Double page-advance (page skipping) after New Survey-Taking Experience — jQuery + header code conflict?

Hi all,

I run memory-research surveys with heavy custom JavaScript, both in question-level code and in the survey header. After the New Survey-Taking Experience was enforced, they broke, and I've been fighting a navigation problem I can't fully explain. Hoping someone here has seen this.

Background

My question-level code advances pages with jQuery, e.g.:

javascript

jQuery('input[type="radio"]').on('click', function(){ jQuery('#NextButton').click(); });

After the update, this silently stopped working for many respondents — my hypothesis is that jQuery is no longer preloaded. To compensate, I added header JavaScript that advances 1-radio "continue" pages itself. This is the current header:

javascript

Qualtrics.SurveyEngine.addOnload(function() {
    var radios = document.querySelectorAll('input[type="radio"]');
    if (radios.length === 2) return;                 // coding item — keyboard only, leave alone
    if (window.Event && Event.stopObserving) {
        Event.stopObserving(document, 'keydown');    // clear a keydown listener that survives between pages
    }
});

Qualtrics.SurveyEngine.addOnReady(function() {
    var that = this;
    var radios = document.querySelectorAll('input[type="radio"]');
    if (radios.length !== 1) return;                 // only "continue" pages (single radio)
    if (radios[0].dataset.bound) return;
    radios[0].dataset.bound = '1';
    var r = radios[0];
    function handler(e){
        if(!document.contains(r)){ document.removeEventListener('click',handler,true); return; }
        if(e.target.closest('input[type="radio"], label, td, .q-radio')){
            document.removeEventListener('click',handler,true);
            r.checked = true;
            try { that.setChoiceValue(parseInt(r.value, 10), true); } catch(err){}
            requestAnimationFrame(function(){
                requestAnimationFrame(function(){ that.clickNextButton(); });
            });
        }
    }
    document.addEventListener('click', handler, true);
});

The problem

For a portion of respondents, pages now advance twice on a single click — a page flashes and is skipped automatically. It corrupts data, including on critical response-time pages. My working theory: for respondents where jQuery IS still active, both the header code above and the legacy question code fire on one click → double advance.

What I can't explain

I compared affected and unaffected survey versions directly — header, question code, and page structure are all identical — yet skip rates differ dramatically: from ~0.1% of pages up to ~36% of pages, measured from exported Page Submit timings (near-zero = skipped). Same code, wildly different skip rates. This is the part that baffles me.

My questions

1. Has anyone confirmed under exactly what conditions jQuery is / isn't available in the new experience? Could it vary per respondent (e.g. gradual rollout, browser, cache)?
2. Is the "header advances + legacy code advances" double-fire the likely cause, and what's the cleanest fix — remove the header advance and rely on local code, or load jQuery explicitly and remove the header advance?
3. Any idea what could make identical survey versions skip at 0.1% vs 36%?

I can share more code and the timing data. Any insight appreciated — thanks!