Getting debrief block to analyze only the certain responses

[rbruno1]

Hi. So my experiment is asking people to answer trivia questions, and then they report their confidence in their answers.
I want something like this to appear at the end of the experiment:

You were right 33% of the time when you felt 55% confident in your answers
You were right 48% of the time when you felt 65% confident in your answers
You were right 82% of the time when you felt 75% confident in your answers
You were right 58% of the time when you felt 85% confident in your answers
You were right 80% of the time when you felt 95% confident in your answers

Here is the code. I need some way for it to filter and analyze only responses that the participant reported __% confidence in:

[var debrief_block = {
  type: jsPsychHtmlKeyboardResponse,
  stimulus: function() {
    var trials_55 = jsPsych.data.get().filter({task: "confidence"},{response: 1});
    var trials_65 = jsPsych.data.get().filter({task: "confidence"},{response: 2});
    var trials_75 = jsPsych.data.get().filter({task: "confidence"},{response: 3});
    var trials_85 = jsPsych.data.get().filter({task: "confidence"},{response: 4});
    var trials_95 = jsPsych.data.get().filter({task: "confidence"},{response: 5});
    var correct_trials_55 = trials_55.filter({task: 'response'},{correct: true});
    var correct_trials_65 = trials_65.filter({task: 'response'},{correct: true});
    var correct_trials_75 = trials_75.filter({task: 'response'},{correct: true});
    var correct_trials_85 = trials_85.filter({task: 'response'},{correct: true});
    var correct_trials_95 = trials_95.filter({task: 'response'},{correct: true});
    var accuracy_55 = Math.round(correct_trials_55.count() / trials_55.count() * 100);
    var accuracy_65 = Math.round(correct_trials_65.count() / trials_65.count() * 100);
    var accuracy_75 = Math.round(correct_trials_75.count() / trials_75.count() * 100);
    var accuracy_85 = Math.round(correct_trials_85.count() / trials_85.count() * 100);
    var accuracy_95 = Math.round(correct_trials_95.count() / trials_95.count() * 100);
    return `<p>You are right ${accuracy_55}% of the time when you are 55% sure of your answer.</p>
            <p>You are right ${accuracy_65}% of the time when you are 65% sure of your answer.</p>
            <p>You are right ${accuracy_75}% of the time when you are 75% sure of your answer.</p>
            <p>You are right ${accuracy_85}% of the time when you are 85% sure of your answer.</p>
            <p>You are right ${accuracy_95}% of the time when you are 95% sure of your answer.</p>
      <p>Press any key to complete the experiment. Thank you!</p>`;
  }
};
timeline.push(debrief_block);

and here is the link to the experiment : https://alskuvye1v.cognition.run/

[becky-gilbert]

Hi @rbruno1, to do this, it helps to have some way of linking the response that you're scoring (the trivia question answer) with the confidence rating, since these two responses are stored as separate trials. What I would do is edit your confidence rating trial so that it also stores the response accuracy from the previous trial. This will make it easier to get accuracy summaries for the different confidence ratings. I think one way to do this is by making the data parameter for the confidence rating trial into a function:

var AnimalsTrials = {
  timeline: [
    {
      type: jsPsychHtmlButtonResponse,
      stimulus: jsPsych.timelineVariable('stimulus'),
      choices: ['True', 'False'],
      data: {
        task: 'response',
        correct_response: jsPsych.timelineVariable('correct_response')
      },
      on_finish: function(data){
        //data.correct = jsPsych.pluginAPI.compareKeys(data.response, data.correct_response);  
        // NOTE: compareKeys is just for scoring keyboard responses
        data.correct = data.response == data.correct_response;
      }
    },
    { 
      type: jsPsychHtmlButtonResponse,
      stimulus: ["How confident are you in your answer?"],
      choices: ["55%", "65%", "75%", "85%", "95%"],
      data: function() {
        // get the response accuracy from the last trial and store it in the data for this trial
        var correct = jsPsych.data.getLastTrialData().values()[0].correct;
        return {task: 'confidence', correct: correct};
      },
    }
  ]
};

Then, in your debrief block, you can filter the confidence rating trials based on the responses, and get the previous trial's response accuracy from that same data:

var debrief_block = {
  type: jsPsychHtmlKeyboardResponse,
  stimulus: function() {
    var trials_55 = jsPsych.data.get().filter({task: "confidence", response: 1});
    // ...
    // 'correct' is the response accuracy from the previous trial (trivia question)
    var correct_trials_55 = trials_55.filter({correct: true}); 
    // ...
  }
};

Does that solution work for you?

[rbruno1]

Thank you so much! Here is the final link: https://ji8wsy8tcu.cognition.run/, in case you wanted to play. It is a trivia game that, upon finishing, tells you if you are overconfident in your guessing. Thanks either way!