Integrate different types of data from different studies?

[GN1]

Hi,

Is it possible to integrate different types of data from different studies? If yes, what precautions we must take?

For example, let us assume there are 3 research papers.
Research paper 1 gives information about variable X
Research paper 2 gives information about variable Y
Research paper 3 gives information about variable Z

I am interested in combined effect of these 3 variables.

Can someone help me please?

Thanks!

 

[mmercker]

I think this is pretty difficult, since (I guess that) each paper considers only the main effect of the corresponding variable. But these variables can be correlated or there could be interactictions between these variables, i.e., effects are not additive.

In an extreme case for instance, all three variables could result from the same underlying factor, which would mean that all three variables are highly correlated and the effect of X + Y + Z equals to the effect of only X, Y, or Z.

To evaluate this, one have to perform the analysis e.g. with a regression model with all three variables (and their interactions) simultaneously, i.e.:

outcome ~ X + Y + Z + X:Y + Y:Z + X:Z + X:Y:Z,