DATA COLLECTION IDENTIFYING VARIABLES LIKELY TO BE IMPORTANT TO INCLUDE IN MY ANALYSIS 5/1/17

Question 1

What does a DAG stand for?

Answer 1

directed acyclic graphs

Question 2

What does a DAG allow us to do?

Answer 2

1. summarise the functional, theoretical and speculative relationships between variables relevant to our question
2. identify any co-variates acting a confounders so that these can be adjusted for
3. identify any measured/able covariates
4. identify competing exposures

Question 3

What are confounders?

Answer 3

they cause both the outcome and the exposure

Question 4

Why do we need to adjust for confounders?

Answer 4

they create a pseudo-causal path between the outcome and the exposure which will generate statical relationship between the 2 even when none exists

Question 5

What are mediators?

Answer 5

they case the outcome and are caused by the exposure

Question 6

What are competing exposures?

Answer 6

• they cause the outcome
• but no relationship with the exposure
• might have to adjust

Question 7

What can a causal relationship be? (4)

Answer 7

• functional
  e. g. no contraception –> teenage mother
• empirical
  e. g. based on previous statical analysis
• theoretical
  e. g. teenage grandmother -> teenage mother
• speculative
  e. g. teenage grandfather -> teenage mother

Question 8

When is a causal relationship only possible? (2)

Answer 8

• if the cause variable precedes the effect

- assumed to be present between variables that are temporally distributed