Workflow leaderboard / General / Data collection
Collect reproducible behavioral experiment data
采集可复现的行为实验数据
Build and pilot a behavioral experiment with versioned stimuli, explicit randomization and row-level provenance before collecting participants.
Variants
Free & local / 免费本地
~60 min setup
A current laptop is sufficient for most keyboard and visual tasks. Timing-sensitive audio, eye tracking or external devices require hardware validation.
| Component | Role | Price | Link |
|---|---|---|---|
| PsychoPy 2026.2.1 | experiment construction, presentation and local data capture | Free / 免费GPL-3.0 open-source software; online hosting or credits are separatechecked 2026-08-02 | https://github.com/psychopy/psychopy/releases/tag/2026.2.1 |
Experiment provenance schema
required_columns:
- participant_pseudonym
- session_id
- experiment_git_commit
- psychopy_version
- condition
- block
- trial_index
- stimulus_id
- response
- correct
- reaction_time_seconds
- trial_started_utc
- completion_status
randomization:
seed_recorded_per_session: true
allocation_test_before_launch: true
quality_flags:
- focus_lost
- frame_drops
- duplicate_session
- incomplete
Freeze stimuli and run a pilot gate
mkdir -p provenance
sha256sum stimuli/* > provenance/stimuli-SHA256SUMS
git rev-parse HEAD > provenance/experiment-commit.txt
python --version > provenance/runtime.txt
python -c "import psychopy; print(psychopy.__version__)" >> provenance/runtime.txt
# Run at least five complete pilot sessions, then inspect condition counts and missing fields before recruitment.
Known pitfalls
- Local success does not prove timing accuracy on participant hardware.
- Do not collect direct identifiers in the trial table.
- Randomization must be tested for balance and recorded per session.
- PsychoPy version compatibility and device drivers can affect timing.