Show question in randomly 20% of all trials

[nora-ho]

Hi, I am currently setting up a picture naming experiment via jsPsych. Participants see 160 pictures in a random order and their task is to name them. Half of the pictures come from the related condition and the other half from the unrelated condition. Now I want to implement a question that appears randomly after 20% of the related pictures and randomly after 20% of the unrelated picture. The way I have done this so far is the following:

var voice_trial = {
type: "picnaming-voice-record-wav",
stimulus: jsPsych.timelineVariable('item'),
stimulus_duration: 2000,
trial_duration: rec_duration,
object: jsPsych.timelineVariable('object'),
stim_num: jsPsych.timelineVariable('stim_num'),
data: jsPsych.timelineVariable("data"),
on_finish: function(data) {
var context_related = jsPsych.randomization.sampleWithReplacement(["context", "nocontext"], 1, [2,8])[0]
var context_unrelated = jsPsych.randomization.sampleWithReplacement(["context", "nocontext"], 1, [2,8])[0]
if(jsPsych.timelineVariable('relatedness', true) == 'relatiert'){
data.context = context_related;
} else {
data.context = context_unrelated;
}
jsPsych.data.addDataToLastTrial(data);

This way, after each trial I assign the data of the timelineVariable that was shown in this trial a "context-label" that eiter says "context" (in 20% of all cases) or "nocontext" (in 80% of all cases). I do this separately for the related and the unrelated cases.

Later, I define the question and create the following timeline:
var context = {
timeline:[context_question],
conditional_function: function(){
// get the data from the previous trial,
// and check which option was selected
var data = jsPsych.data.get().last(1).values()[0];
if (data.context == "context"){
return true;
} else {
return false;
}
}
}
timeline.push([...], voice_trial, context, [...]);

By this, the question only appears if the data of the last trial before the current trial had been (randomly) tagged with "context", which should be the case for 20% of all trials.

However, by this I cannot make sure that each participant sees the question after exactly 20% of all trials. This is because for the first trial, the conditional function cannot draw on the context-label of a last trial (because there is no last trial before the first) and the last trial in the timeline does not have a next trial that will draw upon its context-label. Thus, it may happen, that some participants see the question after 31 trials, some after 32 (20%) and some after 33, which is not ideal.
So, my question is, whether there is a way to randomly assign each timeline Variable randomly with either 'context' (20% of related and 20% of unrelated) or 'nocontext' (20% of related and 20% of unrelated), e.g. at the very beginning of the experiment and later base the conditional function on the context-label of the current (instead of the last) trial. This would make sure, that the question is shown after exactly 20% of all trials. Or do you have any other ideas how I could achieve my goal?

Thanks in advance and for all the other super helpful discussions here!
Best,
Nora