I am having trouble coding this problem. Thanks in advance to anyone who can help! Let's say we have individual-level data. The dataset has families that live in neighborhoods so the dataset would look something like this: Have: ID Fam_ID Neigh_ID Income 1 A CC 100,000 2 A CC 200,000 3 B CC 60,000 4 B CC 200,000 5 B CC 120,000 6 C DD 35,000 7 C DD 40,000   I am trying to find the median income of a neighborhood for each family that excludes the income of those in said family.  For example, family A would have a med_neighborhood_income of 120,000 since the only other family in neighborhood CC is family B and the median income in family B is 120,000. The med_neighborhood_income value for family B would be ((100,000+200,000)/2) or 150,000.  Hope this is clear.    ... View more

Stratified proc phreg seems to only use discordant pairs within strata as informative data points for estimation. Is there an option to tag the data used in the analyses? I want to see what data is actually providing information for the output. The observations used is misleading because SAS says it "uses" the whole dataset, when I think only discordant pairs are informative. ... View more