Lecture – Advanced Research
Methods
Inhoudsopgave
Lecture 1 – Introduction to causal inference...................................................................1
Knowledge video 1 – DAG (Directed Acyclic Graphs)...................................................8
Lecture 2 – OLS and moderations.................................................................................10
Knowledge video 2: OLS regression...........................................................................13
Knowledge Video 3 – Logistic Regression..................................................................13
Knowledge Video 4 – OLS vs. Logistic Regression......................................................15
Knowledge Video 5 – Table 2 Fallacy..........................................................................16
Lecture 3 – OLS and Logistic Regression.......................................................................18
Lecture 4 – Qualitative Methods...................................................................................20
Knowledge video 6 – Discourse analysis in healthcare..............................................21
Knowledge video 7 – Taking discourse into analysis into the field.............................22
Knowledge Video – Ethnography in healthcare research..............................................23
Lecture – Organizational ethnography..........................................................................24
How to assess quality in qualitative research...............................................................25
Lecture 1 – Introduction to causal inference
Learning Goals:
Why examine (statistical) associations?
1. Descriptions: patterns X and Y
2. Predicition: Y given X (characteristics of something on the research
question (RQ))
3. Causal inference: Effect X on Y
,Example
Would you buy this foundation or not?
- Small sample size (n=41)
Is it always a problem?
not always, depends on what you want to know.
- Study performed or financed by commercial company
Is this a problem? Not always, but it can be frowned
upon. Make sure to have a contract which says you can
publish anything, positive and negative
- No control group
Is it a problem?
Essential data is missing, what would happen
without treatment, potential regression to the mean
What do we want to know in causal inreference:
- We are not interested in the outcome (Y, 70% less imperfections) but
- We are interested in the role of the treatment (X, without the
foundation) in achieving this outcome
Conclusion
- We do not have that information
- No causal claim can be made base on L’Oréal study
Causal effect
Formal definition by Hernan and Robins (2020)
In an individual, a treatment has a causal effect if the outcome
under treatment 1 would be different from the outcome under treatment
2.
To assess this, we need information on:
what would have happened, had this not happened
Assume that we have this information in relation to the foundation study:
- Woman A treated with the foundation: 2 bad spots
- Had Woman A not been treated with the foundation: 5 bad spots
Individual treatment effect: -3 spots (or 60% less imperfections)
Average treatment effect: average of individual effects in a population
Formal notation of a causal effect:
Y = outcome
A = treatment
, i = individual
1 = yes (received treatment)
0 = no (received no treatment)
Does not equal
Not all potential outcomes are observed
- Counterfactual outcome: potential outcome that is not observed
because the subject did not experience the treatment (counter the
fact)
- Potential outcome is factual (or observed) for some subjects, and
counterfactual (or not observed) for others
Fundamental problem in causal inference
Individual causal effect cannot be observed:
- No information on counterfactual
- Except under extremely strong (and generally unreasonable)
assumptions
Average causal effect cannot be determined based on individual estimates
- Causal inference as a missing data problem
So, we need a different approach to estimate causal effects.
Identifiability conditions
Average causal effect van be determined if, and only if, 3 identifiability
conditions are met:
1. Positivity
2. Consistency
3. Exchangeability
If all conditions are met (and an association is found in the data), the
association between exposure and outcome is an unbiased estimate of a
causal effect and you can make a causal claim
Positivity
Methods
Inhoudsopgave
Lecture 1 – Introduction to causal inference...................................................................1
Knowledge video 1 – DAG (Directed Acyclic Graphs)...................................................8
Lecture 2 – OLS and moderations.................................................................................10
Knowledge video 2: OLS regression...........................................................................13
Knowledge Video 3 – Logistic Regression..................................................................13
Knowledge Video 4 – OLS vs. Logistic Regression......................................................15
Knowledge Video 5 – Table 2 Fallacy..........................................................................16
Lecture 3 – OLS and Logistic Regression.......................................................................18
Lecture 4 – Qualitative Methods...................................................................................20
Knowledge video 6 – Discourse analysis in healthcare..............................................21
Knowledge video 7 – Taking discourse into analysis into the field.............................22
Knowledge Video – Ethnography in healthcare research..............................................23
Lecture – Organizational ethnography..........................................................................24
How to assess quality in qualitative research...............................................................25
Lecture 1 – Introduction to causal inference
Learning Goals:
Why examine (statistical) associations?
1. Descriptions: patterns X and Y
2. Predicition: Y given X (characteristics of something on the research
question (RQ))
3. Causal inference: Effect X on Y
,Example
Would you buy this foundation or not?
- Small sample size (n=41)
Is it always a problem?
not always, depends on what you want to know.
- Study performed or financed by commercial company
Is this a problem? Not always, but it can be frowned
upon. Make sure to have a contract which says you can
publish anything, positive and negative
- No control group
Is it a problem?
Essential data is missing, what would happen
without treatment, potential regression to the mean
What do we want to know in causal inreference:
- We are not interested in the outcome (Y, 70% less imperfections) but
- We are interested in the role of the treatment (X, without the
foundation) in achieving this outcome
Conclusion
- We do not have that information
- No causal claim can be made base on L’Oréal study
Causal effect
Formal definition by Hernan and Robins (2020)
In an individual, a treatment has a causal effect if the outcome
under treatment 1 would be different from the outcome under treatment
2.
To assess this, we need information on:
what would have happened, had this not happened
Assume that we have this information in relation to the foundation study:
- Woman A treated with the foundation: 2 bad spots
- Had Woman A not been treated with the foundation: 5 bad spots
Individual treatment effect: -3 spots (or 60% less imperfections)
Average treatment effect: average of individual effects in a population
Formal notation of a causal effect:
Y = outcome
A = treatment
, i = individual
1 = yes (received treatment)
0 = no (received no treatment)
Does not equal
Not all potential outcomes are observed
- Counterfactual outcome: potential outcome that is not observed
because the subject did not experience the treatment (counter the
fact)
- Potential outcome is factual (or observed) for some subjects, and
counterfactual (or not observed) for others
Fundamental problem in causal inference
Individual causal effect cannot be observed:
- No information on counterfactual
- Except under extremely strong (and generally unreasonable)
assumptions
Average causal effect cannot be determined based on individual estimates
- Causal inference as a missing data problem
So, we need a different approach to estimate causal effects.
Identifiability conditions
Average causal effect van be determined if, and only if, 3 identifiability
conditions are met:
1. Positivity
2. Consistency
3. Exchangeability
If all conditions are met (and an association is found in the data), the
association between exposure and outcome is an unbiased estimate of a
causal effect and you can make a causal claim
Positivity
