Interpretation of -pstest- after -psmatch2-

[Man Yang]

Hi folks, I am using -psmatch2- to analyze a data and I was able to get the results with no question. However, I need a little bit of help interpreting the results I got after -pstest-. Below is what I got from the output. I am looking at the continous variable base_funskl after matching. People said that pstest will give you standardized mean difference but which one is it? Thanks!

. pstest base_funskl, both treated(_treated) graph label 

----------------------------------------------------------------------------------------
                Unmatched |       Mean               %reduct |     t-test    |  V(T)/
Variable          Matched | Treated Control    %bias  |bias| |    t    p>|t| |  V(C)
--------------------------+----------------------------------+---------------+----------
base_funskl            U  | 9.3652   8.8219     18.1         |   2.75  0.006 |  0.97
                       M  | 9.3406   9.5826     -8.1    55.5 |  -0.99  0.321 |  0.85
                          |                                  |               |
----------------------------------------------------------------------------------------
* if variance ratio outside [0.80; 1.24] for U and [0.80; 1.25] for M

-----------------------------------------------------------------------------------
 Sample    | Ps R2   LR chi2   p>chi2   MeanBias   MedBias      B      R     %Var
-----------+-----------------------------------------------------------------------
 Unmatched | 0.006      7.53    0.006     18.1      18.1      18.1    0.97      0
 Matched   | 0.001      0.99    0.320      8.1       8.1       7.8    0.85      0
-----------------------------------------------------------------------------------
* if B>25%, R outside [0.5; 2]

[David Radwin]

It is "%bias". Since you have only one variable, it is also reported in "MeanBias" and "MedBias".

There is more than one measure of standardized difference. An alternative is the difference in sample means or proportions of the two groups expressed as a percentage of the square root of the mean of the sample standard deviations of the two groups (Ho et al. 2007; Stuart and Rubin 2008).

Ho, D.E., Imai, K., King, G., and Stuart, E.A. (2007). Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference. Political Analysis, 15(3): 199–236.

Stuart, E.A., and Rubin, D.B. (2008). Best Practices in Quasi-Experimental Designs: Matching Methods for Causal Inference. In J.W. Osborne (Ed.), Best Practices in Quantitative Methods (pp. 155–176). Thousand Oaks, CA: Sage.

[Man Yang]

Hi David, thanks a lot for your response.