Testing for Attrition Bias

[Zelina Stevens]

Hello. I am estimating a SEM (path analysis) using two waves of data. How can I test for attrition bias between the two waves? Thank you.

[Felix Bittmann]

You can start with a graphical analysis, assuming your data are in long format:

twoway (kdensity var if wave == 1) (kdensity var if wave ==2)

If the shapes are similar, this means that no bias is present.

You can also conduct a t-test:

ttest var, by(wave)

To amend the issue of attrition bias using SEM you can apply FIML:

sem ..., method(mlmv).

For more information, please share data and code, see https://www.statalist.org/forums/for...-for-statalist

[Zelina Stevens]

Thank you very much!