iPad-data-cleaning

Script to munge ipad data for soc-xsit project

Load libraries for data manipulation

library(plyr)
library(dplyr)
library(directlabels)
library(stringr)

Read in the data, which are stored in separate .txt files for each participant.

# Create empty arrays for binding 
all.data <- as.data.frame(matrix(ncol = 0, nrow = 0))

#grab all file names from data dir. 
#These are all of the kids for whom we have ipad data
all_results <- list.files(path = "/Users/kmacdonald/Documents/Projects/SOC_XSIT/raw_data/child/",
                          pattern = 'results_*', all.files = FALSE)

#function to munge the data
#takes in a list of filenames (.txt of each kid's data), 
#strips html escape characters, and grabs the relevant info for each trial
#returns data frame with all the data
bing_clean <- function(filename) {
        x <- readLines(paste("/Users/kmacdonald/Documents/Projects/SOC_XSIT/raw_data/child/", 
                                filename,sep=""),warn=FALSE)
        x <- unlist(strsplit(x,'<li>'))
        
        #grab condition 
        condition <- x[3]

        if (grepl(pattern="Social", condition)) {
                condition <- str_extract(pattern="Social", condition)
        } else {
                condition <- str_extract(pattern="No-social", condition)
        }
                
        # grab trial information 
        x <- x[11:150]
        
        #strip html characters
        x <- gsub('\\"',"", x)
        x <- gsub("\\\\\"","",x)
        x <- gsub("</li>","",x)
        x <- gsub("</ul>\\},\\{<ul>","",x)
        x[length(x)] <- gsub("</ul>\\}]","",x[length(x)])
        x <- gsub("^.*?: ","",x)
        
        #grab date 
        date <- str_sub(filename,20,36)
        date <- gsub('_','',date)
        date <- gsub('\\.','',date)
        #grab id
        id <- str_sub(filename,0,2)
        id <- gsub('-','',id)
        #bind to data frame
        x <- as.data.frame(matrix(x,14,10,byrow=TRUE), stringsAsFactors=F)
        x$V11 <- date
        x$V12 <- id
        x$V13 <- condition
        return(x) ## need this return to get the data frame!
}

#apply munging function to each kid's data file
all.data <- ldply(
                .data = all_results,
                .fun = bing_clean
                )

#add variable names to columns
names(all.data) <- c("itemNum","trialType","samePos","chosen","chosen_idx",
              "kept","kept_idx","rt","faceVid","faceIdx", "date", "id", "condition")

Tag the different trial types:
* example * exposure * test

Arrange the data file with sensible column order.

#dplyr syntax
all.data <- all.data %>% 
                group_by(date) %>%
                mutate(trial = c(1:14)) %>%
                mutate(example_trial = ifelse(trial %in% c(1:2),1,0),
                       exposure_trial = ifelse(trial %in% c(3,5,7,9,11,13),1,0),
                       test_trial = ifelse(trial %in% c(4,6,8,10,12,14),1,0)) %>%
                arrange(date, trial)

Clean up the data: * figure out if child answered correctly on that trial * relabel variables and change variable types for analysis

#find whether child's choice was correct 
#for exposure trials, we check the index of eye gaze against child's choice
#for test trials, we check the kept image against the child's choice 
all.data <- all.data %>% 
                group_by(id) %>%
                mutate(correct = ifelse(exposure_trial == 1, chosen_idx == faceIdx, 
                                 ifelse(example_trial == 1, chosen[1] == "flower" & chosen[2] == "truck",
                                        chosen == kept))) 

#relabel variables and variable types for analysis
all.data$rt <- as.integer(all.data$rt)
all.data$condition <- as.factor(all.data$condition)

Add demographic data for each child (age and gender).

#read in demo csv
demo <- read.csv("/Users/kmacdonald/Documents/Projects/SOC_XSIT/raw_data/child/soc-xsit-ipad-demo.csv")

all.data <- join(all.data,demo)

## Joining by: id

#reorder columns to put id and demo variables first
all.data <- all.data[c("id","birthday","gender","age","age.group","date", "condition", 
                   "trial","itemNum","trialType","samePos","chosen","chosen_idx","kept",
                   "kept_idx","rt","faceVid","faceIdx","example_trial",
                   "exposure_trial","test_trial", "correct")]

Fix trial type labels -> not tracked correctly during experiment.

exposure.trials <- all.data %>%
                filter(exposure_trial == 1) %>%
                mutate(trial_type_redo = ifelse(chosen == kept,1,2)) %>%
                select(id, itemNum, trial_type_redo, chosen_exposure = chosen, 
                       kept_exposure = kept, correct_exposure = correct)

test.trials <- all.data %>%
                filter(test_trial == 1) 

test.trials <- join(test.trials,exposure.trials, by=c("id","itemNum"))

test.trials$trial_type_redo <- factor(test.trials$trial_type_redo, 
                                      labels = c('Same','Switch'))

Save tidy data files.

write.csv(all.data, file="/Users/kmacdonald/Documents/Projects/SOC_XSIT/processed_data/child/soc-xsit-ipad-alldata.csv")
write.csv(test.trials, file="/Users/kmacdonald/Documents/Projects/SOC_XSIT/processed_data/child/soc-xsit-ipad-testdata.csv")