Data file (.csv) calculation problem

[ontinglo]

hi all

I want to calculate the average value of a column of data and save the average value in a new column in the .csv file. Like what is shown below:

Before

After

I don't know in where (now i write codes in jsPsych.init() and what codes should I add to achieve such purposes. What I have right now is like the following, apparently it does not work. could you please give me some advice?

thanks a lot!

 jsPsych.init({
        timeline: timeline,
        preload_images: [all_StimSet, all_StimSetT],
        show_preload_progress_bar: true,
        show_progress_bar: true,
        minimum_valid_rt: 300,
        default_iti: 50,

        on_data_update: function(data) {
            var TrustA_data = jsPsych.data.get().select('Trial_Trust_Value').mean();
            T_data = {
                meannn: TrustA_data,
            }
            jsPsych.data.write(T_data);
        },

        exclusions: {
            min_width: 800,
            min_height: 600,
        },
        on_finish: function() {
            jsPsych.data.get().localSave('csv', 'pilot_v5_mydata_Murk.csv');
        },

    })

[becky-gilbert]

Hi @ontinglo, I wouldn't do this in the on_data_update function, because that function runs every time information is saved to the jsPsych data, i.e. after each trial. Instead, I think you just need to do the mean calculation once, after all of the relevant trials have finished. Some options for this are in a call-function trial, in the on_timeline_finish function, which runs after a set of trials is finished, or at the very end of the experiment in the jsPsych.init on_finish function.

I also wouldn't suggest using the jsPsych.data.write function to save the mean to the data - instead you can use jsPsych.data.addProperties, which will add the mean to the data for all trials, or jsPsych.data.get().addToLast, which will add the mean to the last trial in the data collection.

Here are the steps for calculating the mean and saving it to the data:

1. Use the jsPsych data filter function to select the relevant trial data
2. Use the jsPsych data select function to select the relevant data column from those trials
3. Use the jsPsych data subset function to remove any missing data (because you can't calculate a mean when some of the values you're averaging over are null, empty, non-numeric etc.)
4. Use the jsPsych data mean function to get the mean of the remaining values
5. Save this mean value to the data using either jsPsych.data.addProperties or jsPsych.data.get().addToLast

Here's an example of averaging the RT values from responses to a series of images and saving the mean RT to the data.

var trial_procedure = {
  timeline: [fixation, trial],
  timeline_variables: trial_info,
  on_timeline_finish: function() {
    var all_rt_values = jsPsych.data.get().filter({trial_type:'image-keyboard-response'}).select('rt').subset(function(x){ return x !== null; });
    var rt_mean = all_rt_values.mean();
    jsPsych.data.get().addToLast({mean: rt_mean});
  }
};

In your case, I'm not sure that you can easily filter the trials with a trial trust value in your data. So I suggest adding a flag in the data for those trials, something like data: {task_part: 'trial_trust'}. Then you can do this somewhere in your experiment after all of the relevant trials are finished, e.g. in your experiment's on_finish function:

on_finish: function() {
  var all_trial_trust_values = jsPsych.data.get().filter({task_part: 'trial_trust'}).select('Trial_Trust_Value').subset(function(x){ return x !== null; });
  var trial_trust_mean = all_trial_trust_values .mean();
  jsPsych.data.get().addToLast({trial_trust_mean: trial_trust_mean});
  jsPsych.data.get().localSave('csv', 'pilot_v5_mydata_Murk.csv');
}

I hope this helps!