Testing the joint significance of the mean with clustering option

[Yiling Xu]

I am replicating one of Krueger's papers.

I would like to test whether the mean of freelunchk given three groups ctypek==1/2/3 are jointly equal to 0, with the clustering option (assume freelunchk has the same standard deviation with the same school schid). I try reg with cluster option, estout command, ttest for several days, but that does not work. Could you please help me solve the problem? Thank you!

[Carlo Lazzaro]

Yiling:
why not going -regress-? (please note that in the following toy-example clustered standard errors are totally misleading, as they are way below the 30 rule of thumb threshold).

. sysuse auto.dta
. regress weight i.rep78 price, vce(cluster foreign)

Linear regression                               Number of obs     =         69
                                                F(0, 1)           =          .
                                                Prob > F          =          .
                                                R-squared         =     0.4895
                                                Root MSE          =     588.53

                                (Std. err. adjusted for 2 clusters in foreign)
------------------------------------------------------------------------------
             |               Robust
      weight | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       rep78 |
          2  |   50.37226   14.41343     3.49   0.177    -132.7678    233.5123
          3  |  -71.28614   238.7848    -0.30   0.815    -3105.334    2962.762
          4  |   -448.434   732.0831    -0.61   0.650    -9750.432    8853.564
          5  |  -972.7328   29.62635   -32.83   0.019    -1349.171   -596.2943
             |
       price |   .1449463   .0102724    14.11   0.045     .0144233    .2754692
       _cons |   2438.393   46.88828    52.00   0.012     1842.621    3034.165
------------------------------------------------------------------------------

. mat list e(b)

e(b)[1,7]
            1b.          2.          3.          4.          5.                        
         rep78       rep78       rep78       rep78       rep78       price       _cons
y1           0   50.372265  -71.286141  -448.43403  -972.73277   .14494627   2438.3928

. test 2.rep78=3.rep78=4.rep78

 ( 1)  2.rep78 - 3.rep78 = 0
 ( 2)  2.rep78 - 4.rep78 = 0
       Constraint 1 dropped

       F(  1,     1) =    0.48
            Prob > F =    0.6133

. regress weight i.rep78 price

      Source |       SS           df       MS      Number of obs   =        69
-------------+----------------------------------   F(5, 63)        =     12.08
       Model |  20924760.3         5  4184952.07   Prob > F        =    0.0000
    Residual |  21820955.6        63  346364.375   R-squared       =    0.4895
-------------+----------------------------------   Adj R-squared   =    0.4490
       Total |  42745715.9        68   628613.47   Root MSE        =    588.53

------------------------------------------------------------------------------
      weight | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
       rep78 |
          2  |   50.37226   466.5591     0.11   0.914    -881.9716    982.7161
          3  |  -71.28614   432.2574    -0.16   0.870    -935.0836    792.5113
          4  |   -448.434   440.2368    -1.02   0.312    -1328.177    431.3091
          5  |  -972.7328   453.6275    -2.14   0.036    -1879.235   -66.23058
             |
       price |   .1449463   .0246846     5.87   0.000     .0956181    .1942745
       _cons |   2438.393   431.1349     5.66   0.000     1576.838    3299.947
------------------------------------------------------------------------------

. mat list e(b)

e(b)[1,7]
            1b.          2.          3.          4.          5.                        
         rep78       rep78       rep78       rep78       rep78       price       _cons
y1           0   50.372265  -71.286141  -448.43403  -972.73277   .14494627   2438.3928

. test 2.rep78=3.rep78=4.rep78

 ( 1)  2.rep78 - 3.rep78 = 0
 ( 2)  2.rep78 - 4.rep78 = 0

       F(  2,    63) =    2.99
            Prob > F =    0.0573

.

[Yiling Xu]

Dear Mr. Lazzaro,

I wanted to express my heartfelt thanks for your quick and helpful response. I just solved the problem. Thank you!!! Have a nice day!!!

Best,
Yiling

[Carlo Lazzaro]

Yiling:
Carlo is enough! ! Thanks.