Reaction Time Survey Questions

Reaction Time measures how quickly a respondent reacts to a stimulus — a word, phrase, or image — instead of just what they choose. Domandata timestamps the moment a stimulus appears and the moment the respondent answers using performance.now(), the same high-resolution clock browsers use for animation timing, rather than the wall-clock timestamps most survey timing features rely on. It is well suited to lexical decision and categorization tasks, implicit-attitude style paradigms, simple detection tasks, and any study where latency itself is the dependent variable, not just the response.

One question runs a full sequence of trials — stimulus, timed response, repeat — the same way a MaxDiff question runs multiple trials internally. Reaction Time requires Standard plan or above to appear in the question type picker.

• Trials: each trial is a text or image stimulus. Add an optional category label (for example, "congruent" / "incongruent") to group trials during analysis — respondents never see it.
• Correct response: mark a trial's correct response to get per-trial accuracy scoring alongside timing. Leave it unset for tasks with no correct answer, such as simple detection.
• Practice trials: mark a trial as practice to still show and time it for the respondent, but keep it out of the speeding check, Instrument Review's category-count guidance, and every summary/export statistic — useful for a warm-up phase before the trials that actually count.
• Blocks (optional): group trials into named phases — for example, an IAT-style pairing block — presented in sequence with a short pause between them. Leave this empty for one continuous sequence; it's only worth setting up for multi-phase designs.
• Counterbalance block order: with 2+ blocks, half of respondents (split deterministically, not by chance per page load) see blocks in authored order and half see them reversed, so the effect you're measuring isn't confounded with which pairing a respondent met first — standard practice in IAT-style designs.
• Response mode: respondents can answer by keypress (map keys like F and J to response labels), on-screen click, or both. On a phone or tablet with no keyboard, a keypress-only question shows on-screen response buttons instead, so touch respondents can still finish; those trials record their input method as "click".
• Error feedback: optionally flash a brief incorrect indicator on a trial with a correct response configured — the classic IAT "red X" feedback loop — and optionally require the correct response before the trial advances, instead of the indicator being purely informational.
• Fixation and inter-trial timing: configure the fixation-cross duration shown before each stimulus and the blank-screen gap between trials.
• Timeout: optionally cap how long a respondent has to answer before a trial is recorded as timed out and the sequence moves on.
• Trial order: randomized per respondent by default (within each block, if you're using them), for order counterbalancing — turn it off if trial order itself is part of the design.

Domandata's timing is precise, but it's still collected in a respondent's own browser, not dedicated lab software (E-Prime, Inquisit, PsychoPy). Published comparisons of browser-based timing to lab hardware (de Leeuw & Motz, 2016; jsPsych's own hardware validation) find browser-collected reaction times run a small, fairly constant offset above lab-software measurements — generally tens of milliseconds. That's fine for within-subject and relative comparisons (a respondent's own trials against each other, or one block against another), which is what most Reaction Time designs actually analyze — but avoid comparing raw absolute reaction times from Domandata directly to published lab-software norms without accounting for it. The same caution applies to comparing keypress responses against touchscreen/click responses, which have their own input-device latency profile. Each trial records its input method, so you can separate or control for touch responses in analysis.

If your design is an IAT-style implicit-association task, know that the IAT's validity at predicting individual-level behavior is a genuinely live debate in the field, not a settled question — it's part of why alternatives like the Affect Misattribution Procedure and the Go/No-Go Association Task exist. Domandata gives you the trial/block infrastructure and raw per-block reaction-time and accuracy figures (see below); it doesn't compute a combined IAT D-score, and doesn't take a position on which paradigm or scoring approach is right for your study.

1. Step 1: Open Block Builder. Open the survey and select the block where the timed task belongs.

   Step 1: Open Block Builder

2. Step 2: Add a Reaction Time question. Choose Reaction Time from the question type selector.

   Step 2: Choose Reaction Time question type

3. Step 3: Build the trial list. Add each stimulus, mark a correct response where applicable, set an optional category for analysis grouping, and mark any warm-up trials as practice.

4. Step 4 (optional): Add blocks. For a multi-phase design, add named blocks and assign each trial to one — then turn on block-order counterbalancing if the design calls for it.

5. Step 5: Set response mode and labels. Choose keypress, click, or both, then name each response and assign a key if keypress is enabled.

6. Step 6: Tune timing and feedback. Set the fixation duration, the gap between trials, an optional response timeout, and whether incorrect responses should show feedback (and whether the correct response is required to continue).

7. Step 7: Name and code it. Add a readable variable name before piloting so the per-trial columns are easy to interpret in export.

   Step 7: Variable Name

8. Step 8: Preview the sequence. Run through the fixation → stimulus → response cycle yourself to confirm timing feels right before fielding.

   Step 8: Preview the sequence

Each response exports one column group per authored trial — stimulus, category, block, response, reaction time in milliseconds, correctness, and whether it timed out — keyed by the trial's authored position, not the shuffled order any individual respondent saw. This keeps a given trial's column meaning consistent across every respondent even though trial order is randomized per person. Domandata also computes summary columns, all excluding practice trials: mean reaction time, accuracy (when any trial has a correct response configured), and a timeout count.

When a question uses 2 or more categories, export also includes a mean-RT and accuracy column per category (for example, separate figures for "congruent" and "incongruent"); when it uses 2 or more blocks, export includes the same pair of columns per block. These are each block's or category's own raw figures — Domandata doesn't combine them into a single IAT-style D-score (see the methodology note above).

Reaction Time isn't one of the question types Silicon Sample can answer on an author's behalf — an LLM persona has no genuine millisecond-scale reaction to report, so the column is left blank rather than filled with a fabricated number.

• MaxDiff Survey Questions for another trial-based question format
• Attention and Manipulation Checks
• Survey Variable Names and Recodes
• Export Research Survey Responses