Covered in lecture on 8/31
Ways to get data
- Observational studies
- Experiments
- Synthetic - Generated by a computer, as an alternative to real-world data
Observational studies
Kinds of observational studies:
- Cross-sectional - Look at data in a single point in time
- Retrospective (case control) - Looking at studies of events in the past
- Prospective (aka longitudinal or cohort)
- Follow group around and observe closely
- High dropout rates
Surveys (kind of observational study)
Issues:
- Wording of questions can introduce bias
- Ordering of questions
- Convenience samples/self-selected samples
- Desire of respondents to please
- Confidentiality concerns may influence response
- Non-response bias
Experiments
- Observe effects on subjects after the application of some treatment
- Might want to compare a treatment versus a control or multiple treatments
- The more variables you can control for, the better your experiment
- Want the control to be as close to the experiment group as possible, except for the thing you’re testing
Fundamental Rule of Data Collection
Your data must be representative of the population you want to study