Advanced Research Methods
General info
Expected knowledge:
- Grasp of regression analysis
- Statistical testing
- Descriptive analysis
Bonus for the exam in the last week of the workgroups!!
Workgroup session really important – exam questions from there!!
Before workgroup:
- Read cases
- Answer questions
- Connection lecture – lit
Final week: presentation on quali or quanti topic (topics will be assigned)
One group presents, other asks critical questions and vice versa.
Causal inference – drawing the lines between causes and efects
Modern approach to quantitative research
- 1960-2005: methodological development focused on statistical methods
o development new techniques
o improvements in computers
o standardized tests, objectivity
o helpful and harmful (not entirely correct always)
- newer developments (not black and white! Data does not tell everything)
o causal theory
o what should be part of a quantitative analysis?
o Interpretation: meaning of results depends on context
o More than just observing data, it has to have a causality
- The biggest part of quantitative analyses is not about numbers (but about the design
and how you got there)
- Miguel Hernan (using his terminology)
- Epidemiology: “is the study of the distribution of health-related states and events in
the population.”
Crucial question in causal inference: what would have happened to their skin if they had not
used a product e.g. (also over a longer term of time). With and without. What is an efect?
Determining that as well (health, looks etc). But central question is what would have
happened with and without?
Improving implies a causal efect: A leads to B, using a product leads to beter results.
, Problems
- Small sample
o Is that always a problem?
o Less precise and certain estimates (doesnnt necessarily lead to bias)
- Study performed or fnanced by commercial company
o Always legit?
- No control group
o Essential omission
o What would have happened without treatment?
o Potential regression towards the mean
Causation
- Formal defnition (Hernan/Robins):
o In an individual, a treatment has a causal efect if the outcome under
treatment 1 would be diferent from the outcome under treatment 2
Potential outcomes
- Causal efects
-
Counterfactual outcome: potential outcome that is not observed because the subject did
not experience the treatment (counter the fact)
Fundamental problem
- Individual causal efect cannot be observed
o Expect under extremely strong (and generally unreasonable) assumptions
- Causal inference as a missing dara problem
- We need a diferent defnition of causal efects
- Average causal efects can be determined under
o No assumptions (randomized studies)
o Strong assumptions (observational studies)