Repository navigation
Meadows
This document provides a practical overview for lab members on how to design, implement and launch behavioral experiments using Meadows Research.
Meadows is a web‑based platform for building and hosting experiments. Researchers can configure tasks, upload and select stimuli, set up participant batches, recruit participants via services (e.g., Prolific or custom links) and export data. The platform is modular: each experiment contains one or more tasks presented to participants in sequence.
- Sign up. Go to Meadows and create an account using your lab or institutional email. If you plan to share projects with multiple researchers, use a shared email.
- Create a new experiment. After logging in, open your profile and click Create a new experiment. Give the experiment a descriptive name and short description; this will later appear in recruitment links and the internal dashboard.
- Version management. Every experiment has a design version that you edit. Once you’re ready to lock the design, go to the Versions tab and click Save the current version. Locked versions are numbered (Version 1, Version 2, etc.) and cannot be modified. You can continue editing the design version while still having access to data collected under previous versions.
Tasks determine what participants will do. Commonly used tasks include rating, matching and information pages, but Meadows offers many templates. To add a task:
- Go to the Tasks tab and click Add a new task. Give the task a unique name (e.g., task1).
- Choose a design from the list. A small question‑mark icon links to documentation for each task type.
- Configure task parameters (e.g., number of trials, trial duration, instructions). In rating and arrangement tasks you’ll also control how stimuli are displayed (size, mute videos, highlight on hover, etc.).
- Add General Information and Consent tasks when appropriate. The Consent task allows you to present terms and checkboxes; participants must tick all terms to proceed. You can customize the main text, add a fullscreen button and specify multiple terms. Remember it’s your responsibility to ensure the consent language meets your institution’s requirements.
- Use General Information or Slideshow tasks to provide instructions or examples before the main task. You can rearrange tasks by dragging them up or down in the list.
A Stimulus Set is a collection of media (images, videos, sounds) or text items that tasks draw from. Every task that presents stimuli must be linked to exactly one stimulus set. Stimulus sets can be reused across experiments.
- Individual files: Drag and drop jpg, png, ogv, mp4, mp3, wav or flac files onto the Media tab. Previews appear and you can remove items. After uploading, click Save Changes. At most 100 stimuli are displayed on the page; for larger sets use zip uploads.
- Zip uploads: Compress a folder of media into a .zip file and upload it. This method supports up to about 40 000 files per zip and can be repeated to upload multiple subsets.
- Finalize: Once satisfied with your set, click Finalize to lock editing and make the set available for experiments.
Text stimuli can be words, phrases or sentences in any language.
- To add a few items, open your stimulus set and use the Text tab to add each entry manually.
- For many items, create a table (e.g., in Google Sheets or Excel) where each row represents a stimulus: column 1 = “short” (default) version; column 2 = “long” (highlight) version. Export as CSV or TSV and upload via Import from file. Check for unexpected characters (quotes, semicolons) before uploading.
- You can use text stimuli for instructions inside tasks like Slideshow and combine them with sampling options to vary instructions across participants.
Meadows offers “feature stimuli” extracted from other sets using decomposition algorithms. These are currently used with the Diagnostic Feature Mapping task and require contacting the Meadows team.
When editing a task, go to the Stimuli tab. You can choose a subset of the stimulus set and control presentation options.
- Source: Choose whether to use (1) no stimuli; (2) stimuli from a preceding task (copy all or only those selected by the participant); or (3) a stimulus set
- Trials: By default Meadows samples stimuli randomly per trial. To specify exact trial sequences, select upload file when creating batch and later provide a tab‑separated “trial file” when opening a participant batch . A trial file has one row per trial per participant; the first column is the participant ID/token and subsequent columns list stimulus names.
- Exclude preceding: Remove stimuli that were already presented in earlier tasks.
- Filters: Filter stimuli by file type (e.g., .jpg, .mp4) or by specific file names.
- Sampling: • All: Use up to the first 1,000 stimuli. • Random: Each participant sees a random subset. • Balanced: Allocate stimuli so that items selected less often are shown preferentially in subsequent participants, aiming for equal sampling across your sample. • Configuration options: Control how stimuli appear. Settings include muting video sound (mute video), hiding images or videos until hovered (only show highlighted), adjusting drag boxes and text size, controlling audio play mode, and toggling highlight behavior.
After adding tasks and stimuli:
- Use Preview Task on the right side of the design page to see how participants will experience the task. For example, in a Drag‑rate task, you drag stimuli onto an arena and then click the Finish button to end the trial.
- Adjust parameters if necessary and preview again.
- Run small pilot tests with colleagues to ensure that stimuli load quickly (< 100 MB is recommended per session) and data are recorded correctly.
- Once a version is saved, go to its Participants tab. The design version does not have participant management; ensure you’re on a saved version.
- A Batch defines a set of participant slots with specific recruitment rules. The page lists existing batches (letters A, B, C…) with progress bars showing started and completed participants.
- To create a new batch, click Create a new Batch. Select a template (e.g., Share link or a recruitment service). Configure options such as requiring participants to have a Meadows account or storing the referring page. Set the number of slots (recommend larger than your target sample to account for drop‑outs) . Click Create to finalize the batch.
- After creation, copy the Entry URL to share with participants. Short or batch‑specific versions of the link can be generated; the “any” link forwards to any batch with available slots.
If you use a third‑party service like Prolific, each platform has its own setup.
Prolific • Use the Entry URL from your batch as the Study URL in Prolific. • Add entry variables to capture metadata (recommended: PROLIFIC_PID for participant ID, STUDY_ID and SESSION_ID). • In Meadows’ Exit URL, paste the completion URL from Prolific. The code after cc= automatically approves submissions; use NOCODE if you prefer to review manually. • On the Exit page you can choose to display a return button, the completion code, the participant’s Meadows alias, and automatically forward to Prolific after a few seconds.
Other services (Sona) follow similar patterns; refer to the specific recruitment guides when needed. Using JHU Sona might actually be simpler and more convenient if you want to have the JHU students be study participants. It's an easy and no-cost way to get data, since you pay participants in course/research credits. It uses a very similar, if not exactly the same, structure as Prolific, and there is more information on their website about it.
For custom setups (e.g., emailing participants directly), use the Share link template. You can also store a participant token in the URL and reference it in trial files to pre‑assign stimuli.
- In the Participants tab, click on a participant nickname (e.g., purple‑elephant) to view their tasks and times. Each task has a download icon. You can download data for individual tasks, individual participants, or the whole experiment.
- When exporting data, decide which range (all participants vs. a single task) and which structure (tree, table, annotations, matrix, events, metadata). The tree structure retains all variables; tables simplify values into rows/columns.File formats include CSV (for spreadsheets), MAT (for Matlab), JSON (general programming) and LOG for event streams.
- Browser performance: Loading more than ~100 stimuli can slow down the preview, particularly when using video or audio. Consider splitting large stimulus sets into multiple tasks or using zip uploads for efficient loading.
- Randomization: Use the built‑in sampling options for simple random or balanced sampling. For complex counterbalancing, create trial files to pre‑define stimulus sequences.
- Instructions and Consent: Always provide clear instructions (General Information or Slideshow tasks) and a Consent form that meets your ethics requirements.
-
Testing: Pilot your experiment on different browsers (Chrome, Firefox, Safari) and devices. Verify that videos autoplay (set them to muted if needed) and that audio plays correctly. (IMPORTANT: If your stimulus set consists of videos, you may need to use workarounds like
ffmpegto use the appropriate CODEC for videos to be playable across all browsers. It is difficult to make it available all three major browsers, but shoot for Firefox and Chrome. Safari is very difficult and it may prevent usability across more widely-used browsers like Chrome. Make sure you test it with the correct browser.) - Budgeting: Meadows charges a small fee per participant minute, with a free allowance for the first 120 minutes per researcher. Plan your batch sizes accordingly.
- Meadows documentation: https://docs.meadows-research.com/
- Tutorial videos on the Meadows YouTube channel.
- If your desired paradigm isn’t available, contact the Meadows team about Custom tasks.
- The quickest point of contact, especially if your study needs a feature that does not quite exist on the platform already, is to contact Jasper van den Bosch (jasper@meadows-research.com). He has worked with me (@kathy) personally before, and he has been super helpful. He should be able to help you with whatever you need!