R Notebook for Fink, Fiehn, Wald-Fuhrmann (2024) analyses

list.of.packages <- c("tidyverse", "ggpubr", "ez", "corrplot", "car", "stats","PerformanceAnalytics", "afex", "emmeans", "mlma", "ggplot2", "ggExtra", "psych", "ggpubr", "here", "gridExtra", "ggridges")
new.packages <- list.of.packages[!(list.of.packages %in% installed.packages()[,"Package"])]
if(length(new.packages)) install.packages(new.packages) 

here::i_am("Fink_Kentler_analyses.Rmd")
library(here)
source(here("dependencies.R"))
data <- read.csv(here("data.csv")) 
fig_path <- here("figures/") # path to save generated figures to
op = 0 # flag if want to save output plots to files (1) or plot inline (0)

# remove variables we don't need (Labvanced still outputting old vars from earlier exp. version)
data_new <- subset(data, select = - c(Block_Name, Task_Nr, aud_tend_unix, aud_ts, aud_ts_unix, condi3, copy_of_condi2, copy_of_condi, frame_end_unix, frame_rt, Name.Audio._copy_4b49, Name.Audio._copy_93a7, Name.Audio._copy_bd6f, Name.Audio._copy_c958, Name.Audio._copy_c96c, Name.Audio., subj_counter_global, subj_counter_per_group, time_delay_offset, time_measure_std, pixelDensityPerMM,unlocked))

# filter out people who have seen/heard stimuli before (some of our pilots) or who don't have normal hearing / vision
exposed = table(data$prev_exposure)
data_new <- data_new[-c(which(data_new$prev_exposure == "option_0")), ]
normSenses = table(data$Normal.Vision.) # no one without
data_new <- data_new[-c(which(data_new$completed == "no")), ] # remove incompletes
data = data_new # overwrite old data so can't accidentally use it

#create a dataframe just with the task block (for further analysis )
data_task <- data_new %>% 
  select(Trial_Nr, Trial_Id, correspondence_, enjoy_, Frame.Time., interestingness_, mood.congruency_, move_, subject_code, Block_Nr, group_name, exp_subject_id)

data_task <- data_task[, c("exp_subject_id", "Block_Nr", "Trial_Nr", "Trial_Id","group_name", "Frame.Time.", "correspondence_", "enjoy_", "interestingness_", "mood.congruency_", "move_" )]

# put time spent in secs (instead of ms)
data_task$Frame.Time. = data_task$Frame.Time. / 1000

# Add column for condition 
# Trial_Id tells which condition it is (1 - 4 = A; 5 - 8 = V; 9 - 12 = AVI; 13 - 16 = AVR)
# with Trial_Id and the group number we can look which specific stimuli are used 
data_task$Trial_Id = as.numeric(data_task$Trial_Id)
data_task$condition = NA
data_task$condition[data_task$Trial_Id <= 4] = 'Audio' 
data_task$condition[data_task$Trial_Id >= 5 & data_task$Trial_Id <= 8] = 'Visual' 
data_task$condition[data_task$Trial_Id >= 9 & data_task$Trial_Id <= 12] = 'AV-Intended' 
data_task$condition[data_task$Trial_Id >= 13] = 'AV-Random' 

# create subset data frame of task data
#just keep Block Number = 2 (these are the experimental trials)
data_task <- data_task %>%
  filter(Block_Nr == 2)

# create subset data frame with just participant info
data_participant <- data_new %>%
  select(exp_subject_id, group_name, age_input, gender, country_current, country_home, Modality_Preference, Ollen, overall_enjoy, "prev_exposure", realWorldLikelihood)

data_participant = na.omit(data_participant)

# calculate area scores for each participant and add to participant data
# AReA scale scoring
#Aesthetic Appreciation: Items 1, 2, 3, 4, 6, 9, 13, 14
#Intense Aesthetic Experience: Items 8, 11, 12, 13
#Creative Behavior: Items 5, 7, 10
area_data = data[data$Task_Name == 'AReA',]
area_data$area_aes_apprec = area_data$area_01 + area_data$area_02 + area_data$area_03 + area_data$area_04 + area_data$area_06 + area_data$area_09 + area_data$area_13 + area_data$area_14
area_data$area_aes_apprec = area_data$area_aes_apprec / 8 # divide by num of items
area_data$area_intense_aes_exp = area_data$area_08 + area_data$area_11 + area_data$area_12 + area_data$area_13
area_data$area_intense_aes_exp = area_data$area_intense_aes_exp / 4
area_data$area_creative_beh = area_data$area_05 + area_data$area_07 + area_data$area_10
area_data$area_creative_beh = area_data$area_creative_beh / 3
area_data$area_composite = (area_data$area_aes_apprec + area_data$area_intense_aes_exp + area_data$area_creative_beh) / 3

# add area scores to mean table (join based on exp_subj_id)
area_data = area_data[,c("exp_subject_id", "area_aes_apprec", "area_intense_aes_exp", "area_creative_beh", "area_composite")]

# merge area with participant df
data_participant = merge(x=data_participant,y=area_data, 
                         by="exp_subject_id", all=TRUE)

# clean up modality pref data to have meaningful labels
temp = data_participant$Modality_Preference
temp = gsub("option_0", "prefer_audio", temp)
temp = gsub("option_1", "prefer_visual", temp)
temp = gsub("option_2", "prefer_audiovisual", temp)
temp = gsub("option_3", "no_preference", temp)
data_participant$Modality_Preference_cat = temp

# join participant info with trial level data for final full data frame
full_data = merge(x=data_task,y=data_participant, 
                  by="exp_subject_id", all=TRUE)

# something wrong! only 195 obs for people who saw endQs ?! (these were required questions. not sure how labvanced let that through)
# table(data_new$Task_Name)

# check what going on with 6 subs
check = subset(full_data, is.na(full_data$Ollen))
check2 = subset(data, is.na(full_data$Ollen))

nsubs = length(unique(full_data$exp_subject_id))
nsubs_missingDemo = length(unique(check$exp_subject_id))

gender_dist = table(data_participant$gender)
age_mean = mean(data_participant$age_input, na.rm = TRUE)
age_std = sd(data_participant$age_input, na.rm = TRUE)
ncountries = length(unique(data_participant$country_current))
countries = table(data_participant$country_current)
modality_pref = table(data_participant$Modality_Preference_cat) 

print("Gender: man, woman, non-binary, prefer not to say")

[1] "Gender: man, woman, non-binary, prefer not to say"

print(gender_dist)

  0   1   2   3 
117  73   4   1 

print("Age (mean, sd, min, max)")

[1] "Age (mean, sd, min, max)"

print(age_mean)

[1] 31.01538

print(age_std)

[1] 11.81973

print(min(data_participant$age_input, na.rm = TRUE))

[1] 18

print(max(data_participant$age_input, na.rm = TRUE))

[1] 68

print("Unique countries:")

[1] "Unique countries:"

print(ncountries)

[1] 23

print(countries)

     Australia        Belgium         Canada          Chile Czech Republic        Ecuador 
             2              1              3              1              2              1 
       Germany         Greece        Hungary          India        Ireland          Italy 
             1              9              4              1              2             11 
        Mexico        Nigeria         Poland       Portugal   South Africa          Spain 
            12              2             23             34             15              9 
   Switzerland        Ukraine United Kingdom  United States 
             2              1             47             12 

print("AReA (mean, sd, min, max):")

[1] "AReA (mean, sd, min, max):"

print(mean(data_participant$area_composite))

[1] 2.62576

print(sd(data_participant$area_composite))

[1] 0.7410346

print(min(data_participant$area_composite))

[1] 1.125

print(max(data_participant$area_composite))

[1] 4.763889

print("AReA AA (mean, sd, min, max):")

[1] "AReA AA (mean, sd, min, max):"

print(mean(data_participant$area_aes_apprec))

[1] 3.439055

print(sd(data_participant$area_aes_apprec))

[1] 0.69193

print(min(data_participant$area_aes_apprec))

[1] 1.125

print(max(data_participant$area_aes_apprec))

[1] 5

print("AReA IAE (mean, sd, min, max):")

[1] "AReA IAE (mean, sd, min, max):"

print(mean(data_participant$area_intense_aes_exp))

[1] 2.461443

print(sd(data_participant$area_intense_aes_exp))

[1] 0.9202953

print(min(data_participant$area_intense_aes_exp))

[1] 1

print(max(data_participant$area_intense_aes_exp))

[1] 5

print("AReA CB (mean, sd, min, max):")

[1] "AReA CB (mean, sd, min, max):"

print(mean(data_participant$area_creative_beh))

[1] 1.976783

print(sd(data_participant$area_creative_beh))

[1] 0.9174315

print(min(data_participant$area_creative_beh))

[1] 1

print(max(data_participant$area_creative_beh))

[1] 5

# clean up and plot ollen data
ollenScale = c("non-musician", "music-loving non-musician", "amateur musician", "serious amateur musician", "semi-professional musician", "professional musician")

ollen = table(data_participant$Ollen)
ollen = data.frame(ollen)
names(ollen) = c("Response","Participants")
ollen$Response = ollenScale[1:5] # nobody chose professional!

print(ollen)

describe(ollen$Participants)

# order Ollen properly
ollen <- within(ollen, Response <- factor(Response, levels=ollenScale))

# create barplot
# png("Kentler_ollen.png", units="in", width=8, height=5, res=300)
ggplot(ollen, aes(x=Response, y=Participants)) + 
  geom_bar(stat = "identity") + coord_flip() + theme_classic() + theme(text = element_text(size = 20))

# Plot area data
ggplot(data_participant, aes(x=area_composite)) + 
  geom_histogram(bins = 20) + ggtitle("AReA composite score")

# Plot enjoyment data
ggplot(data_participant, aes(x=overall_enjoy)) + 
  geom_histogram(bins = 25) + ggtitle("Overall enjoyment of the experiment")

ggplot(data_participant, aes(x=realWorldLikelihood)) + 
  geom_histogram(bins = 25) + ggtitle("Likelihood of seeking similar pieces in real world")

print("Enjoyment (mean, sd)")

[1] "Enjoyment (mean, sd)"

print(mean(data_participant$overall_enjoy, na.rm=T))

[1] 84.58974

print(sd(data_participant$overall_enjoy, na.rm=T))

[1] 20.67136

print("Likelihood (mean, sd)")

[1] "Likelihood (mean, sd)"

print(mean(data_participant$realWorldLikelihood, na.rm=T))

[1] 54.81026

print(sd(data_participant$realWorldLikelihood, na.rm=T))

[1] 31.5477

# plot relation between enjoyment and likelihood
p <- ggplot(data_participant, aes(x=overall_enjoy, y=realWorldLikelihood)) +
  geom_point() + geom_smooth(method=lm , color="black", fill="gray", se=TRUE) +
  theme(legend.position="none") + theme_classic()

# Custom marginal plots:
p2 <- ggMarginal(p, type="densigram", fill = "gray") 
p2 

# Self-reported modality prefs
# create barplot of modality preferences
mpref = data.frame(modality_pref)
names(mpref) = c("Response","Participants")
mpref$percent = mpref$Participants / sum(mpref$Participants)


# png("Kentler_preference.png", units="in", width=8, height=5, res=300) 
ggplot(mpref, aes(x=Response, y=Participants)) + 
  geom_bar(stat = "identity") + coord_flip() + theme_classic() + theme(text = element_text(size = 20))

# create data frame for correlation
# need to convert categorical to numeric 
corrdf = data_participant[ , which(names(data_participant) %in% c("Ollen", "Modality_Preference", "overall_enjoy", "realWorldLikelihood", "area_aes_apprec", "area_intense_aes_exp", "area_creative_beh", "area_composite"))] 

corrdf$Modality_Preference = as.numeric(as.factor(corrdf$Modality_Preference))
corrdf$Ollen = as.numeric(as.factor(corrdf$Ollen))

chart.Correlation(corrdf, histogram=TRUE, pch=25, text.panel=labs)

# create zscore function
zscore <- function(input){
  out = (input - mean(input))/sd(input)
}

# zscore data
zscored = data_task %>%
  group_by(exp_subject_id) %>%
  mutate(z_enjoy = zscore(enjoy_), z_interesting = zscore(interestingness_), z_mood_cong = zscore(mood.congruency_), z_moved = zscore(move_))
# NOTE zscoring correspondence needs to be done only for AV conditions

# get means of zscored data
z_means = zscored %>%
  group_by(exp_subject_id, condition) %>%
  summarise(mean_timespent = mean(Frame.Time.), mean_enjoy = mean(z_enjoy), mean_interesting = mean(z_interesting), mean_mood_cong = mean(z_mood_cong), mean_moved = mean(z_moved))


# get means of zscored data
# z_means_participant = z_means %>%
#   group_by(exp_subject_id) %>%
#   summarise(mean_timespent = mean(mean_timespent), mean_enjoy = mean(mean_enjoy), mean_interesting = mean(mean_interesting), mean_mood_cong = mean(mean_mood_cong), mean_moved = mean(mean_moved))
# 

# get raw means in case want to compare to zscored data
means = data_task %>%
  group_by(exp_subject_id, condition) %>%
  summarise(mean_timespent = mean(Frame.Time.), mean_enjoy = mean(enjoy_), mean_interesting = mean(interestingness_), mean_mood_cong = mean(mood.congruency_), mean_moved = mean(move_), mean_corresp = mean(correspondence_))

# get raw sds in case want to compare to zscored data
sds = data_task %>%
  group_by(exp_subject_id, condition) %>%
  summarise(mean_timespent = sd(Frame.Time.), mean_enjoy = sd(enjoy_), mean_interesting = sd(interestingness_), mean_mood_cong = sd(mood.congruency_), mean_moved = sd(move_), mean_corresp = sd(correspondence_))

# Function to plot violins for each condition
# Input: dataframe, variable to plot, title for plot
plotViolin = function(data, groupvar, var, titstr, xlab, ylab){
  
  # define colors
  AVR = '#8c1aff'
  AVI = '#cc99ff'
  V = '#ff4d4d'
  A = '#6699ff'
  ourcolors = c(A, AVI, AVR, V)
  
  # only AV conditions relevant
  if (var == "z_corresp"){
    ourcolors = c(AVI, AVR) 
  }
  
  # Plot
  var = sym(var)
  groupvar = sym(groupvar)

  pl = data %>%
    ggplot(aes(x=!!groupvar, y=!!var, fill=!!groupvar)) +
    geom_violin(legend = FALSE, trim = FALSE, adjust = 1, draw_quantiles = c(0.25, 0.75), size=.5, linetype = "dashed", color="black") + #.8 
    geom_violin(legend = FALSE, trim = FALSE, adjust = 1, draw_quantiles = 0.5, fill="transparent", size=.85, color="black") + #1.2
    
    geom_jitter(size=0.4, alpha=0.6, color="gray") +
    scale_fill_manual(values=ourcolors) +
    #theme_classic() +
    theme_ridges() +
    theme(
      legend.position="none",
      plot.title = element_text(size=11)
    ) +
    ggtitle(titstr) +
    xlab(xlab) +
    ylab(ylab)
  
  return(pl)
  
}



# Function to create density ridge plots for each condition
# Input: dataframe, variable to plot, title for plot
plotRidges = function(data, groupvar, var, titstr, xlab, ylab){
  
  # define colors
  AVR = '#CCBB44'
  AVI = '#228833'
  V = '#CC6677'
  A = '#88CCEE'
  ourcolors = c(A, AVI, AVR, V)
  
  # only AV conditions relevant
  if (var == "z_corresp"){
    ourcolors = c(AVI, AVR) 
  }
  
  # Plot
  var = sym(var)
  groupvar = sym(groupvar)

  pl = data %>%
    ggplot(aes(x=!!var, y=!!groupvar, fill=!!groupvar)) +
    
   # stat_density_ridges(quantile_lines=TRUE, quantiles = c(0.025, 0.975), color="black", alpha = 1) +
    
    geom_density_ridges(inherit.aes = TRUE, jittered_points = TRUE, stat ="density_ridges",
      position = position_points_jitter(width = 0.05, height = 0),
      point_shape = '|', point_size = 3, point_alpha = .9, alpha = 0.7) +
    
    geom_density_ridges(alpha = 0.85, quantile_lines=TRUE,
                        color="white", quantile_fun=function(var,...)mean(var)) +
    
    scale_fill_manual(values=ourcolors) +
    xlab(ylab) +
    theme_ridges() + theme(legend.position = "none") +
    theme(axis.title.x = element_text(hjust = 0.5)) + 
          theme(axis.title.y=element_blank())
  
  return(pl)
  
}


# Function to run rm anova and output stats
runaov = function(id, dv, data, w){
  fit = aov_ez(id,dv,data,within=w)
  print(fit)
  fit2 <- lsmeans(fit,specs = c(w))
  contrast(fit2,method="pairwise",adjust="holm")
}

# Conditions to compare after ANOVAs
my_comparisons <- list( c("Audio", "AV-Intended"), c("Audio", "AV-Random"), c("Audio", "Visual"), c("AV-Intended", "AV-Random"),  c("AV-Intended", "Visual"), c("AV-Random", "Visual"))
my_comparisons = rev(my_comparisons) # reverse order so more readable on plots

# Choose plotting style (violin or ridge) for future plots
customplot = plotRidges
hide_ns = FALSE

z_means$logtime = log(z_means$mean_timespent)
p = customplot(z_means, "condition", "mean_timespent", "", "Condition", "Mean Time Spent (secs)")
plotTimeSecs = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired = "true", na.rm=TRUE, hide.ns=hide_ns)

p = customplot(z_means, "condition", "logtime", "", "", "Mean Time Spent (log secs)")
plotTimeLog = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired = "true", na.rm=TRUE, hide.ns=hide_ns)
z_means$exp_subject_id = as.factor(z_means$exp_subject_id)

runaov("exp_subject_id","mean_timespent", z_means, "condition")

Anova Table (Type 3 tests)

Response: mean_timespent
     Effect           df    MSE         F  ges p.value
1 condition 2.00, 400.21 237.00 96.93 *** .100   <.001
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

Sphericity correction method: GG 
 contrast                estimate    SE  df t.ratio p.value
 AV.Intended - AV.Random   -0.616 0.948 200  -0.650  0.5165
 AV.Intended - Audio       -4.908 0.800 200  -6.134  <.0001
 AV.Intended - Visual      15.067 1.469 200  10.256  <.0001
 AV.Random - Audio         -4.292 0.942 200  -4.554  <.0001
 AV.Random - Visual        15.683 1.514 200  10.361  <.0001
 Audio - Visual            19.975 1.600 200  12.481  <.0001

P value adjustment: holm method for 6 tests 

# print means and sds
cond_means = z_means %>%
  group_by(condition) %>%
  summarise(mean_timespent = mean(mean_timespent, na.rm=T))   

cond_sds = z_means %>%
  group_by(condition) %>%
  summarise(sd_timespent = sd(mean_timespent, na.rm=T))

print(cond_means)

print(cond_sds)

plotTimeSecs

plotTimeLog

plotTimeSecs = plotTimeSecs + labs(title = "Time Spent", tag = "A") + theme(plot.title = element_text(hjust = 0.5, face='bold'), plot.tag = element_text(face='bold'))

p = customplot(z_means, "condition", "mean_enjoy", "", "Condition", "Mean Enjoyment (z-score)")
plotEnjoy = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired = "true", na.rm=TRUE, hide.ns=hide_ns)
runaov("exp_subject_id","mean_enjoy", z_means, "condition")

Anova Table (Type 3 tests)

Response: mean_enjoy
     Effect           df  MSE         F  ges p.value
1 condition 2.76, 552.98 0.33 13.83 *** .065   <.001
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

Sphericity correction method: GG 
 contrast                estimate     SE  df t.ratio p.value
 AV.Intended - AV.Random -0.01486 0.0484 200  -0.307  1.0000
 AV.Intended - Audio      0.28382 0.0524 200   5.416  <.0001
 AV.Intended - Visual    -0.00333 0.0544 200  -0.061  1.0000
 AV.Random - Audio        0.29869 0.0546 200   5.475  <.0001
 AV.Random - Visual       0.01154 0.0544 200   0.212  1.0000
 Audio - Visual          -0.28715 0.0654 200  -4.388  0.0001

P value adjustment: holm method for 6 tests 

plotEnjoy

plotEnjoy = plotEnjoy + labs(title = "Enjoyment", tag = "C") + theme(plot.title = element_text(hjust = 0.5, face='bold'), plot.tag = element_text(face='bold'))

p = customplot(z_means, "condition", "mean_interesting", "", "Condition", "Mean Interestingness (z-score)")
plotInterest = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired = "true", na.rm=TRUE, hide.ns=hide_ns)
runaov("exp_subject_id","mean_interesting", z_means, "condition")

Anova Table (Type 3 tests)

Response: mean_interesting
     Effect           df  MSE         F  ges p.value
1 condition 2.65, 526.68 0.35 13.06 *** .062   <.001
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

Sphericity correction method: GG 
 contrast                estimate     SE  df t.ratio p.value
 AV.Intended - AV.Random  -0.0412 0.0460 199  -0.897  0.5345
 AV.Intended - Audio       0.2781 0.0530 199   5.249  <.0001
 AV.Intended - Visual      0.0637 0.0573 199   1.112  0.5345
 AV.Random - Audio         0.3193 0.0537 199   5.943  <.0001
 AV.Random - Visual        0.1050 0.0532 199   1.973  0.1496
 Audio - Visual           -0.2144 0.0680 199  -3.154  0.0075

P value adjustment: holm method for 6 tests 

plotInterest

plotInterest = plotInterest + labs(title = "Interestingness", tag = "D") + theme(plot.title = element_text(hjust = 0.5, face='bold'), plot.tag = element_text(face='bold'))

p = customplot(z_means, "condition", "mean_mood_cong", "", "Condition", "Mean Mood Congruency (z-score)")
plotMood = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired = "true", na.rm=TRUE, hide.ns=hide_ns)
runaov("exp_subject_id","mean_mood_cong", z_means, "condition")

Anova Table (Type 3 tests)

Response: mean_mood_cong
     Effect           df  MSE      F  ges p.value
1 condition 2.84, 564.37 0.30 3.61 * .018    .015
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

Sphericity correction method: GG 
 contrast                estimate     SE  df t.ratio p.value
 AV.Intended - AV.Random  -0.0256 0.0454 199  -0.563  0.5910
 AV.Intended - Audio       0.0896 0.0526 199   1.703  0.3607
 AV.Intended - Visual     -0.0827 0.0540 199  -1.533  0.3809
 AV.Random - Audio         0.1152 0.0519 199   2.219  0.1380
 AV.Random - Visual       -0.0571 0.0545 199  -1.049  0.5910
 Audio - Visual           -0.1723 0.0609 199  -2.831  0.0307

P value adjustment: holm method for 6 tests 

plotMood

plotMood = plotMood + labs(title = "Mood Congruency", tag = "E") + theme(plot.title = element_text(hjust = 0.5, face='bold'), plot.tag = element_text(face='bold'))

p = customplot(z_means, "condition", "mean_moved", "", "Condition", "Mean Feeling Moved (z-score)")
runaov("exp_subject_id","mean_moved", z_means, "condition")

Anova Table (Type 3 tests)

Response: mean_moved
     Effect           df  MSE        F  ges p.value
1 condition 2.83, 561.24 0.30 9.12 *** .044   <.001
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

Sphericity correction method: GG 
 contrast                estimate     SE  df t.ratio p.value
 AV.Intended - AV.Random  -0.0265 0.0453 198  -0.584  1.0000
 AV.Intended - Audio       0.1667 0.0539 198   3.092  0.0068
 AV.Intended - Visual      0.1987 0.0544 198   3.651  0.0013
 AV.Random - Audio         0.1932 0.0520 198   3.714  0.0013
 AV.Random - Visual        0.2252 0.0535 198   4.209  0.0002
 Audio - Visual            0.0320 0.0611 198   0.523  1.0000

P value adjustment: holm method for 6 tests 

plotMoved = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired = "true", na.rm=TRUE, hide.ns=hide_ns)
plotMoved

plotMoved = plotMoved + labs(title = "Feeling Moved", tag = "F") + theme(plot.title = element_text(hjust = 0.5, face='bold'), plot.tag = element_text(face='bold'))

# only zscore within relevant AV conditions
correspdata = data_task[data_task$condition == "AV-Intended" | data_task$condition == "AV-Random", ]
zscored_corresp = correspdata %>%
  group_by(exp_subject_id) %>%
  mutate(z_corresp = zscore(correspondence_)) 

mean_z_corresp = zscored_corresp %>%
  group_by(exp_subject_id, condition) %>%
  summarise(z_corresp = mean(z_corresp))

# plot and calculate stats
p <- customplot(mean_z_corresp, "condition", "z_corresp", "", "Condition", "Mean Correspondence (z-score)")
plotAVcorresp = p + stat_compare_means(comparisons = list(c("AV-Intended", "AV-Random")), label = "p.signif", method = "t.test", paired="true", label.x.npc=.5)
compare_means(z_corresp ~ condition,  data = mean_z_corresp, method = "t.test", hide.ns=hide_ns)

plotAVcorresp

plotAVcorresp = plotAVcorresp + labs(title = "Audiovisual Correspondence", tag = "B") + theme(plot.title = element_text(hjust = 0.5, face='bold'), plot.tag = element_text(face='bold'))
 

if (op) {tiff(paste(fig_path, "combined_DVs_ridge_r1.tiff", sep=""), units="in", width=9, height=11, res=300)}
plotAll = grid.arrange(plotTimeSecs, plotAVcorresp, plotEnjoy, plotInterest, plotMood, plotMoved, ncol = 2)

plotAll

TableGrob (3 x 2) "arrange": 6 grobs

ggsave("combined_DVs_ridge_r1.svg", plotAll, width=9, height=11, units = "in")

# remove ratio diff and other columns we don't need 
ratings = z_means[ , -which(names(z_means) %in% c("exp_subject_id","condition", "mean_corresp", "logtime"))] # drop label columns and ones not using
ratings$mean_timespent = log(ratings$mean_timespent) # TODO check if want to do this or not

# Capture the plot so we can use it again later
chart.Correlation(ratings, histogram=TRUE, pch=25, text.panel=labs)

# remove ratio diff and other columns we don't need 
ratings = subset(z_means, condition == 'Audio')
ratings = ratings[ , -which(names(ratings) %in% c("exp_subject_id","condition", "mean_corresp", "logtime"))] # drop label columns and ones not using
ratings$mean_timespent = log(ratings$mean_timespent)

chart.Correlation(ratings, histogram=TRUE, pch=25, text.panel=labs)

# remove ratio diff and other columns we don't need 
ratings = subset(z_means, condition == 'Visual')
ratings = ratings[ , -which(names(ratings) %in% c("exp_subject_id","condition", "mean_corresp", "logtime"))] # drop label columns and ones not using
ratings$mean_timespent = log(ratings$mean_timespent)

chart.Correlation(ratings, histogram=TRUE, pch=25, text.panel=labs)

# remove ratio diff and other columns we don't need 
ratings = subset(z_means, condition == 'AV-Intended')
corresp = subset(mean_z_corresp, condition == 'AV-Intended')
#ratings$mean_corresp = mean_z_corresp
ratings <- merge(x=ratings,y=corresp, 
                by="exp_subject_id")
ratings <- ratings %>% # rename
  rename(mean_corresp = z_corresp)
ratings = ratings[ , -which(names(ratings) %in% c("exp_subject_id","condition", "condition.x", "condition.y", "logtime"))] 
ratings$mean_timespent = log(ratings$mean_timespent)

chart.Correlation(ratings, histogram=TRUE, pch=25, text.panel=labs)

# remove ratio diff and other columns we don't need 
ratings = subset(z_means, condition == 'AV-Random')
corresp = subset(mean_z_corresp, condition == 'AV-Random')
ratings <- merge(x=ratings,y=corresp, 
                by="exp_subject_id")
ratings <- ratings %>% # rename
  rename(mean_corresp = z_corresp)
ratings = ratings[ , -which(names(ratings) %in% c("exp_subject_id","condition", "condition.x", "condition.y", "logtime"))] 
ratings$mean_timespent = log(ratings$mean_timespent)


chart.Correlation(ratings, histogram=TRUE, pch=25, text.panel=labs)

# merge participants means and demo data
pall = merge(data_participant, z_means, by="exp_subject_id")

# remove columns we don't need 
pall = pall[ , -which(names(pall) %in% c("country_current", "country_home","prev_exposure","area_intense_aes_exp","Modality_Preference_cat", "area_creative_beh","area_aes_apprec"))]  
# "gender","Modality_Preference","Ollen",, "exp_subject_id","group_name",
pall$mean_timespent = log(pall$mean_timespent)

# make sure variable right type
# Remove the 'option_' prefix and keep only the numbers for the ordered categorical columns
data$Modality_Preference <- gsub('option_', '', data$Modality_Preference)
# Convert the column to numeric
data$Modality_Preference <- as.numeric(data$Modality_Preference)
# Convert to ordered factor
pall$Modality_Preference = factor(pall$Modality_Preference)

# Do same for ollen scale
data$Ollen <- gsub('option_', '', data$Ollen)
data$Ollen <- as.numeric(data$Ollen)
pall$Ollen = factor(pall$Ollen)

# gender is unordered
pall$gender = factor(pall$gender, ordered = F)

#chart.Correlation(pall, histogram=TRUE, pch=25, text.panel=labs)

# organize data
forfa = z_means[ , -which(names(z_means) %in% c("mean_timespent","logtime", "condition","exp_subject_id", "factor_score"))] 

# check for missing values
nna = sum(is.na(forfa)) # 16 nas in the df
# N.B. NAs have been introduced by z-scoring, due to dividing by zero (no variance)
# Since the true mean was zero, it is fair to impute with zero 
forfa$mean_enjoy[is.na(forfa$mean_enjoy)] <- 0
forfa$mean_mood_cong[is.na(forfa$mean_mood_cong)] <- 0
forfa$mean_moved[is.na(forfa$mean_moved)] <- 0
forfa$mean_interesting[is.na(forfa$mean_interesting)] <- 0

# Do tests to determine suitability for FA
print("KMO")

[1] "KMO"

KMO(r=cor(forfa))

Kaiser-Meyer-Olkin factor adequacy
Call: KMO(r = cor(forfa))
Overall MSA =  0.8
MSA for each item = 
      mean_enjoy mean_interesting   mean_mood_cong       mean_moved 
            0.73             0.75             0.88             0.90 

print("Bartlett")

[1] "Bartlett"

cortest.bartlett(forfa)

$chisq
[1] 2127.938

$p.value
[1] 0

$df
[1] 6

print("Determinant")

[1] "Determinant"

det(cor(forfa))

[1] 0.07014754

fafitfree <- fa(forfa, nfactors = ncol(forfa), rotate = "none")
n_factors <- length(fafitfree$e.values)
scree     <- data.frame(
  Factor_n =  as.factor(1:n_factors), 
  Eigenvalue = fafitfree$e.values)
ggplot(scree, aes(x = Factor_n, y = Eigenvalue, group = 1)) + 
  geom_point() + geom_line() +
  xlab("Number of factors") +
  ylab("Initial eigenvalue") +
  labs( title = "Scree Plot", 
        subtitle = "(Based on the unreduced correlation matrix)")

fa.none <- fa(r=forfa, 
              nfactors = 1, 
              fm="pa", # type of factor analysis we want to use (“pa” is principal axis factoring)
              max.iter=100,
              rotate="varimax") # none rotation
print(fa.none)

Factor Analysis using method =  pa
Call: fa(r = forfa, nfactors = 1, rotate = "varimax", max.iter = 100, 
    fm = "pa")
Standardized loadings (pattern matrix) based upon correlation matrix

                PA1
SS loadings    2.75
Proportion Var 0.69

Mean item complexity =  1
Test of the hypothesis that 1 factor is sufficient.

df null model =  6  with the objective function =  2.66 with Chi Square =  2127.94
df of  the model are 2  and the objective function was  0.07 

The root mean square of the residuals (RMSR) is  0.03 
The df corrected root mean square of the residuals is  0.05 

The harmonic n.obs is  804 with the empirical chi square  9.47  with prob <  0.0088 
The total n.obs was  804  with Likelihood Chi Square =  53.1  with prob <  2.9e-12 

Tucker Lewis Index of factoring reliability =  0.928
RMSEA index =  0.178  and the 90 % confidence intervals are  0.139 0.221
BIC =  39.72
Fit based upon off diagonal values = 1
Measures of factor score adequacy             
                                                   PA1
Correlation of (regression) scores with factors   0.96
Multiple R square of scores with factors          0.93
Minimum correlation of possible factor scores     0.85

z_means['factor_score'] <- fa.none$scores

p = plotRidges(z_means, "condition", "factor_score", "", "Condition", "Mean Factor Score")
runaov("exp_subject_id","factor_score", z_means, "condition") 

Anova Table (Type 3 tests)

Response: factor_score
     Effect           df  MSE         F  ges p.value
1 condition 2.72, 543.46 1.29 11.85 *** .056   <.001
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

Sphericity correction method: GG 
 contrast                estimate     SE  df t.ratio p.value
 AV.Intended - AV.Random  -0.0506 0.0916 200  -0.552  0.9428
 AV.Intended - Audio       0.5237 0.1047 200   5.001  <.0001
 AV.Intended - Visual      0.0774 0.1073 200   0.722  0.9428
 AV.Random - Audio         0.5743 0.1058 200   5.429  <.0001
 AV.Random - Visual        0.1280 0.1047 200   1.222  0.6691
 Audio - Visual           -0.4463 0.1300 200  -3.434  0.0029

P value adjustment: holm method for 6 tests 

factorPlot = p + stat_compare_means(comparisons = my_comparisons, label = "p.signif", method = "t.test", paired="True", na.rm=TRUE)
factorPlot = factorPlot + labs(y = "Mean Factor Score (Aesthetic Experience)")
factorPlot

ratings = z_means[ , -which(names(z_means) %in% c("exp_subject_id","condition", "mean_corresp"))] # drop label columns and ones not using
ratings$mean_timespent = log(ratings$mean_timespent)

par(mfrow = c(3,1))
chart.Correlation(ratings, histogram=TRUE, pch=25, text.panel=labs)

fa.diagram(fa.none, digits=2, cut = .3, simple=TRUE)

#factorPlot

# merge mean table and participant level table
df = pall

# plot relation between area and feeling moved
p <- ggplot(df, aes(x=area_composite, y=mean_moved)) +
  geom_point() + geom_smooth(method=lm , color="black", fill="gray", se=TRUE) +
  theme(legend.position="none") + theme_classic()

# Custom marginal plots:
p2 <- ggMarginal(p, type="densigram", fill = "gray") 
p2 

# plot relation between area and timespent
p <- ggplot(df, aes(x=area_composite, y=mean_timespent)) +
  geom_point() + geom_smooth(method=lm , color="black", fill="gray", se=TRUE) +
  theme(legend.position="none") + theme_classic()

# Custom marginal plots:
p2 <- ggMarginal(p, type="densigram", fill = "gray") 
p2 

# merge mean table and participant level table
# pvis = pall[, pall$condition == 'Visual']
# df = pvis


pvis <- pall %>%
  filter(condition == "Visual")

df = pvis

# plot relation between area and feeling moved
p <- ggplot(df, aes(x=area_composite, y=mean_moved)) +
  geom_point() + geom_smooth(method=lm , color="black", fill="gray", se=TRUE) +
  theme(legend.position="none") + theme_classic()

# Custom marginal plots:
p2 <- ggMarginal(p, type="densigram", fill = "gray") 
p2 

# plot relation between area and timespent
p <- ggplot(df, aes(x=area_composite, y=mean_timespent)) +
  geom_point() + geom_smooth(method=lm , color="black", fill="gray", se=TRUE) +
  theme(legend.position="none") + theme_classic()

# Custom marginal plots:
p2 <- ggMarginal(p, type="densigram", fill = "gray") 
p2 

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