I'm helping with a study that is looking at gestational cytokine levels of children with neurodevelopmental disorders versus neurotypical children. I've been given a dataset looking at sibling pairs - one has a neurodevelopmental disorder, the other is neurotypical, and we have gestational samples for each. Someone else is looking at the non-sibling paired population data and is running logistic regression (adjusting for covariates that may influence the data, such as maternal weight at the time of blood draw, maternal age, etc.). I initially ran a paired t-test on the sibling data to see if there were differences in the sibling pair, but the PIs want me to run linear regression on the sibling dataset instead.

My question is: how do I run linear regression on this dataset without losing the sibling pair information? Someone briefly mentioned doing something that involved adjusting the intercept for each sibling pair, but I have no idea how to do that.

I'll be using SPSS to analyze this data as well - not sure if that matters (I'm relatively new to stats that aren't ANOVAs or ttests, so please don't judge my naivety).