Many, many thanks, Fugue!! It worked very well with my data! I have over 25000 observations, and actually I just have (for now), 2 ethnicities as well! But in the future I will have at least 4 more and more observations. This query did not take too long to run in my machine (about 3 minutes), which is also windows-based but in mac (partition), so I assume that will take a bit longer with more ethnicity groups but it's fine! Thank you again! ... View more

Original data looks like this: cluster Year Measurement Ethnicity Individual height income 1 2000 1 A 1 1 2000 1 A 2 1 2000 1 A 3 1 2000 1 B 1 1 2000 1 B 2 1 2000 1 B 3 1 2000 1 B 4 1 2000 1 B 5 Resulting table should look like this: Cluster Year Measurement Ethnicity Individual height weight income Sum of weights of individuals of the ethnicity A larger than the subject individual Sum of weights of individuals of the ethnicity B larger than the subject individual 1 2000 1 A 1 1 2000 1 A 2 1 2000 1 A 3 1 2000 1 A 4 1 2000 1 A 5 1 2000 1 B 1 1 2000 1 B 2 1 2000 1 B 3 1 2000 1 B 4 It would be the same data, all cells populated. In this case, it would be the sum of all heights, weights, etc, of all individuals of ethnicity A and ethnicity B (separate) larger than the individual 1, 2, 3, etc, for that specifc measurement, year and cluster; that is, comparison of individual 1 with ethnicity A, in the measurement 1, year 2000 and cluster 1, with heights/weights, etc of individuals of ethnicity A and with individuals of ethnicity B fo that specific cluster, year and measurement. Does that make sense? Thank you! ... View more

Yes, I did get it. Thank you! ... View more

Thank you very much, Fugue!! It worked well for my data set! Missing data shouldn't be a problem. One more question. I have an additional classification variable (in the same cluster, year, and measurement), let's say ethnicity or race, and I want again to obtain the sum of weights, heights, etc for those individuals of ethnicity 'x' and individuals of ethnicity 'y' (separate) larger than my subject individual, is there any way to modify this code in order to do that? I really appreciate your help! ... View more

Hi Fugue! Thanks for both replies! I was going to try with a 'retain' statement. I am going to try with the code you posted, hopefull it will work with my data set, which has about 25,000 individual observations. Thank you very much again! ... View more

Hi Fugue, 1. All variables (cluster, year, measurement) are populated for every individual (row) in the data set, I just did not do it in the example. 2. Yes, I only want individuals from cluster=1, year=2000, and measurement=1 that are greater than the specified individual. 3. I would like to do both, a many-to-many and a many-to-one; however, I want to use this info in an 'individual regression model' in which I will use the sum of heights/weights of individuals greater than the subject as an independent variable, to explain variation in other attributes. Thanks for your help! ... View more