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 at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: Why do I get two different results from the same specification and the same dataset?


From   Richard Williams <richardwilliams.ndu@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Why do I get two different results from the same specification and the same dataset?
Date   Sun, 06 Nov 2011 12:03:42 -0500

Like everyone else, I strongly suspect you computed the interactions wrong. Maybe you accidentally used a wrong variable in the calculation. I would especially double-check the calculation of the units interaction, as it is by far and away the most different coefficient in your two models: 128,569.4 in the first model (and highly significant) and -.2338972 (and insignificant) in the 2nd. Also the main effect of units switches from large positive to large negative.

If you are convinced the coding is right I might try rescaling bid_win, e.g. dividing by 1000. Their might be some sort of precision issue.

At 07:24 AM 11/6/2011, Yuval Arbel wrote:
Dear statalist participants,

when I run the following regression

reg bid_win dev_cost bid_num year area units min min_price
c.dev_cost#i.min c.bid_num#i.min c.year#i.min c.area#i.min
c.units#i.min

I get the following output:


Source | SS df MS Number of obs = 6802
-------------+------------------------------           F( 12,  6789) = 2891.19
Model | 7.0107e+17 12 5.8423e+16 Prob > F = 0.0000 Residual | 1.3719e+17 6789 2.0207e+13 R-squared = 0.8363
-------------+------------------------------           Adj R-squared =  0.8361
Total | 8.3826e+17 6801 1.2326e+14 Root MSE = 4.5e+06

------------------------------------------------------------------------------
bid_win | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
dev_cost | -.0782451 .1319286 -0.59 0.553 -.3368666 .1803764 bid_num | 53637.98 18087.12 2.97 0.003 18181.54 89094.41 year | 61991.65 44544.85 1.39 0.164 -25330.21 149313.5 area | 204.3705 105.0742 1.95 0.052 -1.607834 410.3488 units | 52691.04 11756.39 4.48 0.000 29644.82 75737.26 min | 1.04e+08 9.74e+07 1.06 0.288 -8.73e+07 2.95e+08 min_price | 3.956053 .0241168 164.04 0.000 3.908777 4.00333
             |
         min#|
  c.dev_cost |
1 | -.460682 .1391518 -3.31 0.001 -.7334631 -.187901
             |
         min#|
   c.bid_num |
1 | -43194.68 19552.09 -2.21 0.027 -81522.9 -4866.457
             |
  min#c.year |
1 | -51639.35 48543.57 -1.06 0.287 -146800 43521.27
             |
  min#c.area |
1 | 186.6343 110.1857 1.69 0.090 -29.36416 402.6327
             |
 min#c.units |
1 | -128569.4 12642.23 -10.17 0.000 -153352.2 -103786.7
             |
_cons | -1.25e+08 8.94e+07 -1.39 0.163 -3.00e+08 5.05e+07
------------------------------------------------------------------------------


But when I define directly the interaction variables, and run the
regression, I get different outcomes:

. reg bid_win dev_cost bid_num year area units min min_price
dev_cost_int bid_num_int year_int area_int units_int

Source | SS df MS Number of obs = 6802
-------------+------------------------------           F( 12,  6789) = 2840.90
Model | 6.9905e+17 12 5.8254e+16 Prob > F = 0.0000 Residual | 1.3921e+17 6789 2.0505e+13 R-squared = 0.8339
-------------+------------------------------           Adj R-squared =  0.8336
Total | 8.3826e+17 6801 1.2326e+14 Root MSE = 4.5e+06

------------------------------------------------------------------------------
bid_win | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
dev_cost | .3458744 .1259233 2.75 0.006 .0990254 .5927235 bid_num | 50612.77 18218.04 2.78 0.005 14899.71 86325.84 year | 26138.3 44731.32 0.58 0.559 -61549.1 113825.7 area | 796.392 88.98841 8.95 0.000 621.9468 970.8371 units | -56322.78 4522.886 -12.45 0.000 -65189.06 -47456.51 min | 2.11e+07 9.78e+07 0.22 0.829 -1.71e+08 2.13e+08 min_price | 3.914549 .0241269 162.25 0.000 3.867252 3.961845
dev_cost_int |  -.9575921   .1316807    -7.27   0.000    -1.215728   -.6994567
bid_num_int | -40191.13 19694.51 -2.04 0.041 -78798.54 -1583.728 year_int | -10450.87 48755.43 -0.21 0.830 -106026.8 85125.05 area_int | -443.5883 91.15801 -4.87 0.000 -622.2866 -264.89 units_int | -.2338972 .131622 -1.78 0.076 -.4919176 .0241233 _cons | -5.29e+07 8.98e+07 -0.59 0.556 -2.29e+08 1.23e+08
------------------------------------------------------------------------------

My question is why do I get two different results from the same specification?
Just to exemplify: note that the coefficient of "dev_cost" has
modified signs and became significant

--
Dr. Yuval Arbel
School of Business
Carmel Academic Center
4 Shaar Palmer Street, Haifa, Israel
e-mail: yuval.arbel@gmail.com
*
*   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/

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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