Propensity score matching - ATT, ATU, ATE

[Tracy Lam]

Hi there --

I am trying to perform propensity score matching for two groups (treatment and control) on reading achievement. My code is:

 psmatch2 trt rd2, outcome(rd3) logit caliper(0.01) noreplacement neighbor(1) ate

My questions:

1. Is the above the correct code if I want to match the two groups on reading achievement?

2. If so, how is Stata calculating ATU and ATE? I understand that for ATT it is treatment - control for the difference, but why does it calculate ATU as control - treatment? Also, why are the standard errors missing for ATU and ATE?


----------------------------------------------------------------------------------------
Variable Sample | Treated Controls Difference S.E. T-stat
----------------------------+-----------------------------------------------------------
rd3 Unmatched | .020863288 -.153141158 .174004445 .070884959 2.45
ATT |-.233408342 .073888208 -.30729655 .074629273 -4.12
ATU | .031247096 -.212995888 -.244242984 . .
ATE | -.275003491 . .
----------------------------+-----------------------------------------------------------


3. I want to the plot the means of the test (rd) for time 1, 2, and 3. Can I use the following to do that:

. pstest rd1 rd2 rd3

--------------------------------------------------------------------
| Mean | t-test
Variable | Treated Control %bias | t p>|t|
------------------------+--------------------------+----------------
rd1 |-.25352 .30533 -55.3 | -6.90 0.000
rd2 |-.15621 -.03315 -12.4 | -1.75 0.081
rd3 |-.23341 .14141 -37.9 | -4.84 0.000
--------------------------------------------------------------------

Any help is greatly appreciated!!