Response Funnel

Every response row Qualtrics has ever created for this survey, including previews, partial sessions, and rows later soft-deleted. This is the broadest count Qualtrics exposes and includes traffic that never became a real submission.

Responses Qualtrics considers eligible for export and audit. Excludes preview-mode and obviously-spammed rows but still includes incompletes and quality-flagged submissions before TABS triage.

Soft-deleted rows in Qualtrics. Includes test runs from the build phase, withdrawn duplicates, and any rows the survey administrator removed. Counted but not part of the analysis dataset.

TABS-reconciled response total after joining the Qualtrics export with Prolific submission data. This is the canonical denominator the daily pipeline starts from.

Distinct Prolific PIDs in the reconciled dataset. Differs from total responses when a participant submits more than once (we keep the first complete attempt).

Submissions dropped because the same Prolific PID had a more complete attempt elsewhere in the data. Currently zero — Prolific blocks reentry by default.

The participant cap configured for this Prolific study. Once filled, the study automatically pauses recruitment.

Slots Prolific has handed out — every participant who at least clicked through to start the survey, regardless of whether they finished, returned, or timed out.

Cash payment Prolific pays each approved participant. Stored in pence (UK) or cents (US) by Prolific; displayed here in major-unit currency.

Mean minutes a participant spent on the survey, computed by Prolific across all submitted attempts.

Implied hourly pay Prolific reports based on reward and average time. Prolific uses this to show participants whether a study meets fair-pay thresholds.

Submissions the TABS team has accepted on Prolific. Approval pays the participant and locks the row into the analysis dataset. This is the headline 'Surveys Completed' number on the homepage.

Submitted attempts that haven't been approved or rejected yet. The daily pipeline messages quality flags and approves clean submissions automatically; this bucket should drain to zero between runs.

Submissions the team rejected on Prolific because they failed quality gates (multiple failed attention checks plus speed). Participants are not paid; the response is not part of the analysis dataset.

Participants who claimed a slot and then returned the study without submitting. Returned slots are recycled back into the pool for new participants.

Slots Prolific reclaimed because the participant started but did not submit within the allowed window. The slot is recycled for a new participant.

Participants currently in the survey at the moment the daily pipeline ran. Snapshot only — usually zero except mid-burst.

Participants the daily pipeline sent a Prolific message to (typically about a quality flag). Counted by the TABS messaging script, not Prolific itself.

Passed every quality check: all 3 attention checks, no straightlining, duration above threshold, reCAPTCHA fine. Indicates a response is eligible for the standard approval path when submitted and included in the primary analysis sample.

Completed within the 5-9 minute Smeal benchmark band. Borderline-fast but otherwise passes checks; almost always approved after manual review.

Finished in under 5 minutes. Flagged for review because that pace is hard to reconcile with thoughtful answers, but not auto-excluded.

Showed low within-person variance on at least one Likert block (e.g., picked the same point on every readiness item). Not auto-excluded because some respondents legitimately answer that way.

Failed exactly 1 of the 3 instructed-response (attention check) items. One miss is usually inattention, not a bot — flagged for review and typically approved.

reCAPTCHA score sat in the borderline range. Flagged so the team can decide if the response should still count.

Failed multiple quality gates (typically 2 or 3 attention checks, often with speed). Removed from the analysis dataset; Prolific outcome may vary and is often reviewed separately.

Started the survey but did not reach the end. Includes both Prolific-returned and Prolific-timed-out attempts. Excluded from analysis but kept for funnel accounting.

Failed all 3 instructed-response items. Strong signal of inattention or automation; auto-excluded.

Failed all 3 attention checks AND completed too fast. The combination is the highest-confidence auto-exclude.

Failed 2 of 3 attention checks. Two misses crosses the threshold from 'inattention' to 'pattern,' so the response is auto-excluded.

Failed 2 attention checks AND completed too fast. Treated as auto-exclude.

Speed flag combined with at least one IRI failure that, together, crosses the auto-exclude threshold. Tracks responses where speed is the leading signal.

Share of evaluable submissions that selected the expected option ('Major Barrier') on the embedded barriers attention-check item.

Share that selected the expected option ('Low Readiness/Capability') on the readiness attention-check item.

Share that selected the expected option ('Level 2: Developing/Repeatable') on the maturity attention-check item.

Submissions evaluated for IRI pass rates (excludes incompletes and rows missing the IRI columns).

Distinct question IDs the Qualtrics survey definition reports. Matrix Likert blocks (Barriers / Readiness / Maturity) collapse to 1 ID even though they present many items.

What a participant actually sees: 19 barriers + 18 readiness + 9 maturity + 9 demographics + 1 open-ended feedback + 1 reCAPTCHA = 57. Derived from the TABS instrument constants.