README.md

In martin-vasilev/readingET: Automatic pre-processing of eye-movement reading data

This is an R package that performs automatic pre-processing and mapping of fixations to text stimuli on the screen. The current implementation of this package only works with reading experiments recorded with the Eyetrack software (http://blogs.umass.edu/eyelab/software/). However, support for other types of software packages and experimental scripts will be added very soon. The goal is to create a general and easy-to-use tool that can perform data pre-processing with just a few commands by using the raw .asc data. For users unfamiliar with R, a web interface with Rshiny will also be created. Currently, most of the development is focused on working with single line reading studies. However, support for multiple-line reading data is also on the way.

I'm doing my best to debug the script and to test it rigouously. Nevertheless, bugs can always slip through. If you notice a problem or want any new functionality to be added, please email me at [e-mail] or open an issue in the repository. Also, if the package doesn't work with your software/ equipment, please let me know and I would do my best to add support for it.

Installation:

To install the package, please run the following code in R:

if('devtools' %in% rownames(installed.packages())==FALSE){
    install.packages('devtools')
    library(devtools)
  }else{
    library(devtools)
  }
install_github('martin-vasilev/EMreading')

If the above doesn't work, try this:

if('remotes' %in% rownames(installed.packages())==FALSE){
    install.packages('remotes')
    library(remotes)
}else{
    library(remotes)
}
install_url(url="https://github.com/martin-vasilev/EMreading/archive/master.zip", INSTALL_opt= "--no-multiarch")

Current functionality:

Pre-processing of single-line reading experiments recorded with Eyetrack

NEW! You can now also pre-process your data online without using any commands! Visit: https://mvasilev.shinyapps.io/shinyapp/ If you prefer to run the graphical interface locally (this is faster since you don't upload any files), you can run the GUI() function from the package.

To pre-process the data, simply use the SingleLine() function of the package. You will need to provide some basic information, such as directory containing the data files and some details about the experiment: e.g.,

# preprocess data
data<- SingleLine(data_list= "C:/Users/Martin Vasilev/My Data", ResX= 1920, ResY=1080, maxtrial= 120)

# save raw data so that you don't have to re-do this later on:
save(data, file = "data.Rda")

This will give you a data frame containing all fixations in addition to most variables that you will need for later analysis. For a full description of all output variables, see here .

To perform a complete clean-up of the data, you can use the cleanData() function:

dataN<- cleanData(data)

This performs a complete clean-up of the raw data that is standardly done in the field of eye-movements during reading. If you don't specify any parameters, it will do the default (conventionally done) clean-up: removal of fixations outside the screen or text area, blink removal, combining of short fixations (< 80ms) that occur within one letter of another fixation, removal of any remaining fixations < 80ms, removal of outlier fixations (> 800ms). All of these paramaters can be turned on or off, and the specific cut-off values can also be modified. Additionally, the script also supports removing outliers via the std method (i.e., removing all fixations that are x standard deviations above the subject's mean). The function also reports a summary containing the percentage of fixations removed from each category for easy reporting.

Next, if you are planning to do word-level analyses, you can calculate the standard fixation durations measures (FFD, SFD, GD, TVT) using the cleaned-up data. This is done with the wordMeasures() function. Simply provide the data frame containing the cleaned-up fixation data:

FD<- wordMeasures(dataN)

martin-vasilev/readingET documentation built on Jan. 31, 2023, 3:38 p.m.