Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: interpreting -test, accumulate- output


From   Tom Moliterno <moliterno@moore.sc.edu>
To   statalist@hsphsun2.harvard.edu
Subject   st: interpreting -test, accumulate- output
Date   Wed, 17 Feb 2010 16:00:24 -0500

Hi Statalisters,

Hope I could get some help interpreting output from a test command,
using the accumulate option.  All the searches I've done make it seem
like it's straightforward, but I'm a bit puzzled ...

First here's the model:  I'll give you just the results for the
variables I'm interested in.  It's an -xtreg, fe-

rY_RDistSAQ1 |  -.5356885   .2935574    -1.82   0.069    -1.114338    .0429612
rY_RDistSAQ2 |  -.6776527   .2659942    -2.55   0.012    -1.201971   -.1533346
rY_RDistSAQ3 |  -.2888348   .3427794    -0.84   0.400    -.9645092    .3868396
rY_RDistSAQ4 |  -.8127006   .4679536    -1.74   0.084    -1.735114    .1097126

So let's call these var1-var4, using the last number of the variable
names. As a side bar, these are 4 linear splines from a continuous
variable made using the -mkspline- command.

Now my objective is to be able to interpret the relationship between
these coefficients.  Obviously, var2 is sig at p<0.05, and var4 is
marginally sig at p<0.10.  But what more can I say ... so I ran
-test-:

. test (rY_RDistSAQ1-rY_RDistSAQ2)=0

 ( 1)  rY_RDistSAQ1 - rY_RDistSAQ2 = 0

       F(  1,   213) =    0.16
            Prob > F =    0.6886

So I interpret this to say that there is not a significant difference
between the coefficient for var1 and var2.  (right?)

Now ... I ran the test command using the accumulate option ... and
this is what I'm not sure how to interpret.  Here is the output:

.         foreach var in rY_RDistSAQ1 rY_RDistSAQ3 rY_RDistSAQ4 rY_RDistSAQ2{
  2.         test `var', accumulate
  3.         }

 ( 1)  rY_RDistSAQ1 = 0

       F(  1,   213) =    3.33
            Prob > F =    0.0694

 ( 1)  rY_RDistSAQ1 = 0
 ( 2)  rY_RDistSAQ3 = 0

       F(  2,   213) =    1.68
            Prob > F =    0.1897

 ( 1)  rY_RDistSAQ1 = 0
 ( 2)  rY_RDistSAQ3 = 0
 ( 3)  rY_RDistSAQ4 = 0

       F(  3,   213) =    1.85
            Prob > F =    0.1395

 ( 1)  rY_RDistSAQ1 = 0
 ( 2)  rY_RDistSAQ3 = 0
 ( 3)  rY_RDistSAQ4 = 0
 ( 4)  rY_RDistSAQ2 = 0

       F(  4,   213) =    2.44
            Prob > F =    0.0481

So do I have this right:

1st iteration --> var1 (marginally) improves model fit
2nd iteration --> adding var3 doesn't improve model fit, conditioned
on having var1 in the model
3rd iteration --> adding var 4 doesn't improve model fit, conditioned
on having var1 and var3 in the model
4th iteration --> adding var2 DOES improve model fit, conditioned on
the other three vars being in the model

Is that right?  Is there anything else interesting I can say about
that last iteration? I'm theoretically interested in var2  ...  I'm
just not sure what the F-test is describing, exactly, in that last
iteration.

Any help would be most appreciated!

Tom
-
**********************************************************
Thomas P. Moliterno, PhD
Moore School of Business
University of South Carolina
moliterno@moore.sc.edu
**********************************************************
"The way to succeed is to double your error rate."
        -- Thomas J. Watson
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index