library(rstatix)
library(dplyr) 
library(ggplot2) 

# A small function helping us to round  all numbers (0.123456 -> 0.123)
round_df <- function(df, digits = 3) {
    nums <- vapply(df, is.numeric, FUN.VALUE = logical(1))
    df[,nums] <- round(df[,nums], digits = digits)
    (df)
}

# Loading the CSV file
df <- read.csv("data-wpm-onefactorial.csv") 

# Show header if everything is OK
head(df)
 
# Check if data is normal distributed
round_df(df %>% group_by(Conditions) %>% shapiro_test(WPM))

# Nothing is significant, so we can run a three-way repeated-measures ANOVA
anova <- anova_test(data = df, dv = WPM, wid = SubjectID, within = Conditions,  effect.size = "pes")

# Auto correction of degree of freedom (if necessary)
get_anova_table(anova, correction = "auto")

# Pairwise Posthoc Comparison to understand where the differences occur
pairwise.t.test(df$WPM, df$Conditions, p.adj = "bonf", paired = TRUE)

# Let's aggregate the data
means <- df %>% group_by(Conditions) %>% summarize(
    Mean = mean(WPM),
    SD = sd(WPM),
    N = n(), .groups = 'drop',) %>% mutate(
    SE = SD / sqrt(N),
    CI = qt(1 - (0.05 / 2), N - 1) * SE) 

# Let's plot the data
plot <- ggplot(means, aes(y=Mean, x=Conditions, fill=Conditions)) +  
                    geom_bar(stat = "identity") + 
                    geom_errorbar(aes(ymin=Mean-CI, ymax=Mean+CI, width=.1)) + 
                    ylab("words per minute (WPM)") + 
                    xlab("")+
                    ggtitle("Typing performance") +
                    theme_minimal() +  
                    theme( 
                        legend.margin=margin(t = -5, unit='mm'),
                        plot.margin=unit(x=c(1,2,1,1),units="mm"),
                        legend.position = "none",
                        legend.title = element_blank(),
                        plot.title = element_text(hjust = 0.5), 
                        panel.background = element_rect(size=0.5, color="transparent", fill="gray98" )
                    ) 

options(repr.plot.width = 5, repr.plot.height = 5)

plot
ggsave("Swipe-WPM-BarChart.pdf", plot, width=5, height=5, device=cairo_pdf)

A data.frame: 6 × 4

	SubjectID	Sequence	Conditions	WPM
	<int>	<int>	<chr>	<dbl>
1	1	1	NoSwipe	44.559
2	1	2	Swipe	75.381
3	1	3	VibroSwipe	88.695
4	2	1	NoSwipe	42.951
5	2	2	VibroSwipe	95.790
6	2	3	Swipe	76.255

A tibble: 3 × 4

Conditions	variable	statistic	p
<chr>	<chr>	<dbl>	<dbl>
NoSwipe	WPM	0.888	0.111
Swipe	WPM	0.979	0.979
VibroSwipe	WPM	0.927	0.350

A anova_test: 1 × 7

	Effect	DFn	DFd	F	p	p<.05	pes
	<chr>	<dbl>	<dbl>	<dbl>	<dbl>	<chr>	<dbl>
1	Conditions	2	22	24.255	2.73e-06	*	0.688

	Pairwise comparisons using paired t tests 

data:  df$WPM and df$Conditions 

           NoSwipe Swipe  
Swipe      0.00012 -      
VibroSwipe 0.00022 1.00000

P value adjustment method: bonferroni 

library(rstatix)
library(dplyr) 
library(ggplot2) 

# A small function helping us to round  all numbers (0.123456 -> 0.123)
round_df <- function(df, digits = 3) {
    nums <- vapply(df, is.numeric, FUN.VALUE = logical(1))
    df[,nums] <- round(df[,nums], digits = digits)
    (df)
}

# Loading the CSV file
df <- read.csv("data-tlx-onefactorial.csv") 

# Show header if everything is OK
head(df)

# TLX = Mental Demand + Phyiscal Demand + Temporal Demand + Performance + Effort + Frustration
df$TLX <- df$MD + df$PD + df$TD + df$PE + df$EF + df$FR

# Check if data is normal distributed
round_df(df %>% group_by(Conditions) %>% shapiro_test(TLX))

# Nothing is significant, so we can run a three-way repeated-measures ANOVA
anova <- anova_test(data = df, dv = TLX, wid = SubjectID, within = Conditions,  effect.size = "pes")

# Auto correction of degree of freedom (if necessary)
get_anova_table(anova, correction = "auto")

# Pairwise Posthoc Comparison to understand where the differences occur
pairwise.t.test(df$TLX, df$Conditions, p.adj = "bonf", paired = TRUE)

# Let's aggregate the data
means <- df %>% group_by(Conditions) %>% summarize(
    Mean = mean(TLX),
    SD = sd(TLX),
    N = n(), .groups = 'drop',) %>% mutate(
    SE = SD / sqrt(N),
    CI = qt(1 - (0.05 / 2), N - 1) * SE) 

# Let's plot the data
plot <- ggplot(means, aes(y=Mean, x=Conditions, fill=Conditions)) +  
                    geom_bar(stat = "identity") + 
                    geom_errorbar(aes(ymin=Mean-CI, ymax=Mean+CI, width=.1)) + 
                    ylab("words per minute (WPM)") + 
                    xlab("")+
                    ggtitle("Typing performance") +
                    theme_minimal() +  
                    theme( 
                        legend.margin=margin(t = -5, unit='mm'),
                        plot.margin=unit(x=c(1,2,1,1),units="mm"),
                        legend.position = "none",
                        legend.title = element_blank(),
                        plot.title = element_text(hjust = 0.5), 
                        panel.background = element_rect(size=0.5, color="transparent", fill="gray98" )
                    ) 

options(repr.plot.width = 5, repr.plot.height = 5)

plot
ggsave("Swipe-WPM-BarChart.pdf", plot, width=5, height=5, device=cairo_pdf)

A data.frame: 6 × 9

	SubjectID	Sequence	Conditions	MD	PD	TD	PE	EF	FR
	<int>	<int>	<chr>	<int>	<int>	<int>	<int>	<int>	<int>
1	1	1	NoSwipe	1	11	12	9	6	0
2	1	2	Swipe	10	11	12	9	10	7
3	1	3	VibroSwipe	12	3	7	0	10	10
4	2	1	NoSwipe	1	6	3	2	2	10
5	2	2	VibroSwipe	12	4	13	8	1	9
6	2	3	Swipe	4	8	5	8	5	8

A tibble: 3 × 4

Conditions	variable	statistic	p
<chr>	<chr>	<dbl>	<dbl>
NoSwipe	TLX	0.929	0.368
Swipe	TLX	0.910	0.216
VibroSwipe	TLX	0.961	0.804

A anova_test: 1 × 7

	Effect	DFn	DFd	F	p	p<.05	pes
	<chr>	<dbl>	<dbl>	<dbl>	<dbl>	<chr>	<dbl>
1	Conditions	2	22	4.405	0.025	*	0.286

	Pairwise comparisons using paired t tests 

data:  df$TLX and df$Conditions 

           NoSwipe Swipe
Swipe      0.110   -    
VibroSwipe 0.099   1.000

P value adjustment method: bonferroni