Causal inferences in surveys??

[aquary]

Hi,

I am looking into the general problems of making causal inferences in surveys. There are a few points I am still confused:

a)In what ways can design of a study address these problems?
b)What analytical approaches in data analysis can aid in developing causal explanations?

Can anyone give me some suggestions/relevant reference on this pliz? thankx

 

[BioStatMatt]

Perhaps you could look at an introductory epidemiology text book. These generally deal with the requirements for a causal relationship. For example, for an action A to be causatory for event B, action A must occur before event B. This is called temporality, and cannot be deduced from data analysis, but must be built into the design of a study. There are other general rules for developing causality. An analytical aid to developing causality is the correlation of action A and event B. While this is perhaps required for most causal relationships, it should not be taken as evidence for a causal relationship (it could be that event B causes action A, or another explanation).

~Matt