People who learn
Understand how instruction, feedback, AI assistance, and independent transfer are evaluated.
Nonprofit Computational Research Lab
Better Learning and Teaching Through Verifiable Research Methods
For people who study, teach, and learn, we analyze how research methods and AI-assisted practice affect learning, teaching, scientific reasoning, evidence synthesis, code, and data analysis. Every case file makes its outcomes, assumptions, human review, and current status explicit.
Current public research record
- Portfolio
- 7 study records Two completed syntheses and five planned studies
- Current stage
- Evidence and protocol work Status is shown on every case file
- Completed outputs
- 2 synthesis records No Santaros participant outcomes claimed
- Last reviewed
- 29 August 2026 Status changes require a documented milestone
MISSION
Improve Learning and Teaching Through Better Methods
Santaros Labs is a nonprofit lab for people who learn, teach, and study education and scientific practice. We conduct analysis and research-methods work on instruction, feedback, independent transfer, evidence quality, and reproducibility, including planned work in Lithuanian learning contexts. The goal is a better learning and teaching experience supported by evidence, not faster output alone. Our Lithuania-facing work uses public Lithuanian education guidance and local language, access, and classroom conditions to frame questions that are relevant to learners, educators, and research teams.
WHO THIS IS FOR
A Research Home for People Who Learn and Teach
We bring learners, educators, learning scientists, and research-methods teams into the same clear record. Each group can see the question, the analysis, the limits, and the practical decision a study is meant to support.
People who teach
Use evidence and practical methods to improve lessons, assessment, access, and learner agency.
People who study learning
Review research questions, measures, analysis plans, validity risks, and completed evidence records.
People who build methods
Work across statistics, research software, data stewardship, governance, and reproducible analysis.
RESEARCH LIFECYCLE
From Question to Verification
Question
Frame an answerable scientific question and state which observations could change the conclusion.
Protocol
Specify the comparison, candidate outcomes, exclusions, analysis plan, and review requirements.
Execution
Record sources, data transformations, code, model details, and material human decisions.
Verification
Use independent checks, sensitivity analysis, reproduction, or replication where the design permits.
MEASUREMENT DOMAINS
What We Measure
Reasoning quality
Are hypotheses testable, assumptions visible, and confidence aligned with correctness?
Candidate measures: testability, calibration, expert error ratingsReproducibility
Can an independent researcher reconstruct the workflow and reproduce the reported result?
Candidate measures: trace completeness, environment recovery, outcome agreementEvidence support
Does each scientific claim remain faithful to its cited source and the uncertainty in the literature?
Candidate measures: citation support, contradiction detection, omission rateLearning transfer
Can learners apply a scientific practice independently after AI-supported instruction ends?
Candidate measures: delayed transfer, error detection, appropriate escalationEVIDENCE STATUS
Research Status Stays Visible
Study concepts, protocols, registrations, data collection, analyses, and completed outputs are distinct states. Each public record should show what exists, what does not yet exist, and what changed.
Preregistration-ready protocols
Reproducible computational workflows
Mixed-methods evaluation
Transparent limitations
Cross-disciplinary collaboration
COLLABORATION AND SUPPORT
Support Method Development and Independent Verification
Santaros Labs welcomes grants, compute credits, open-data partnerships, methodological review, domain expertise, educators, learning scientists, and replication partners for AI-assisted research training.
Support enables:
- Protocol development
- Research staff
- Secure compute
- Open-source tooling
- Replication studies
RESEARCH COLLABORATION
Bring a Scientific Workflow Worth Testing
Share the research question, current workflow, data constraints, and the decision that better evidence would support. We will begin with scope and method, not a predetermined claim.
No study begins until scientific accountability, data terms, and required oversight are documented.