Dear all,

I will please like to find out the explanation for the issue below.

I have a cross sectional model which includes a family ID (family-id) because there are participants who are siblings in the study. For my analysis i drop those without siblings and estimate a family fixed effects using the code below. (m_famrnd )

xtset family-id
xtreg..............., fe

Conditional fixed-effects logistic regression Number of obs = 236
Group variable: family-id Number of groups = 106
Obs per group:
min = 2
avg = 2.2
max = 4
LR chi2(6) = 24.99
Log likelihood = -70.920178 Prob > chi2 = 0.0003
noenjoy Coef. Std. Err. z P>z [95% Conf. Interval]
bmi -.0365734 .0357937 -1.02 0.307 -.1067278 .0335811

I will like to know if the effect i find with this analysis is the effect within siblings in each family (I.e result is for sibling a compared to sibling b while holding unobserved family variables constant) or if it is between the different families (i.e result is for family a compared to the other families holding unobserved family variables constant). I will really appreciate some clarification on this.

I will also like to find out what is the difference between "within sibling analysis" and "between sibling analysis" as i saw this in a paper but not sure i understand. Thanks.