tricky randomization and block breaking

[viviantic]

Hello everyone! I have a question about randomization and block breaking.

In my experiment, I have 72 sets of sentences as critical items and 120 filler items. For each set of the critical sentences, there are 6 conditions (i.e., 1a, 1b, 1c, 1d, 1e, 1f for set #1, and 2a, 2b, 2c, 2d, 2e, 2f for set #2,... all the way to 72a, 72b,... for set#72). Fillers are just 120 random sentences, so no condition differences for fillers.

I'd like to counterbalance these 72 sets of critical sentences among my participants in 6 experimental lists that contain only one version of each sentence, so each experimental list will contain 192 sentences (72 critical + 120 fillers) and will be divided into 3 blocks (64 sentences per block). Right now I am kinda doing it manually by myself, but I was just wondering if there are some plugins that can save me a lot of time and trouble.

I am stuck when I tried to write a randomization command and a block-breaking command that might work for this tricky situation. Any suggestions on how I can approach this?

Thanks!

[becky-gilbert]

Hi @viviantic! Sorry for the delay in responding to your question. Would you still like some help with this? If so could you share the way that you've structured your trial data? E.g. something like this:

var trial_info = [
  {condition: '1a', set: '1', type: 'critical', filler_number: 'NA'},
  {condition: '1b', set: '1', type: 'critical', filler_number: 'NA'},
  ...
  {condition: '2a', set: '2', type: 'critical', filler_number: 'NA'},
  {condition: '2b', set: '2', type: 'critical', filler_number: 'NA'},
  ...
  {condition: 'NA', set: 'NA', type: 'filler', filler_number: '1'},
  {condition: 'NA', set: 'NA', type: 'filler', filler_number: '2'}
];

Of course it doesn't need to be set up this way - this is just an example.

And could you also explain how you're setting up your experiment, e.g. using timeline variables or a loop, or share your code? Thanks!

[viviantic]

Hello @becky-gilbert! Thanks for your reply.

I started with copying your code. It did assign a version number (e.g. 1-6) and a condition letter (e.g., a-f) to all the stimuli, which was super cool. However, instead of randomly selecting one version (e.g., version 5) and going through all subset items in that version (e.g., 72 sets of critical sentences, only 1 condition per set, like 1a for set 1, 2b for set 2, etc.), the script still ran through every single stimulus under the variable (var critical_stimuli) regardless of what version the stimulus belonged to. (So it won't stop until it ran through all conditions of the 72 sets of critical sentences).

Is there a way to only show one condition per set for each version (i.g., set 1 condition a in version 1, and set 1 condition b in version 2, ...set 1 condition f in version 6, etc.)? So if a participant gets version 2, he/she will only see 1b (and never 1a, 1c to 1f) for set 1?

Something like this:
version 1:
1a, 2b, 3c, 4d, 5e, 6f, 7a, 8b, 9c, 10d, 11e, 12f, ...72f (critical) + 120 fillers, which will then be randomized and divided into 3 blocks (64 sentences each).

version 2:
1b, 2c, 3d, 4e, 5f, 6a, 7b, 8c, 9d, 10e, 11f, 12a,....72a(critical) + 120fillers, randomized and then divided into 3 blocks

... and 1c for version 3, 1d for version 4, 1e for version 5, and then 1f for version 6. (with randomization and block-breaking).

[becky-gilbert]

Hi @viviantic, the code that I wrote to create the stimuli versions assumes that the trials are listed in a particular order in the critical_stimuli array: 1a, 1b, 1c, 1d, 1e, 1f, 2a, 2b, 2c, 2d, 2e, 2f, ... 72a, 72b, 72c, 72d, 72e, 72f (same as in this comment). Is that the same as the order you're using?

When I test the code with this array:

    var critical_stimuli= [
        {words: ['+','This','is','critical','stimuli','1','a.'], question: "question for critical stimuli 1a here", correct_response: "j"}, 
        {words: ['+','This','is','critical','stimuli','1','b.'], question: "question for critical stimuli 1b here", correct_response: "f"}, 
        {words: ['+','This','is','critical','stimuli','1','c.'], question: "question for critical stimuli 1c here", correct_response: "j"}, 
        {words: ['+','This','is','critical','stimuli','1','d.'], question: "question for critical stimuli 1d here", correct_response: "f"},
        {words: ['+','This','is','critical','stimuli','1','e.'], question: "question for critical stimuli 1e here", correct_response: "j"}, 
        {words: ['+','This','is','critical','stimuli','1','f.'], question: "question for critical stimuli 1f here", correct_response: "f"}, 
        {words: ['+','This','is','critical','stimuli','2','a.'], question: "question for critical stimuli 2a here", correct_response: "j"}, 
        {words: ['+','This','is','critical','stimuli','2','b.'], question: "question for critical stimuli 2b here", correct_response: "f"}, 
        {words: ['+','This','is','critical','stimuli','2','c.'], question: "question for critical stimuli 2c here", correct_response: "j"}, 
        {words: ['+','This','is','critical','stimuli','2','d.'], question: "question for critical stimuli 2d here", correct_response: "f"},
        {words: ['+','This','is','critical','stimuli','2','e.'], question: "question for critical stimuli 2e here", correct_response: "j"}, 
        {words: ['+','This','is','critical','stimuli','2','f.'], question: "question for critical stimuli 2f here", correct_response: "f"}, 
        // all the way to 72a, 72b,...for set 72.  
    ];

Here's what I get in the stimuli_subset array:
version 1: 1a, 2b
version 2: 1b, 2c
version 3: 1c, 2d
version 4: 1d, 2e
version 5: 1e, 2f
version 6: 1f, 2a
Is that what you're going for? Sorry if I've misunderstood...

[viviantic]

Hello @becky-gilbert! Yes, I use the same order. The assignment seems fine (so 1a, 2b are in version 1, and 1b, 2c in version 2, etc.), but when I run the script, it runs all the critical sentences (1a, 1b, 1c, 1d, 1e, 1f for set 1 and all conditions for all 72 sets) instead of just running the sublet of the critical stimuli (e.g., just 1a, then 2b...).

Here is a simplified version revised from my code. Could you run it and see if you get the same result? When I run it, I can see all 6 conditions of set 1 (i.e., 1a, 1b, 1c,...1f) and all 6 conditions of set 2. I was hoping that this script will run through only 5 sentences (1a, 2b, and 3 fillers if version 1 is selected) but not all (1a to 1f, 2a to 2f, 3 fillers).

`

<title>Self Paced Reading Experiment </title> <script src="jspsych/jspsych.js"></script> <script src="jspsych/plugins/jspsych-html-button-response.js"></script> <script src="jspsych/plugins/jspsych-moving-window-button-phrase.js"></script> <script src="jspsych/plugins/jspsych-html-keyboard-response.js"></script> <script src="jspsych/plugins/jspsych-instructions.js"></script> <script src="jspsych/plugins/jspsych-image-keyboard-response.js"></script>

<script>
    
// create timeline
var timeline = [];
    
    
// create fixation 
var fixation = {
    type: 'html-keyboard-response',
    stimulus: '<span style="font-size:40px;">+</span>',
    choices: jsPsych.NO_KEYS,
    trial_duration: 2000,
    data: {
            trial_part: 'fixation', 
            task_part: jsPsych.timelineVariable('task_part')}
    };


//create feedback
var feedback = {
    type: 'html-keyboard-response',
    stimulus: function() {
        var feedback_text = '<span style="color:red;font-size:30px;">Incorrect！</span>';
        var last_resp_acc = jsPsych.data.getLastTrialData().values()[0].accuracy;
        if (last_resp_acc == true) {
            feedback_text = '<span style="color:green;font-size:30px;">Correct！</span>'
        }
        return feedback_text;
    },
    choices: jsPsych.NO_KEYS,
    trial_duration: 2000
    };

// create critical_stimuli
var critical_stimuli= [
    {words: ['+','This','is','critical','stimuli','1','a.'], question: "question for critical stimuli 1a here", correct_response: "j"}, 
    {words: ['+','This','is','critical','stimuli','1','b.'], question: "question for critical stimuli 1b here", correct_response: "f"}, 
    {words: ['+','This','is','critical','stimuli','1','c.'], question: "question for critical stimuli 1c here", correct_response: "j"}, 
    {words: ['+','This','is','critical','stimuli','1','d.'], question: "question for critical stimuli 1d here", correct_response: "f"},
    {words: ['+','This','is','critical','stimuli','1','e.'], question: "question for critical stimuli 1e here", correct_response: "j"}, 
    {words: ['+','This','is','critical','stimuli','1','f.'], question: "question for critical stimuli 1f here", correct_response: "f"}, 
    {words: ['+','This','is','critical','stimuli','2','a.'], question: "question for critical stimuli 2a here", correct_response: "j"}, 
    {words: ['+','This','is','critical','stimuli','2','b.'], question: "question for critical stimuli 2b here", correct_response: "f"}, 
    {words: ['+','This','is','critical','stimuli','2','c.'], question: "question for critical stimuli 2c here", correct_response: "j"}, 
    {words: ['+','This','is','critical','stimuli','2','d.'], question: "question for critical stimuli 2d here", correct_response: "f"},
    {words: ['+','This','is','critical','stimuli','2','e.'], question: "question for critical stimuli 2e here", correct_response: "j"}, 
    {words: ['+','This','is','critical','stimuli','2','f.'], question: "question for critical stimuli 2f here", correct_response: "f"}, 
    // all the way to 72a, 72b,...for set 72.
];

//create filler_stimuli
var filler_stimuli=[
    {words:['+','this','is','filler','1'],question:"filler 1 question here",correct_response:"f",filler_number: '1'},
    
    {words:['+','this','is','filler','2'],question:"filler 2 question here",correct_response:"f",filler_number: '2'},
    
    {words:['+','this','is','filler','3'],question:"filler 3 question here",correct_response:"f",filler_number: '3'},
    
];

//randomization
var version = jsPsych.randomization.sampleWithoutReplacement([1,2,3,4,5,6],1)[0];
var n_conditions = 6; // a, b, c, d, e, f
var stimuli_subset = [];
var index = version-1; // if version = 1 then the first trial from the      critical_stimuli is at index 0, for version 2 it's index 1, etc.
var current_condition = version-1; // this will go from 0 to 5, incrementing for each condition a-f, then start over at 0
    for (var i = 0; i < critical_stimuli.length; i++) {
        if (i == index) {
        stimuli_subset.push(critical_stimuli[i])
        current_condition++;
            if (current_condition == n_conditions) {
            // once we get to the last condition ("f"), the next critical trial will be the first trial in the next set, so increment by 1
            index++;
            // we need to go back to the first condition ("a")
            current_condition = 0;
            } else {
            // if this isn't the last condition ("f"), then we add n_conditions to get to the next set, then 1 to get to the next condition
            index = index + n_conditions + 1;
            }
        }     
    }
console.log('stimuli subset: ', stimuli_subset);
    
// combine critical and filler stimulous arrays into a single array to use for timeline variables
var main_stimuli = critical_stimuli.concat(filler_stimuli);
console.log(main_stimuli);   
    
// create main_task
var main_task = {
	type: "moving-window-button-phrase",
    words: jsPsych.timelineVariable("words"),
	data: {
        trial_part: "main_task",
        correct_response: jsPsych.timelineVariable("correct_response"),
        stimulus: jsPsych.timelineVariable("stimulus"),
        question: jsPsych.timelineVariable("question")
    },
};

// create practice_question
var main_question = {
	type: "html-keyboard-response",
	stimulus: function() {
            return '<span style="font-size:30px;">' + jsPsych.timelineVariable('question', true) + '</span>';
        },
	choices: ["f", "j"],
	data: {
        trial_part: "main_question",
        task_part: jsPsych.timelineVariable('task_part'),
        correct_response: jsPsych.timelineVariable("correct_response"),
        stimulus: jsPsych.timelineVariable("stimulus"),
        question: jsPsych.timelineVariable("question")
    },
	prompt: '<p>F=False &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; J=True</p>',
    post_trial_gap: 500,
    on_finish: function(data) {
            var acc = false;
            if (data.correct_response == jsPsych.pluginAPI.convertKeyCodeToKeyCharacter(data.key_press)) {
                acc = true;
            }
            data.accuracy = acc;
        }
        };
    
var main_procedure = {
    timeline: [fixation, main_task, main_question, feedback],
    timeline_variables: main_stimuli,
    randomize_order: true
};
timeline.push(main_procedure);
    
// run experiment
jsPsych.init({
	timeline: timeline,
	on_finish: function(){
		jsPsych.data.displayData();
	}
});

</script>

`

Please forgive me if I fail to explain the issue clearly. Thank you for helping me out!

[becky-gilbert]

Hi @viviantic, the problem is that you're still using the critical_stimuli array when you create the main_stimuli array that's used as the timeline variables in your main_procedure. Instead you want to use the stimuli_subset array. So try changing this:

var main_stimuli = critical_stimuli.concat(filler_stimuli);

to this:

var main_stimuli = stimuli_subset.concat(filler_stimuli);

[viviantic]

Hello @becky-gilbert! Yes this worked perfectly! Thank you so much for your help and for making it a great learning experience for a novice coder like me.

I have one last question. I need to divide my experiment into 3 blocks separated by 2 breaks. So for each version (i.e., version 2, if selected), which contains 192 sentences (72 critical + 120 fillers, randomized), I'd like to divide the presentation of the stimuli into 3 big blocks (64 sentences per block). These blocks are separated by a 1-minute countdown page, serving as a very short break.

I tried to code it myself using what I have learned, which is why it took me a few days to reply. Here's my code (This might be way off, so a little disclaimer here. Serious headache might occur):

// skip randomization codes in the previous comment above

// combine critical and filler stimulous arrays into a single array to use for timeline variables
var main_stimuli = stimuli_subset.concat(filler_stimuli);
console.log(main_stimuli);   

//create timedbreak(a page showing a 1-minute countdown timer in the middle) <-- I got stuck here
var timedbreak
    
//create blocks (to run 64 sentences, then take 1-minute break, then another 64 sentences, then break, then the rest of the sentences)
var block = {
    timeline: [main_stimuli[1:64], timedbreak, main_stimuli[65:128], timedbreak, main_stimuli[129:192]],
    timeline_variables: main_stimuli,
    randomize_order: false //because we've already set up the randomization command, I suppose?
};

//create main_task (skip for now)
//create main_question (skip for now)

var main_procedure = {
        timeline: [fixation, main_task, main_question, feedback],
        timeline_variables: main_stimuli,
        randomize_order: true
    };
    timeline.push(main_procedure);
//I got stuck here too because I don't know how to incorporate the ```var block``` into ```var main_procedure```

so I guess a summary of my questions is:

1. how to create a variable var timedbreak that shows a page with a 1-minute countdown timer in the middle,
2. does my var block look ok?
3. how to incorporate var block , if it was written correctly, to my var main_procedure for it to run?

Any suggestions on how to tackle these problems would be greatly appreciated.
Thank you.

[becky-gilbert]

Hi @viviantic, I think your general approach to this makes sense. One problem with the way you've set this up is that, in your block procedure, the timeline should be the events that you want to happen during the block, where each event is an object like a trial or trial procedure. Whereas it looks like you're using the stimuli array in the timeline parameter. See the timeline variables docs for a refresher on how this works.

The way I might do this is (1) randomize the main_stimuli array using jsPsych.randomization.shuffle, (2) set up the separate trial blocks by separating the randomized main_stimuli array into chunks, using those chunks as the timeline_variables array for each block, and (3) inserting the break trial onto your timeline, in between each trial block. Note that this method does not guarantee that you have the same numbers of your different trial types (conditions/fillers) in each block.

For your break with the countdown timer, you could use an html-keyboard-response trial, and use the trial's on_load function to store the current time remaining and update the text with that value after every second using JavaScript's setInterval function. You'd need to remember to clear the interval once the time is up.

Here's how it might look:

var main_stimuli = stimuli_subset.concat(filler_stimuli);
var main_stimuli_rand = jsPsych.randomization.shuffle(main_stimuli);

// set up the break screen 
var timedbreak = {
  type: 'html-keyboard-response',
  stimulus: '<div><p>Time for a 60 second break!</p><p>The task will move on when the time below reaches 0.</p><div id="timer">60 seconds left</div>',
  trial_duration: 61000,
  on_load: function() {
    var sec_left = 60;
    var timer_interval = setInterval(function() {
      sec_left--;
      document.getElementById("timer").innerHTML = sec_left.toString() + " seconds left";
      if (sec_left == 0) {
        clearInterval(timer_interval);
      }
    }, 1000);
  }
};

// define all the trial objects that you'll use in the blocks: fixation, main task, main question, feedback 
// using jsPsych.timelineVariable() for any parameters that come from the main_stimuli array

// set up block 1 and add it to the main timeline
var block1 = {
  timeline: [fixation, main_task, main_question, feedback],
  timeline_variables: main_stimuli_rand.slice(0,64); // first chunk of your randomized trial array - remember that indexing starts at 0
};
timeline.push(block1);

// add a break to the main timeline 
timeline.push(timedbreak);

// set up block 2 with the next chunk of main_stimuli_rand as the timeline_variables array and add this block to the main timeline
// then add another break to the main timeline
// etc.

[viviantic]

Hello @becky-gilbert!
Thank you so much for the solution. I tried it, and it worked perfectly.
Thank you for helping me and for being incredibly patient along the way. I've learned a lot for sure!!!