Online Experiments
Date: 1. Oktober 2025
Speaker: Onur Özsoy, Project C03, SFB 1252
Duration: 14:00 - 15:30
Presenter Materials
- Slides (PDF): Online_experiments_RDM_SFB1252_2025.pdf
- Analysis script (R): ResearchDataMethod.R
- Results data (CSV): results_prod.csv
Overview
ntroduction to designing, implementing, and conducting online experiments for linguistic research. Covers platform selection, participant recruitment, data quality control, and ethical considerations.
Learning Objectives
- Design effective online experiments for linguistic research
- Select appropriate platforms and tools
- Implement robust participant recruitment strategies
- Ensure data quality in online settings
- Address ethical considerations specific to online research
- Troubleshoot common online experiment issues
Topics
Platform Overview
- jsPsych for behavioral experiments
- Prolific for participant recruitment
- Amazon Mechanical Turk (MTurk) considerations
- Qualtrics for surveys and simple experiments
- Lab.js and PsyToolkit alternatives
Experiment Design
- Adapting lab experiments for online delivery
- Timing precision and technical limitations
- Browser compatibility and device considerations
- Attention checks and data validation
Participant Management
- Recruitment strategies and screening
- Payment and compensation logistics
- Communication with participants
- Handling technical issues and dropouts
Data Quality
- Identifying low-quality responses
- Attention checks and validation measures
- Statistical approaches to data cleaning
- Handling missing data and exclusions
Ethics and Legal Issues
- Informed consent in online settings
- Data protection and GDPR compliance
- Participant anonymity and privacy
- Cross-border research considerations
Prerequisites
Basic familiarity with experimental design concepts. Programming experience helpful but not required.
Notes (from slides)
Presenter: Onur Özsoy
Date: 1 October 2025 (Uni Köln)
Contact: ooezsoy@uni-koeln.de
Why webcam-based online eye-tracking
- Reach remote and hard-to-reach, non-WEIRD populations (e.g., regions without labs; heritage speakers outside universities).
- Cost: high-end eye-trackers ($10k+) vs. webcam-based (no additional hardware cost).
- Approachability: less technical training than lab setups; online training/support available.
- Context: pandemic accelerated remote data collection.
Core concept and tooling
- Webcam tracks pupil movement; gaze estimation computed locally in participant’s browser.
- WebGazer.js: "Democratizing Webcam Eye Tracking on the Browser" — https://webgazer.cs.brown.edu/
- PCIbex: platform to program and host experiments (free; docs/support available).
- Docs: https://doc.pcibex.net/
- Paper: Zehr, J., & Schwarz, F. (2018). PennController for Internet Based Experiments (IBEX). https://doi.org/10.17605/OSF.IO/MD832
- Typical tasks supported: surveys, AJTs, ratings, Stroop, picture selection, audio recording, eye-tracking, lexical decision, mouse tracking, etc.
Limitations and caveats
- Timing delays: ~300 ms average between true gaze and recorded time (Slim & Hartsuiker, 2022).
- Accuracy limits and varying sampling rate (Zhang et al., 2022).
- Calibration can be difficult.
- Suitability: often "good enough" for Visual World Paradigm (VWP); consider supervised webcam ET to improve data quality/control.
Step-by-step quickstart
1) Create an account: https://farm.pcibex.net/ (you can create multiple accounts if needed). 2) General guidelines for ET: https://hu.berlin/ET 3) HU Berlin farm: https://korpling.german.hu-berlin.de/ibex/ 4) Explore a demo and copy into your account: https://farm.pcibex.net/r/jsBNEF/ 5) Example experiment components: translation tasks, Stroop task, survey.
Analysis
- "Jump to R" indicated for analysis workflow; details not included on slides.
Additional resources and examples
- WebGazer.js: https://webgazer.cs.brown.edu/
- PCIbex documentation: https://doc.pcibex.net/
- Paper: Zehr & Schwarz (2018), https://doi.org/10.17605/OSF.IO/MD832
- Analysis (Turkish case processing): https://osf.io/sehnf/?view_only=a0a306606f064652b14b03e29a52d27d
- Turkish case processing experiment: https://farm.pcibex.net/r/GOsYzn/
- German case processing experiment: https://farm.pcibex.net/r/gcwFgV/
Key takeaways
- Webcam-based ET via PCIbex/WebGazer is accessible, free, and feasible for many psycholinguistic paradigms.
- Plan for timing/accuracy limitations; design tasks accordingly and consider supervision to boost data quality.
Workshop Discussion Points
Platforms and task design
- Randomization and lists (PCIbex)
- Use
Template()with CSV for lists/conditions. - URL parameter filtering for list assignment; enables quick list handling.
- Mobile devices
- Suitability and constraints; consider disabling or gating by screen size/orientation.
- Timing
- Be mindful of webcam ET timing variance; design tasks tolerant to delays.
- Effort and plugins
- Factor in learning curve; use community plugins and examples.
Hosting and data governance
- GDPR/DSGVO compliance
- Pavlovia: EU-friendly hosting option; check DPA and data flows.
- PCIbex Farm: quick start but US-based; consider self-hosting for GDPR.
- jsPsych
- Can use static hosting (GitHub Pages/Netlify) or Pavlovia.
- Data capture: serverless CSV download or custom backend/API.
- PCIbex
- Farm hosting (fast onboarding) or self-host for compliance.
- Built-in result capture and download options.
General online design considerations
- Session duration and fatigue management.
- Response filtering (invalid/too fast/failed calibrations).
- Breaks and pacing to maintain attention.
- Attention checks embedded at intervals.
Participant recruitment
- Balanced sampling across lists/conditions.
- Scaling to large samples when needed.
- Prolific integration
- Pre-screening, approval rate thresholds.
- Payment on completion; use bonus for performance/longer tasks.
Demos and Links
- PCIbex account: https://farm.pcibex.net/
- ET guidelines: https://hu.berlin/ET
- HU Berlin farm: https://korpling.german.hu-berlin.de/ibex/
- Demo (copy to your account): https://farm.pcibex.net/r/jsBNEF/
- WebGazer.js: https://webgazer.cs.brown.edu/
- PCIbex docs: https://doc.pcibex.net/
- Zehr & Schwarz (2018): https://doi.org/10.17605/OSF.IO/MD832
- Turkish analysis (OSF): https://osf.io/sehnf/?view_only=a0a306606f064652b14b03e29a52d27d
- Turkish experiment: https://farm.pcibex.net/r/GOsYzn/
- German experiment: https://farm.pcibex.net/r/gcwFgV/
Additional Resources
- jsPsych Documentation - Comprehensive programming guide
- Prolific Academic - Participant recruitment platform
- Ethics in Online Research - BPS Guidelines
Related Workshops
This workshop complements the ethics approval and preregistration workshops in understanding online research methodology.