Quality control

8.1 Pre-Survey Verification¶

8.1.1 Overview¶

When enabled, respondents must pass a verification screen before starting the survey. This screen can include reCAPTCHA bot protection, an audio device check, and/or a Terms & Conditions consent checkbox.

8.1.2 Configuration¶

Go to Settings > Pre-Survey Verification section. Three options are available:

reCAPTCHA Verification (Bot Protection): Invisible bot protection powered by Google reCAPTCHA Enterprise. It automatically analyzes user behavior and assigns a risk score. No user interaction is required.
Audio/Video Device Check: Plays a test tone so respondents can verify their audio is working before starting the survey. Non-blocking -- the respondent can always continue.
Terms & Conditions Consent: Shows a checkbox the respondent must tick to accept the Terms and Conditions and the Privacy Policy before starting. This provides the legal basis to collect the AI attention index during the survey. The consent text is shown in the respondent's language, with links to both documents. Blocking -- the respondent cannot continue until the box is ticked.

8.1.3 How it works¶

The respondent clicks the survey link.
If verification is enabled, a verification screen appears before the first question.
reCAPTCHA runs automatically in the background.
For the audio check: the respondent plays a test tone and confirms they can hear it.
For the Terms & Conditions consent: the respondent ticks the checkbox accepting the Terms and Conditions and the Privacy Policy.
The respondent clicks "Continue to Survey."
If reCAPTCHA passes (score >= 0.5) and the consent (when enabled) was ticked, the respondent proceeds to the survey.
If reCAPTCHA fails, the respondent is blocked and the attempt is recorded in the Quality Checks Report under the reason reCAPTCHA rejected (bot protection), with the score Google returned, the threshold and the panel of origin, so you can see how many visits the bot protection stopped. The blocked visit is counted under Quality Terminate in the survey list, where the counter links to the report.

Notes: - Verification is skipped for respondents resuming a saved survey. - The verification screen only appears when the survey is in "Open" status (not in test/preview mode). - The reCAPTCHA site key is configured by the platform administrator.

8.2 Cross-Panel Deduplication¶

This feature detects and blocks respondents who attempt to complete the same survey from different research panels (e.g., TickPanel, Netquest, Cint).

8.2.1 How it works¶

The system uses FingerprintJS v3 to generate a unique browser fingerprint based on hardware characteristics (canvas, WebGL, audio context, installed fonts, screen resolution, etc.). The fingerprint is checked when a respondent enters the survey and stored when their participation ends (Done, Early Screen Out, Quota full or Quality terminate). If the same fingerprint appears again for the same survey, the respondent exits through the Early Screen Out redirect and the attempt is recorded — with its panel of origin and panelist id — in the Early Screen Out Report, together with the earlier participation of that device (panelist id, Survey Id, outcome and panel) when it was recorded.

8.2.2 Configuration¶

Go to Settings > Entrance Filter Settings > enable Cross-Panel Deduplication (Browser Fingerprint).

Important: This option is mutually exclusive with "Same Computer Control" (cookies). Enabling one automatically disables the other, as they serve similar but incompatible purposes.

Only the SHA-256 hash of the fingerprint is stored, not the raw data.
Data is linked to the survey, not to a persistent user profile.
Each fingerprint is stored together with the panelist id (pid) and the outcome (Done, Early Screen Out, Quota full, Quality terminate) of the participation that left it, within the same survey only. This link is used only to report which earlier participation a blocked entry matches. It is never shared with the panels and is deleted together with the survey's fingerprints.
Fingerprints can be deleted when the survey is closed.

8.3 Careless Attention¶

Careless-responding detection has two complementary parts in the platform:

In-survey time controls (this section): minimum reading time per step that can eject rushing respondents, and the recording-time warning below.
Careless indices: statistical indices (longstring, IRV, Mahalanobis distance, even-odd, psychometric synonyms/antonyms) computed with the R careless package on demand — at any point during fielding or after the survey is closed — see the Careless Indices Report and its visual summary in the Interactive Dashboard's Data Quality tab.

8.3.1 Attention, We Are Recording Time¶

To inform users that response times are being recorded, you can add a warning at the top of the page. The following image shows the warning that will be visible for the entire duration of the step:

This warning can be configured per step. To include it in a specific step (e.g., Step 4), go to the "Survey structure" form and set the value "1" in the column with the header "Warning."

For "multiple choice" questions, you can display this message on specific cards rather than all cards. To do this, edit the desired questions and enable the "Visualizar dado" checkbox.

8.4 Quality Checks Report¶

A dedicated report displays quality check results and the reasons respondents were terminated. Access it via Results > Quality Checks.

The report shows: - Total respondents terminated by quality checks - Breakdown by reason: minimum time, maximum time, careless response patterns, and visits rejected by the pre-survey reCAPTCHA (reason reCAPTCHA rejected (bot protection), shown with the score and the threshold; the visit never reaches the first question) - Average time per step for completed respondents

Duplicate-device detections (Cross-Panel Deduplication) are reported in the Early Screen Out Report, since those respondents exit through the Early Screen Out redirect.

8.5 Respondent Exit Scenarios¶

When a respondent exits the survey for any reason, they are redirected to the corresponding Exit URL configured in the survey settings (see Section 2.3.6 - Exit URLs). The following table summarizes the possible exit scenarios and what the respondent experiences:

Scenario	Trigger	Redirect URL Used
Survey Closed or Paused	Respondent accesses a survey that is no longer in "Open" status.	The respondent sees a message indicating the survey is not available.
Quota Full	All quota slots are filled and no more completions are needed for the respondent's segment.	Redirected to the Quota Full URL.
Early Screen Out	The respondent does not meet the screening criteria: screen-out answer, missing quota combination, prior participation (pid exclusion) or duplicate device (fingerprint). See the Early Screen Out Report for the per-respondent reason.	Redirected to the Early Screen Out URL.
Quality Terminate	The respondent fails quality checks such as minimum time, maximum time, or careless attention detection.	Redirected to the Quality Terminate URL.
Already Completed	The respondent has already completed the survey and attempts to access it again (detected via cookie or fingerprint).	Redirected to the Already Done URL.
Successful Completion	The respondent finishes all steps of the survey.	Redirected to the Complete (Done) URL.

When distributing through research panels (Netquest, Cint, TickPanel, etc.), replace the default TickStat redirect URLs with the panel's own redirect URLs so that respondent status is properly communicated back to the panel.