Determining matching units to evaluate program

Question

I introduced some treatment to a school (let’s say PS142) and I want to know if my treatment improved students’ math scores compared to the control group. PS142 has 10% minority students and 15 % low income students.

NYC department of education lists 30 peer schools that closely match PS142 on specific characteristics (in terms of % minority and % low income). I don’t want to compare PS142 to all 30 peer schools. I want to just choose 5 or 6 schools from the list of 30 schools. How do I do this? What would be the best method?

Why is it difficult to obtain scores from these schools? – AdamO Nov 18 '15 at 20:50 — AdamO, Nov 18 '15 at 20:50

score 0 · Answer 1 · answered Nov 18 '15 at 20:10

0

Why not just compare it to the mean of the other schools (using a t-test)? You could first look for and omit any outliers from the control group perhaps.

answered Nov 18 '15 at 20:10

DaGu

31
4

Thanks for your response DaGu. Yes I thought about making comparison based on the mean of other schools. The reason I want to just choose 5 or 6 schools from the list is to reduce workload/save time. I want to look up stats for just a few schools vs. 30 schools ( I have a lot of analyses to run!). nj – Nora Jones Nov 18 '15 at 20:21

Determining matching units to evaluate program

1 Answers1