Post-hoc or Secondary Research Guide

This guide helps you design research studies that can be conducted with SafeInsights.

Research Patterns

We've identified six common patterns for post-hoc or secondary research with learning platform data:

Pattern 1: Engagement & Usage Analysis

Research Question Example: "How does time spent on practice problems relate to quiz performance?"

Data Needed:

  • Event logs (page views, interaction timestamps)
  • Assessment data (quiz scores)

Common Methods: Correlation analysis, regression modeling

Example Platforms: OpenStax, ASSISTments

Limitations:

  • Correlation ≠ causation
  • Self-selection bias (students choose how much to practice)

[Link to full example in OpenStax data catalog]

Pattern 2: Performance Relationships

[Similar structure]

Pattern 3: Longitudinal Trends

[Similar structure]

Pattern 4: Subgroup Comparisons

[Similar structure]

Pattern 5: Predictive Modeling

[Similar structure]

Pattern 6: Learning Curve Analysis

[Similar structure]

From Idea to Feasible Study

Use this decision tree to assess feasibility:

  1. Is your question observational?

    • Yes → Continue
    • No (requires intervention) → Not supported in v1

  2. Do you need data from multiple platforms?

    • Yes → Not supported in v1 (fusion coming later)
    • No → Continue

  3. Check the data catalog: Does a member have the data you need?

    • Yes → Continue
    • No → Consider alternative questions or contact us

  4. Can you answer with aggregated outputs?

    • Yes → Likely feasible!
    • No (need individual records) → Redesign study

Next Steps