R part V: analysis of experimental results

• Swirl
• DataCamp
• Statistics in R
• Tidyverse Resources
• R For Data Science

When troubleshooting errors, try googling your error message. Chances are, someone else has had the same error. Another great site for asking questions is https://stackoverflow.com!

• A few miscellaneous R commands
  • compiling data files
  • working with strings
  • saving ggplot2 objects
• Exploring, visualizing and analyzing the results of our stroop experiment
  • compiling individual data files
  • review of steps to summarize and organize data
  • summarizing and visualizing the data using dplyr & ggplot2
  • basic statistical analyses of our results

Last week, we created an experiment in Psychtoolbox, and this week, 4 of us did the experiment and generated a single .csv file with our own results. Today, we're going to analyze that data in R. The first thing we will need to do is compile the data from all 4 files into a single dataframe.

One function that will be useful is list.files(), which lists all files in a given directory matching a specified pattern:

list.files('./data', pattern = '*.csv')

[1] "1_stroop.csv" "2_stroop.csv" "3_stroop.csv" "4_stroop.csv"

We can then use a for-loop to loop over these files (see next slide).

We used for loops extensively in MATLAB, and although I tend to try to minimize my use of for-loops in R, sometimes they are inevitable. For instance, today we will loop over a list of files, and on each iteration of the loop, load one data file. See an example below:

library(data.table)
myFiles <- list.files('./data', pattern = '*.csv')
for (ii in 1:length(myFiles)) {
  this_data_frame <- fread(myFiles[ii])
}

In fact, we will create a loop that not only loads a dataframe on each iteration but also adds that dataframe to a composite dataframe that will store the data for all subjects (e.g. from all iterations).

There are two useful commands if you want to literally bind dataframes together (this is a distinct use case from dplyrs left_join and related join functions):

rbind() binds two dataframes by stacking rows on top of one another

cbind() binds two dataframes by stacking columns next to one another

Sometimes, you'll have string variables that need to be dealt with in more complex ways. For instance, you might want to extract a variable based on the name of a file (which is what we will do today).

We created files such as 1_stroop.csv. Today we will want to extract the subject ID from the filename and store it as a variable in the dataframe that is loaded. So, how can we extract the ID from the filename? The stringr package is useful for this instance (and all kinds of situations when you need to manipulate strings).

Specifically, we will use the str_split() function within the stringr package to split a string at a specific place:

filename <- '1_stroop.csv'
str_split(filename, '_')

[[1]]
[1] "1"          "stroop.csv"

The output is a list, and we can index within the first element ([[1]]) to extract our string of interest:

filename <- '1_stroop.csv'
split <- str_split(filename, '_')
split[[1]][1]

[1] "1"

Download the zipped directory of files for this week. It contains the data files to compile.

Open RStudio, and open the script for today's tutorial.