
Bot Checks
A bot check in Labvanced is a short task placed in your study flow that asks the participant to do something a script cannot easily fake, such as saying a word aloud, facing the screen, or picking the one color word among emotion words. You build the check from the same objects, triggers, and actions as the rest of your study, so it can sit at the start, between tasks, or wherever your design needs it.
Table of Contents
Why Remote Studies Use Bot Checks
In a lab, a researcher sees who sits down at the computer. In an online study that runs from a link, the researcher does not, and a script can submit responses as easily as a person can. A bot check gives you a defined, documented criterion: participants who pass continue, and participants who fail can be removed from the session.
Bot checks belong to the engagement side of Experimental Control, next to fullscreen monitoring and attention checks. They screen for scripted and inattentive responding, and no single check catches everything, so the examples below cover three different kinds of evidence.
Example Studies and the Principles Behind Bot Checks
Each example is a public library study that shows one principle behind bot checks. Open it, try it as a participant, import a copy to see the events behind it, and modify it to fit your own study. The Importing Templates guide shows how to bring a study or block into your own project.
Voice Check with Speech Transcription
In the Voice-controlled Study / Transcription example, participants sort images into a left or right pile by saying the answer aloud. Holding the space bar records the voice, releasing it stops the recording, and the speech is transcribed in the participant's browser with Whisper. The transcript decides where the image goes. See Text Transcription for how transcription works.
The study has five trials. You can edit it down to two trials for a quick check at the start of your own study, or use the same principles to build a new one: record a short spoken response, transcribe it, and compare the transcript with what the task asked for. As shipped, the study sorts images rather than screening participants, so add a Reject/End Session action to the failure branch if you want participants who do not respond correctly to leave the study.
Head Tracking Attention Check
In the Attention Control with Head Tracking example, a video plays only while the participant is facing the screen. The Head Tracking / Face Tracking trigger follows the head through the webcam, and every 100 milliseconds a short JavaScript snippet turns head turn and tilt into a screen attention value between 0 and 1. When the value drops, the video pauses and the screen shows "You are not facing the screen!". It resumes when the participant turns back, and the study moves on when the video ends.
The video is only the example target: you can replace it with an image or any other target the participant has to face. The same logic also works as a pass or fail rule. You can require a participant to turn the head or move closer to the screen at a given moment and record whether they did. Head and face tracking runs in the participant's own browser, see Head & Face Tracking for the measures available.
Multiple Choice Check
The Choose a Color - Bot Check add-on asks the participant to select the response that refers to a color and to avoid the responses that refer to an emotion. The options mix words such as Happy, Angry, Peaceful, Green, and Orange. After the participant submits, the study shows a correct or incorrect screen and then moves on, and you edit the follow-up to fit your design, for example to remove participants who answer incorrectly.
It takes seconds to complete and works on any device.
Choosing a Check
| Check | Example study | What it checks | Needs from the participant | Consider it when | Keep in mind |
|---|---|---|---|---|---|
| Voice and transcription | Voice-controlled Study / Transcription | A live voice and a spoken answer that matches the task | Microphone permission, a place to speak | Your study already collects audio, or you want evidence of a live voice | Participants who cannot or prefer not to speak need an alternative path |
| Head tracking | Attention Control with Head Tracking | A live person at the screen, in real time | Webcam permission | Your study is about attention, video, or anything the participant watches | Mention the webcam in your consent materials |
| Multiple choice and text-based tasks | Choose a Color - Bot Check | Reading comprehension and attention to instructions | Nothing beyond the screen | You want a fast screen that works on every device | A program that reads text can answer a text question, so this screens inattention and simple scripts more than determined ones |
You can use two kinds of check in the same study, for example a text question at the start and a voice or head tracking check later on. A script that can answer the text question usually cannot pass the second one, so the two checks together catch more than either does alone.
Placing a Check in Your Study
A check can be a block of its own at the start of the study, a task between two experimental tasks, or a single frame inside a task. Import the add-on as a block or task, or copy the events from the library example into your own frame. Use a Reject/End Session action to remove participants who fail, and record the result in a variable so that passes and failures stay visible in your data export.
Questions about your study design or platform fit?