1.7 Teaser: A seminar as treatment?

  • This seminar/lecture (treatment D) should affect your knowledge (outcome Y)

  • Discussion

    • Q: What is the general problem with this setup (‘our class’) if we are interested in measuring its impact?
    • Q: Is it sufficient to measure knowledge after the lecture?
    • Q: If yes, when should I measure knowledge after the lecture?
    • Q: Should I measure your knowledge here or at home?
    • Q: Is that a randomized experiment here?
    • Q: What’s the problem if someone knows everything beforehand?
  • Terminology for a starter

    • Treatment (variable), Outcomes/response (variable), treatment/control groups, pretreatment posttreatment

Notes

  • Lessons to be learned
    1. Ultimately, causal inference is about comparisons. Here we don’t have a group with whom we could compare those that took the class, i.e., a control group
    2. Measuring before and after treatment allows us to observe change
    3. Any outcome variables (e.g. knowledge) has a trajectory depending on how it is affected by the treatment. Depending on our measurement time point we may either fail to capture an change that was caused.
    4. Ideally, we also have control over the measurement process
    5. No…
    6. Depending on an initial outcome level we may not see any change/difference at all