Main File for S2 main CAM study

Author

Julius Fenn, Louisa Estadieu

Notes

prepare data

set up data.frame questionnaires

setwd("outputs")
# > pre study
suppressMessages(read_file('preCAM.txt') %>%
                   # ... split it into lines ...
                   str_split('\n') %>% first() %>%
                   # ... filter empty rows ...
                   discard(function(x) x == '') %>%
                   discard(function(x) x == '\r') %>%
                   # ... parse JSON into a data.frame
                   map_dfr(fromJSON, flatten=TRUE)) -> dat_preCAM
# > post first CAM
suppressMessages(read_file('postCAM.txt') %>%
                   # ... split it into lines ...
                   str_split('\n') %>% first() %>%
                   # ... filter empty rows ...
                   discard(function(x) x == '') %>%
                   discard(function(x) x == '\r') %>%
                   # ... parse JSON into a data.frame
                   map_dfr(fromJSON, flatten=TRUE)) -> dat_postCAM

# > post second CAM
suppressMessages(read_file('secondPostCAM.txt') %>%
                   # ... split it into lines ...
                   str_split('\n') %>% first() %>%
                   # ... filter empty rows ...
                   discard(function(x) x == '') %>%
                   discard(function(x) x == '\r') %>%
                   # ... parse JSON into a data.frame
                   map_dfr(fromJSON, flatten=TRUE)) -> dat_secondPostCAM



########################################
# create counter variable for both data sets
########################################
### pre study
dat_preCAM$ID <- NA

tmp_IDcounter <- 0
for(i in 1:nrow(dat_preCAM)){
  if(!is.na(dat_preCAM$sender[i]) && dat_preCAM$sender[i] == "Greetings"){
    # tmp <- dat_preCAM$prolific_pid[i]
    tmp_IDcounter = tmp_IDcounter + 1
  }
  dat_preCAM$ID[i] <- tmp_IDcounter
}



### post study
dat_postCAM$ID <- NA

tmp_IDcounter <- 0
for(i in 1:nrow(dat_postCAM)){
  if(!is.na(dat_postCAM$sender[i]) && dat_postCAM$sender[i] == "CAMfeedbackGeneral"){
    # tmp <- dat_postCAM$prolific_pid[i]
    tmp_IDcounter = tmp_IDcounter + 1
  }
  dat_postCAM$ID[i] <- tmp_IDcounter
}

### second post study
#> fix error in "sender variable"
for(i in 1:nrow(dat_secondPostCAM)){
  if(is.na(dat_secondPostCAM$sender[i])){
    if(!is.na(dat_secondPostCAM$sender[i+1])){
      dat_secondPostCAM$sender[i] <- "adaptiveAnswer"
    }
  }
}



dat_secondPostCAM$ID <- NA
tmp_IDcounter <- 0
for(i in 1:nrow(dat_secondPostCAM)){
  if(!is.na(dat_secondPostCAM$sender[i]) && dat_secondPostCAM$sender[i] == "adaptiveAnswer"){
    # tmp <- dat_secondPostCAM$prolific_pid[i]
    tmp_IDcounter = tmp_IDcounter + 1
  }
  dat_secondPostCAM$ID[i] <- tmp_IDcounter
}


########################################
# keep only complete data sets
########################################
### pre-study
sort(table(dat_preCAM$ID))


 1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 
27 28 29 30 
10 10 10 10

sum(table(dat_preCAM$ID) != max(table(dat_preCAM$ID)))

[1] 0

sum(table(dat_preCAM$ID) == max(table(dat_preCAM$ID)))

[1] 30

dat_preCAM <- dat_preCAM[dat_preCAM$ID %in% names(table(dat_preCAM$ID))[table(dat_preCAM$ID) == max(table(dat_preCAM$ID))],]

### post-study
sort(table(dat_postCAM$ID))


 0 30  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 
 1  4  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5 
25 26 27 28 29 
 5  5  5  5  5

sum(table(dat_postCAM$ID) != max(table(dat_postCAM$ID)))

[1] 2

sum(table(dat_postCAM$ID) == max(table(dat_postCAM$ID)))

[1] 29

# dat_postCAM <- dat_postCAM[dat_postCAM$ID %in% names(table(dat_postCAM$ID))[table(dat_postCAM$ID) == max(table(dat_postCAM$ID))],]
dat_postCAM <- dat_postCAM[dat_postCAM$ID %in% names(table(dat_postCAM$ID))[table(dat_postCAM$ID) >= 4],]

### post-study second
sort(table(dat_secondPostCAM$ID))


 0 30  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 
 1 11 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 
25 26 27 28 29 
12 12 12 12 12

sum(table(dat_secondPostCAM$ID) != max(table(dat_secondPostCAM$ID)))

[1] 2

sum(table(dat_secondPostCAM$ID) == max(table(dat_secondPostCAM$ID)))

[1] 29

# dat_secondPostCAM <- dat_secondPostCAM[dat_secondPostCAM$ID %in% names(table(dat_secondPostCAM$ID))[table(dat_secondPostCAM$ID) == max(table(dat_secondPostCAM$ID))],]
dat_secondPostCAM <- dat_secondPostCAM[dat_secondPostCAM$ID %in% names(table(dat_secondPostCAM$ID))[table(dat_secondPostCAM$ID) >= 11],]



########################################
# json (from JATOS) to 2D data.frame
########################################
################################ pre-study

################################ post-study

################################ post-study second
tmp_numeric <- str_subset(string = colnames(dat_secondPostCAM), pattern = "^GAToRS|^Almere|^LiWang")


vec_ques <- c("ans1",
                tmp_numeric,
                "feedback_critic")

vec_notNumeric = c("ans1",
                   "feedback_critic")

questionnaire_secondPostCAM <- questionnairetype(dataset = dat_secondPostCAM, 
                                        listvars = vec_ques, 
                                        notNumeric = vec_notNumeric, verbose = FALSE)


dim(questionnaire_secondPostCAM)

[1] 30 38

set up CAM data

pre

Load CAM data

setwd("outputs")
suppressMessages(read_file("CAMdata.txt") %>%
  # ... split it into lines ...
  str_split('\n') %>% first() %>%
    discard(function(x) x == '') %>%
    discard(function(x) x == '\r') %>%
  # ... filter empty rows ...
  discard(function(x) x == '')) -> dat_CAM_pre

raw_CAM_pre <- list()
for(i in 1:length(dat_CAM_pre)){
  raw_CAM_pre[[i]] <- jsonlite::fromJSON(txt = dat_CAM_pre[[i]])
}

Create CAM files, draw CAMs and compute network indicators

########################################
# create CAM single files (nodes, connectors, merged)
########################################
CAMfiles_pre <- create_CAMfiles(datCAM = raw_CAM_pre, reDeleted = TRUE)

Nodes and connectors, which were deleted by participants were removed. 
 # deleted nodes:  49 
 # deleted connectors:  13

########################################
# draw CAMs
########################################
CAMdrawn_pre <- draw_CAM(dat_merged = CAMfiles_pre[[3]],
                     dat_nodes = CAMfiles_pre[[1]],ids_CAMs = "all",
                     plot_CAM = FALSE,
                     useCoordinates = TRUE,
                     relvertexsize = 3,
                     reledgesize = 1)

processing 30 CAMs... 
[1] "== ids_CAMs in drawnCAM"

########################################
# draw CAMs
########################################
tmp_microIndicator <- c("Rettungsroboter", "sozialer Assistenzroboter", "Vorteile", "Nachteile")
networkIndicators_pre <- compute_indicatorsCAM(drawn_CAM = CAMdrawn_pre, 
                                           micro_degree = tmp_microIndicator, 
                                           micro_valence = tmp_microIndicator, 
                                           micro_centr_clo = tmp_microIndicator, 
                                           micro_transitivity = tmp_microIndicator, 
                                           largestClique = FALSE)


########################################
# wordlists
########################################
CAMwordlist_pre <- create_wordlist(
  dat_nodes =  CAMfiles_pre[[1]],
  dat_merged =  CAMfiles_pre[[3]],
  order = "frequency",
  splitByValence = FALSE,
  comments = TRUE,
  raterSubsetWords = NULL,
  rater = FALSE
)

processing 30 CAMs... 
[1] "== ids_CAMs in drawnCAM"

DT::datatable(CAMwordlist_pre, options = list(pageLength = 5))

save CAMs as .json files, and as .png (igraph)

save_CAMs_as_pictures = FALSE

if(save_CAMs_as_pictures){
  setwd("outputs")

setwd("savedCAMs_pre")
setwd("png")
### remove all files if there are any
if(length(list.files()) >= 1){
  file.remove(list.files())
  cat('\n!
      all former .png files have been deleted')
}

### if no participant ID was provided replace by randomly generated CAM ID

if(all(CAMfiles_pre[[3]]$participantCAM.x == "noID")){
  CAMfiles_pre[[3]]$participantCAM.x <- CAMfiles_pre[[3]]$CAM.x
}

### save as .json files, and as .png (igraph)
ids_CAMs <- unique(CAMfiles_pre[[3]]$participantCAM.x); length(ids_CAMs)


for(i in 1:length(ids_CAMs)){
  save_graphic(filename = paste0("CAM_", i, "_t1")) #  paste0(ids_CAMs[i]))
  CAM_igraph <- CAMdrawn_pre[[c(1:length(CAMdrawn_pre))[
    names(CAMdrawn_pre) == paste0(unique(CAMfiles_pre[[3]]$participantCAM.x)[i])]]]
  plot(CAM_igraph, edge.arrow.size = .7,
       layout=layout_nicely, vertex.frame.color="black", asp = .5, margin = -0.1,
       vertex.size = 10, vertex.label.cex = .9)
  dev.off()
}

setwd("../json")
### remove all files if there are any
if(length(list.files()) >= 1){
  file.remove(list.files())
  cat('\n!
      all former .json files have been deleted')
}
for(i in 1:length(raw_CAM_pre)){
  if(!is_empty(raw_CAM_pre[[i]]$nodes)){
    if(nrow(raw_CAM_pre[[i]]$nodes) > 5){
      write(toJSON(raw_CAM_pre[[i]], encoding = "UTF-8"),
            paste0(raw_CAM_pre[[i]]$idCAM, ".json"))
    }
  }
}
}

post

Load CAM data

setwd("outputs")
suppressMessages(read_file("secondCAMdata.txt") %>%
  # ... split it into lines ...
  str_split('\n') %>% first() %>%
    discard(function(x) x == '') %>%
    discard(function(x) x == '\r') %>%
  # ... filter empty rows ...
  discard(function(x) x == '')) -> dat_CAM_post

raw_CAM_post <- list()
for(i in 1:length(dat_CAM_post)){
  raw_CAM_post[[i]] <- jsonlite::fromJSON(txt = dat_CAM_post[[i]])
}

Create CAM files, draw CAMs and compute network indicators

########################################
# create CAM single files (nodes, connectors, merged)
########################################
CAMfiles_post <- create_CAMfiles(datCAM = raw_CAM_post, reDeleted = TRUE)

Nodes and connectors, which were deleted by participants were removed. 
 # deleted nodes:  23 
 # deleted connectors:  10

########################################
# draw CAMs
########################################
CAMdrawn_post <- draw_CAM(dat_merged = CAMfiles_post[[3]],
                     dat_nodes = CAMfiles_post[[1]],ids_CAMs = "all",
                     plot_CAM = FALSE,
                     useCoordinates = TRUE,
                     relvertexsize = 3,
                     reledgesize = 1)

processing 30 CAMs... 
[1] "== ids_CAMs in drawnCAM"

########################################
# draw CAMs
########################################
tmp_microIndicator <- c("Rettungsroboter", "sozialer Assistenzroboter", "Vorteile", "Nachteile")
networkIndicators_post <- compute_indicatorsCAM(drawn_CAM = CAMdrawn_post, 
                                           micro_degree = tmp_microIndicator, 
                                           micro_valence = tmp_microIndicator, 
                                           micro_centr_clo = tmp_microIndicator, 
                                           micro_transitivity = tmp_microIndicator, 
                                           largestClique = FALSE)


########################################
# wordlists
########################################
CAMwordlist_post <- create_wordlist(
  dat_nodes =  CAMfiles_post[[1]],
  dat_merged =  CAMfiles_post[[3]],
  order = "frequency",
  splitByValence = FALSE,
  comments = TRUE,
  raterSubsetWords = NULL,
  rater = FALSE
)

processing 30 CAMs... 
[1] "== ids_CAMs in drawnCAM"

DT::datatable(CAMwordlist_post, options = list(pageLength = 5))

save CAMs as .json files, and as .png (igraph)

save_CAMs_as_pictures = FALSE

if(save_CAMs_as_pictures){
  setwd("outputs")

setwd("savedCAMs_post")
setwd("png")
### remove all files if there are any
if(length(list.files()) >= 1){
  file.remove(list.files())
  cat('\n!
      all former .png files have been deleted')
}

### if no participant ID was provided replace by randomly generated CAM ID

if(all(CAMfiles_post[[3]]$participantCAM.x == "noID")){
  CAMfiles_post[[3]]$participantCAM.x <- CAMfiles_post[[3]]$CAM.x
}

### save as .json files, and as .png (igraph)
ids_CAMs <- unique(CAMfiles_post[[3]]$participantCAM.x); length(ids_CAMs)


for(i in 1:length(ids_CAMs)){
  save_graphic(filename = paste0("CAM_", i, "_t2")) #  paste0(ids_CAMs[i], "_t2"))
  CAM_igraph <- CAMdrawn_post[[c(1:length(CAMdrawn_post))[
    names(CAMdrawn_post) == paste0(unique(CAMfiles_post[[3]]$participantCAM.x)[i])]]]
  plot(CAM_igraph, edge.arrow.size = .7,
       layout=layout_nicely, vertex.frame.color="black", asp = .5, margin = -0.1,
       vertex.size = 10, vertex.label.cex = .9)
  dev.off()
}

setwd("../json")
### remove all files if there are any
if(length(list.files()) >= 1){
  file.remove(list.files())
  cat('\n!
      all former .json files have been deleted')
}
for(i in 1:length(raw_CAM_post)){
  if(!is_empty(raw_CAM_post[[i]]$nodes)){
    if(nrow(raw_CAM_post[[i]]$nodes) > 5){
      write(toJSON(raw_CAM_post[[i]], encoding = "UTF-8"),
            paste0(raw_CAM_post[[i]]$idCAM, ".json"))
    }
  }
}
}

analyze data

group comparisons

networkIndicators_pre$timepoint <- "pre"
networkIndicators_post$timepoint <- "post"

networkIndicators <- rbind(networkIndicators_pre, networkIndicators_post)


########################################
# show which robot was on average perceived more positive (overall data set)
########################################
summary(networkIndicators$mean_valence_macro[!is.na(networkIndicators$valence_micro_Rettungsroboter)])

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
-0.59091 -0.01923  0.22065  0.13307  0.38846  0.48387

summary(networkIndicators$mean_valence_macro[!is.na(networkIndicators$valence_micro_sozialerAssistenzroboter)])

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
-0.76923 -0.25417 -0.05757  0.02136  0.24451  1.05263

########################################
# post - pre difference of robot -> average valence
########################################
hist(networkIndicators_post$mean_valence_macro - networkIndicators_pre$mean_valence_macro)

summary(networkIndicators_post$mean_valence_macro - networkIndicators_pre$mean_valence_macro)

     Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 
-0.414438  0.001812  0.091209  0.110002  0.232504  0.502564

ggwithinstats(
  data = networkIndicators,
  x = timepoint,
  y = mean_valence_macro
)

########################################
# post - pre difference of robot -> average number of concepts
########################################
ggwithinstats(
  data = networkIndicators,
  x = timepoint,
  y = num_nodes_macro
)

open text answers

questionnaire_secondPostCAM$meanDifferencesCAMs <- round(x = networkIndicators_post$mean_valence_macro - networkIndicators_pre$mean_valence_macro, digits = 2)
DT::datatable(questionnaire_secondPostCAM[,c("meanDifferencesCAMs", "ans1", "feedback_critic")], options = list(pageLength = 5))