pressure non-blocked","time pressure blocked")), annotations = c("***","***","***"), y_position = c(7.6,7,7.6)) + theme_classic() + theme(legend.position="none", axis.text.x = element_text(size = 11, hjust = 0.5)) + scale_y_continuous(name ="Self-reported Stress Level", breaks=seq(0,7,1)) + scale_x_discrete(name="Trial Type") + scale_color_manual(values=cbPalette)
stress
model <- lmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data[data$success==1,])model <- rlmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data[data$success == 1,])
summary(model)## Robust linear mixed model fit by DAStau
## Formula: Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f)
## Data: data[data$success == 1, ]
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.3742 -0.5921 -0.2638 0.5842 18.7660
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 0.04271 0.2067
## Residual 0.24622 0.4962
## Number of obs: 1043, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) -0.230915 0.049108 -4.702
## trialType.fB 0.126165 0.037929 3.326
## trialType.fC -0.095363 0.041047 -2.323
## trialNumber -0.006072 0.002310 -2.628
##
## Correlation of Fixed Effects:
## (Intr) trlT.B trlT.C
## trialTyp.fB -0.264
## trialTyp.fC -0.218 0.302
## trialNumber -0.648 0.024 -0.019
##
## Robustness weights for the residuals:
## 837 weights are ~= 1. The remaining 206 ones are summarized as
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0717 0.3290 0.5090 0.5530 0.7820 0.9990
##
## Robustness weights for the random effects:
## 42 weights are ~= 1. The remaining 6 ones are
## 13 16 18 19 23 38
## 0.450 0.790 0.423 0.734 0.488 0.333
##
## Rho functions used for fitting:
## Residuals:
## eff: smoothed Huber (k = 1.345, s = 10)
## sig: smoothed Huber, Proposal 2 (k = 1.345, s = 10)
## Random Effects, variance component 1 (subject.f):
## eff: smoothed Huber (k = 1.345, s = 10)
## vcp: smoothed Huber, Proposal 2 (k = 1.345, s = 10)
var.fix <- get_variance_fixed(model)
var.ran <- get_variance_random(model)
var.res <- get_variance_residual(model)
R2m = var.fix/(var.fix+var.ran+var.res)
R2c = (var.fix+var.ran)/(var.fix+var.ran+var.res)
print(R2m)## var.fixed
## 0.02623171
print(R2c)## var.fixed
## 0.1701731
model <- rlmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data_b[data_b$success == 1,])
summary(model)## Robust linear mixed model fit by DAStau
## Formula: Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f)
## Data: data_b[data_b$success == 1, ]
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.3742 -0.5921 -0.2638 0.5842 18.7660
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 0.04271 0.2067
## Residual 0.24622 0.4962
## Number of obs: 1043, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) -0.104750 0.053529 -1.957
## trialType.fA -0.126165 0.037929 -3.326
## trialType.fC -0.221528 0.046725 -4.741
## trialNumber -0.006072 0.002310 -2.628
##
## Correlation of Fixed Effects:
## (Intr) trlT.A trlT.C
## trialTyp.fA -0.466
## trialTyp.fC -0.366 0.546
## trialNumber -0.577 -0.024 -0.036
##
## Robustness weights for the residuals:
## 837 weights are ~= 1. The remaining 206 ones are summarized as
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0717 0.3290 0.5090 0.5530 0.7820 0.9990
##
## Robustness weights for the random effects:
## 42 weights are ~= 1. The remaining 6 ones are
## 13 16 18 19 23 38
## 0.450 0.790 0.423 0.734 0.488 0.333
##
## Rho functions used for fitting:
## Residuals:
## eff: smoothed Huber (k = 1.345, s = 10)
## sig: smoothed Huber, Proposal 2 (k = 1.345, s = 10)
## Random Effects, variance component 1 (subject.f):
## eff: smoothed Huber (k = 1.345, s = 10)
## vcp: smoothed Huber, Proposal 2 (k = 1.345, s = 10)
model <- lmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data[data$success == 1,])
robust_model <- rlmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data[data$success == 1,])
# get coefficients from non-robust model to extract Satterthwaite approximated DFs
coefs <- data.frame(coef(summary(model)))
# get coefficients from robust model to extract t-values
coefs.robust <- coef(summary(robust_model))
# calculate p-values based on robust t-values and non-robust approx. DFs
p.values <- 2*pt(abs(coefs.robust[,3]), coefs$df, lower=FALSE)
p.values < 0.05## (Intercept) trialType.fB trialType.fC trialNumber
## TRUE TRUE TRUE TRUE
p.values < 0.01## (Intercept) trialType.fB trialType.fC trialNumber
## TRUE TRUE FALSE TRUE
p.values < 0.001## (Intercept) trialType.fB trialType.fC trialNumber
## TRUE TRUE FALSE FALSE
p.values## (Intercept) trialType.fB trialType.fC trialNumber
## 5.090279e-06 9.123337e-04 2.036589e-02 8.709625e-03
model <- lmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data_b[data_b$success == 1,])
robust_model <- rlmer(Efficiency.s ~ trialType.f + trialNumber + (1 | subject.f), data = data_b[data_b$success == 1,])
# get coefficients from non-robust model to extract Satterthwaite approximated DFs
coefs <- data.frame(coef(summary(model)))
# get coefficients from robust model to extract t-values
coefs.robust <- coef(summary(robust_model))
# calculate p-values based on robust t-values and non-robust approx. DFs
p.values <- 2*pt(abs(coefs.robust[,3]), coefs$df, lower=FALSE)
p.values < 0.05## (Intercept) trialType.fA trialType.fC trialNumber
## FALSE TRUE TRUE TRUE
p.values < 0.01## (Intercept) trialType.fA trialType.fC trialNumber
## FALSE TRUE TRUE TRUE
p.values < 0.001## (Intercept) trialType.fA trialType.fC trialNumber
## FALSE TRUE TRUE FALSE
p.values## (Intercept) trialType.fA trialType.fC trialNumber
## 5.149490e-02 9.123337e-04 2.436551e-06 8.709625e-03
data_subj_trial_success <- data[data$success==1,] %>%
group_by(subject.f, trialType.f, trial_type.p) %>%
summarise(Efficiency = mean(Efficiency))## `summarise()` has grouped output by 'subject.f', 'trialType.f'. You can
## override using the `.groups` argument.
eff <- ggplot(data = data_subj_trial_success, aes(x=trial_type.p, y=Efficiency, shape=trial_type.p)) + geom_half_boxplot(errorbar.draw = TRUE, outlier.color = NA, data=data_subj_trial_success, aes(x=trial_type.p, y=Efficiency, color=trial_type.p)) + geom_half_violin(side = "r", data=data_subj_trial_success, aes(x=trial_type.p, y=Efficiency, color=trial_type.p)) + geom_half_point(side = "l", alpha = .5, data=data_subj_trial_success, aes(x=trial_type.p, y=Efficiency, color=trial_type.p)) + geom_signif(comparisons = list(c("control","time pressure non-blocked"),c("control","time pressure blocked"),c("time pressure non-blocked","time pressure blocked")), annotations = c("***","*","***"), y_position = c(1.6,1.4,1.6)) + theme_classic() + theme(legend.position="none", axis.text.x = element_text(size = 11, hjust = 0.5)) + scale_y_continuous(name ="Excess Distance") + scale_x_discrete(name="Trial Type") + scale_color_manual(values=cbPalette)
eff
model <- glmer(success.f ~ trialType.f + trialNumber +timeInTrial+ (1 | subject.f) , data = data, family = binomial(link = "logit"))
summary(model)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: success.f ~ trialType.f + trialNumber + timeInTrial + (1 | subject.f)
## Data: data
##
## AIC BIC logLik deviance df.resid
## 365.9 396.0 -177.0 353.9 1112
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -48.030 0.040 0.083 0.195 1.731
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 0.2852 0.534
## Number of obs: 1118, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 9.40851 0.94943 9.910 < 2e-16 ***
## trialType.fB -1.83091 0.47035 -3.893 9.92e-05 ***
## trialType.fC -2.12239 0.42076 -5.044 4.56e-07 ***
## trialNumber 0.02507 0.02206 1.136 0.256
## timeInTrial -0.35885 0.03862 -9.291 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trlT.B trlT.C trlNmb
## trialTyp.fB -0.616
## trialTyp.fC -0.611 0.724
## trialNumber -0.375 0.093 0.033
## timeInTrial -0.883 0.424 0.393 0.100
r.squaredGLMM(model)## R2m R2c
## theoretical 0.5417981 0.5783512
## delta 0.2046708 0.2184791
model <- glmer(success.f ~ trialType.f + trialNumber + (1 | subject.f) , data = data_b, family = binomial(link = "logit"))
summary(model)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: success.f ~ trialType.f + trialNumber + (1 | subject.f)
## Data: data_b
##
## AIC BIC logLik deviance df.resid
## 475.7 500.8 -232.8 465.7 1113
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -11.4889 0.0997 0.1637 0.2388 1.1190
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 1.618 1.272
## Number of obs: 1118, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 2.61523 0.42842 6.104 1.03e-09 ***
## trialType.fA 0.94631 0.36020 2.627 0.00861 **
## trialType.fC -0.97627 0.31557 -3.094 0.00198 **
## trialNumber 0.04971 0.02072 2.399 0.01644 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trlT.A trlT.C
## trialTyp.fA -0.324
## trialTyp.fC -0.447 0.524
## trialNumber -0.537 -0.108 -0.069
success <- ggplot(data = data_subj, aes(x=trial_type.p, y=success, shape=trial_type.p)) + geom_half_boxplot(errorbar.draw = TRUE, outlier.color = NA, data=data_subj, aes(x=trial_type.p, y=success, color=trial_type.p)) + geom_half_violin(side = "r", data=data_subj, aes(x=trial_type.p, y=success, color=trial_type.p)) + geom_half_point(side = "l", alpha = .5, data=data_subj, aes(x=trial_type.p, y=success, color=trial_type.p)) + geom_signif(comparisons = list(c("control","time pressure non-blocked"),c("control","time pressure blocked"),c("time pressure non-blocked","time pressure blocked")), annotations = c("**","***","**"), y_position = c(1.2,1.1,1.2)) + theme_classic() + theme(legend.position="none", axis.text.x = element_text(size = 11, hjust = 0.5)) + scale_y_continuous(name ="Trial Completion Rate", breaks=seq(0,1,0.5)) + scale_x_discrete(name="Trial Type") + scale_color_manual(values=cbPalette)
success
data_shortcut_trialtype <- data_with_si %>%
group_by(subject.f, trial_type.p, trialType) %>%
summarise(shortcut=mean(shortcut),
Efficiency=mean(Efficiency))
model <- glmer(shortcut.f ~ trialType.f.x + trialNumber + (1 | subject.f) , data = data_with_si, family = binomial(link = "logit"))
summary(model)## Generalized linear mixed model fit by maximum likelihood (Laplace
## Approximation) [glmerMod]
## Family: binomial ( logit )
## Formula: shortcut.f ~ trialType.f.x + trialNumber + (1 | subject.f)
## Data: data_with_si
##
## AIC BIC logLik deviance df.resid
## 1089.4 1108.2 -540.7 1081.4 823
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.8512 -0.8449 0.5032 0.8388 2.4759
##
## Random effects:
## Groups Name Variance Std.Dev.
## subject.f (Intercept) 0.6568 0.8104
## Number of obs: 827, groups: subject.f, 48
##
## Fixed effects:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.31906 0.21582 -1.478 0.13931
## trialType.f.xB -0.38383 0.16136 -2.379 0.01738 *
## trialNumber 0.03545 0.01144 3.098 0.00195 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) trT..B
## trlTyp.f.xB -0.244
## trialNumber -0.726 0.007
r.squaredGLMM(model)## R2m R2c
## theoretical 0.02338274 0.1859109
## delta 0.01988255 0.1580817
shortcut <- ggplot(data = data_shortcut_trialtype, aes(x=trial_type.p, y=shortcut, shape=trial_type.p)) + geom_half_boxplot(errorbar.draw = TRUE, outlier.color = NA, data=data_shortcut_trialtype, aes(x=trial_type.p, y=shortcut, color=trial_type.p)) + geom_half_violin(side = "r", data=data_shortcut_trialtype, aes(x=trial_type.p, y=shortcut, color=trial_type.p)) + geom_half_point(side = "l", alpha = .5, data=data_shortcut_trialtype, aes(x=trial_type.p, y=shortcut, color=trial_type.p)) + geom_signif(comparisons = list(c("control","time pressure non-blocked")), annotations = c("*"), y_position = c(1.1)) + theme_classic() + theme(legend.position="none", axis.text.x = element_text(size = 11, hjust = 0.5)) + scale_y_continuous(name ="Shortcut Rate", breaks=seq(0,1,0.5)) + scale_x_discrete(name="Trial Type") + scale_color_manual(values=cbPalette)
shortcut
multi <- ggarrange(stress, success, eff, shortcut,
ncol = 2, nrow = 2, labels = "AUTO")
multi
data2 <- data_other %>%
dplyr::mutate(category =
dplyr::case_when(
PointingError >= median(PointingError) ~ "High",
PointingError < median(PointingError) ~ "Low"
)
)data_high_low <- merge(data_shortcut_trialtype, data2, by=c("subject.f"))data_high_low$category <- factor(data_high_low$category, levels=c("Low", "High"))
s <- ggplot(data = data_high_low, aes(x=category, y=shortcut)) + geom_signif(annotation=c("***", "NS"), y_position = c(1.1, 1.1), xmin=c(0.8,1.8), xmax=c(1.2,2.2)) + geom_half_boxplot(errorbar.draw = TRUE, outlier.color = NA, data=data_high_low, aes(x=category, y=shortcut, color=trial_type.p)) + geom_half_violin(side = "r", data=data_high_low, aes(x=category, y=shortcut, color=trial_type.p)) + geom_half_point(side='l', alpha = .5, data=data_high_low, aes(x=category, y=shortcut, color=trial_type.p)) + theme_classic2() + scale_y_continuous(name ="Shortcut Rate") + scale_x_discrete(name="Pointing Error") + scale_color_manual(values=cbPalette) + labs(color = "Trial Type")
s
res <- t.test(data_high_low$shortcut[data_high_low$category=="Low" & data_high_low$trialType=="A"], data_high_low$shortcut[data_high_low$category=="Low" & data_high_low$trialType=="B"], paired=TRUE)
cohens_d(res)## Cohen's d | 95% CI
## ------------------------
## 0.78 | [0.31, 1.23]
res##
## Paired t-test
##
## data: data_high_low$shortcut[data_high_low$category == "Low" & data_high_low$trialType == "A"] and data_high_low$shortcut[data_high_low$category == "Low" & data_high_low$trialType == "B"]
## t = 3.8049, df = 23, p-value = 0.0009121
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
## 0.07943111 0.26871102
## sample estimates:
## mean difference
## 0.1740711
res <- t.test(data_high_low$shortcut[data_high_low$category=="High" & data_high_low$trialType=="A"], data_high_low$shortcut[data_high_low$category=="High" & data_high_low$trialType=="B"], paired=TRUE)
cohens_d(res)## Cohen's d | 95% CI
## -------------------------
## -0.05 | [-0.45, 0.35]
res##
## Paired t-test
##
## data: data_high_low$shortcut[data_high_low$category == "High" & data_high_low$trialType == "A"] and data_high_low$shortcut[data_high_low$category == "High" & data_high_low$trialType == "B"]
## t = -0.23327, df = 23, p-value = 0.8176
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
## -0.11961279 0.09537036
## sample estimates:
## mean difference
## -0.01212121
# data_high_low$category <- factor(data_high_low$category, levels=c("Low", "High"))
ed <- ggplot(data = data_high_low, aes(x=category, y=Efficiency.x)) + geom_signif(annotation=c("*", "NS"), y_position = c(1.35, 1.35), xmin=c(0.8,1.8), xmax=c(1.2,2.2)) + geom_half_boxplot(errorbar.draw = TRUE, outlier.color = NA, data=data_high_low, aes(x=category, y=Efficiency.x, color=trial_type.p)) + geom_half_violin(side = "r", data=data_high_low, aes(x=category, y=Efficiency.x, color=trial_type.p)) + geom_half_point(side='l', alpha = .5, data=data_high_low, aes(x=category, y=Efficiency.x, color=trial_type.p)) + theme_classic2() + scale_y_continuous(name ="Excess Distance") + scale_x_discrete(name="Pointing Error") + scale_color_manual(values=cbPalette) + labs(color = "Trial Type")
ed
res <- t.test(data_high_low$Efficiency.x[data_high_low$category=="Low" & data_high_low$trialType=="A"], data_high_low$Efficiency.x[data_high_low$category=="Low" & data_high_low$trialType=="B"], paired = TRUE)
cohens_d(res)## Cohen's d | 95% CI
## --------------------------
## -0.48 | [-0.90, -0.05]
res##
## Paired t-test
##
## data: data_high_low$Efficiency.x[data_high_low$category == "Low" & data_high_low$trialType == "A"] and data_high_low$Efficiency.x[data_high_low$category == "Low" & data_high_low$trialType == "B"]
## t = -2.3364, df = 23, p-value = 0.02855
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
## -0.17291598 -0.01050915
## sample estimates:
## mean difference
## -0.09171256
res <- t.test(data_high_low$Efficiency.x[data_high_low$category=="High" & data_high_low$trialType=="A"], data_high_low$Efficiency.x[data_high_low$category=="High" & data_high_low$trialType=="B"], paired=TRUE)
cohens_d(res)## Cohen's d | 95% CI
## -------------------------
## -2.18e-03 | [-0.40, 0.40]
res##
## Paired t-test
##
## data: data_high_low$Efficiency.x[data_high_low$category == "High" & data_high_low$trialType == "A"] and data_high_low$Efficiency.x[data_high_low$category == "High" & data_high_low$trialType == "B"]
## t = -0.010672, df = 23, p-value = 0.9916
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
## -0.1116633 0.1105171
## sample estimates:
## mean difference
## -0.0005731011
data_cor <- data_merge_per_subj[c("success", "Efficiency.x", "shortcut", "stress_level.x", "PointingError", "SpatialAnxiety", "SBSOD")]
data_cor <- data_cor %>% rename("StressLevel" = "stress_level.x",
"TrialCompletion Rate" = "success",
"ExcessDistance" = "Efficiency.x",
"ShortcutRate" = "shortcut",
"PointingError" = "PointingError",
"SpatialAnxiety" = "SpatialAnxiety",
"SBSOD" = "SBSOD")
apa.cor.table(data_cor)##
##
## Means, standard deviations, and correlations with confidence intervals
##
##
## Variable M SD 1 2 3
## 1. TrialCompletion Rate 0.93 0.09
##
## 2. ExcessDistance 0.32 0.23 -.37**
## [-.59, -.10]
##
## 3. ShortcutRate 0.51 0.21 .21 -.85**
## [-.08, .47] [-.91, -.74]
##
## 4. StressLevel 2.74 1.00 -.36* .40** -.29*
## [-.58, -.08] [.14, .62] [-.53, -.01]
##
## 5. PointingError 62.08 21.15 -.44** .45** -.41**
## [-.65, -.18] [.20, .65] [-.62, -.14]
##
## 6. SpatialAnxiety 2.52 0.55 -.05 .42** -.33*
## [-.33, .24] [.15, .63] [-.56, -.05]
##
## 7. SBSOD 4.25 1.02 .13 -.25 .24
## [-.16, .40] [-.50, .03] [-.05, .49]
##
## 4 5 6
##
##
##
##
##
##
##
##
##
##
##
## .33*
## [.05, .56]
##
## .48** .14
## [.23, .67] [-.15, .40]
##
## -.20 -.08 -.52**
## [-.46, .09] [-.35, .21] [-.70, -.28]
##
##
## Note. M and SD are used to represent mean and standard deviation, respectively.
## Values in square brackets indicate the 95% confidence interval.
## The confidence interval is a plausible range of population correlations
## that could have caused the sample correlation (Cumming, 2014).
## * indicates p < .05. ** indicates p < .01.
##
fit <- lm(ShortcutRate ~ PointingError + SBSOD, data = data_cor)
summary(fit)##
## Call:
## lm(formula = ShortcutRate ~ PointingError + SBSOD, data = data_cor)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.33661 -0.14664 0.02071 0.11460 0.46743
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.575452 0.151279 3.804 0.000427 ***
## PointingError -0.003971 0.001336 -2.973 0.004723 **
## SBSOD 0.042547 0.027578 1.543 0.129879
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.1931 on 45 degrees of freedom
## Multiple R-squared: 0.2106, Adjusted R-squared: 0.1755
## F-statistic: 6.001 on 2 and 45 DF, p-value: 0.004896
fit <- lm(ExcessDistance ~ PointingError + SBSOD, data = data_cor)
summary(fit)##
## Call:
## lm(formula = ExcessDistance ~ PointingError + SBSOD, data = data_cor)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.40291 -0.10599 -0.02316 0.08655 0.82632
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.230949 0.161663 1.429 0.16003
## PointingError 0.004841 0.001427 3.392 0.00146 **
## SBSOD -0.049690 0.029471 -1.686 0.09870 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.2063 on 45 degrees of freedom
## Multiple R-squared: 0.254, Adjusted R-squared: 0.2209
## F-statistic: 7.661 on 2 and 45 DF, p-value: 0.001369
data_merge_per_subj$diffSI <- data_shortcut_trialtype[data_shortcut_trialtype$trialType=="A",]$shortcut - data_shortcut_trialtype[data_shortcut_trialtype$trialType=="B",]$shortcut
data_merge_per_subj$diffED <- data_shortcut_trialtype[data_shortcut_trialtype$trialType=="A",]$Efficiency - data_shortcut_trialtype[data_shortcut_trialtype$trialType=="B",]$Efficiency
data_diff <- data_merge_per_subj[c("diffSI", "diffED", "PointingError", "SpatialAnxiety")]
apa.cor.table(data_diff)##
##
## Means, standard deviations, and correlations with confidence intervals
##
##
## Variable M SD 1 2 3
## 1. diffSI 0.08 0.26
##
## 2. diffED -0.05 0.23 -.71**
## [-.83, -.53]
##
## 3. PointingError 62.08 21.15 -.43** .35*
## [-.64, -.17] [.08, .58]
##
## 4. SpatialAnxiety 2.52 0.55 -.12 .03 .14
## [-.39, .17] [-.26, .31] [-.15, .40]
##
##
## Note. M and SD are used to represent mean and standard deviation, respectively.
## Values in square brackets indicate the 95% confidence interval.
## The confidence interval is a plausible range of population correlations
## that could have caused the sample correlation (Cumming, 2014).
## * indicates p < .05. ** indicates p < .01.
##
cor.test(data_diff$diffSI,data_diff$PointingError)##
## Pearson's product-moment correlation
##
## data: data_diff$diffSI and data_diff$PointingError
## t = -3.2764, df = 46, p-value = 0.002004
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.6400139 -0.1721184
## sample estimates:
## cor
## -0.4349805
cor.test(data_diff$diffED,data_diff$PointingError)##
## Pearson's product-moment correlation
##
## data: data_diff$diffED and data_diff$PointingError
## t = 2.5688, df = 46, p-value = 0.01352
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.07789572 0.57995985
## sample estimates:
## cor
## 0.3541912