Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: odd grqreg behavior
From
Jeffrey T Grogger <[email protected]>
To
"[email protected]" <[email protected]>
Subject
Re: st: odd grqreg behavior
Date
Tue, 15 Mar 2011 11:58:20 -0500
Many thanks. The author provided me with a workaround, but I know he
wants to fix the problem, so I'm copying him on this.
On 3/15/11 11:40 AM, "Scott Merryman" <[email protected]> wrote:
>This problem can be replicated with
>
>sysuse auto,clear
>gen w2= weight/10000
>xi i.rep
>sqreg mpg w2 _I*,quantile(.1 .2 .3 .4 .5 .6 .7 .8 .9) reps(10) level(95)
>grqreg w2, ci ols save(grtest) level(95)
>use grtest,clear
>l qtile w2*
>
>The problem appears to be limitation in the wordcount() function that
>is limiting the number variables -grqreg- uses. In the above example
>the results in grtest.dta are for
>
>sqreg mpg w2 _Irep78_2, q(.1 .2 .3 .4 .5 .6 .7 .8 .9)
>
>
>The best thing would be to contact the author, Joao Pedro Azevedo,
>but in the mean time if you alter lines 39/40 of _bsqregtrace.ado and
>lines 102/103 of grqreg.ado from
>
> loc tmp0 = wordcount("`e(eqnames)'")
> loc tmp1 = wordcount("`rhs'")
>
>to
>
> loc tmp0 : word count `e(eqnames)'
> loc tmp1 : word count `rhs'
>
>this should bypass the wordcount() limitation.
>
>Scott
>
>
>On Sun, Mar 13, 2011 at 4:42 PM, Jeffrey T Grogger
><[email protected]> wrote:
>> Dear all:
>> I am trying to estimate quantile treatment effects using sqreg, then
>>plot them using grqreg. To wit:
>>
>> set seed 1234567
>> sqreg allbut_06 treatment _Icl*,quantile(.1 .2 .3 .4 .5 .6 .7 .8 .9)
>>reps(10) level(95)
>> grqreg treatment, ci ols save(grtest) level(95)
>>
>> sqreg produces estimates, and grqreg plots something, but grqreg does
>>not produce the estimates computed by sqreg. The log file is below,
>>edited for clarity. The bottom line is that the numbers that grqreg is
>>plotting (in grtest.dta, at bottom of log) are not the coefficients that
>>sqreg is producing. Any idea what is happening?
>> Thanks.
>>
>>-------------------------------------------------------------------------
>>--------------------------
>> . xi i.clust_match
>> i.clust_match _Iclust_mat_1-50 (naturally coded; _Iclust_mat_1
>>omitted)
>>
>> . set seed 1234567
>>
>> . sqreg allbut_06 treatment _Icl*,quantile(.1 .2 .3 .4 .5 .6 .7 .8 .9)
>>reps(10) level(95)
>> (fitting base model)
>> (bootstrapping ..........)
>>
>> Simultaneous quantile regression Number of obs =
>> 28048
>> bootstrap(10) SEs .10 Pseudo R2 =
>>0.0691
>> .20 Pseudo R2 =
>>0.0816
>> .30 Pseudo R2 =
>>0.0889
>> .40 Pseudo R2 =
>>0.0949
>> .50 Pseudo R2 =
>>0.0980
>> .60 Pseudo R2 =
>>0.0999
>> .70 Pseudo R2 =
>>0.0996
>> .80 Pseudo R2 =
>>0.0978
>> .90 Pseudo R2 =
>>0.1022
>>
>>
>>-------------------------------------------------------------------------
>>-----
>> | Bootstrap
>> allbut_06 | Coef. Std. Err. t P>|t| [95% Conf.
>>Interval]
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q10 |
>> treatment | 185.5 44.24246 4.19 0.000 98.78263
>>272.2174
>> _Iclust_m~_2 | -815.5 205.061 -3.98 0.000 -1217.43
>>-413.5704
>> .
>> .
>> .
>> _Iclust_m~50 | 714.5 311.8173 2.29 0.022 103.3228
>>1325.677
>> _cons | 3560.5 178.9403 19.90 0.000 3209.768
>>3911.232
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q20 |
>> treatment | 195 46.32869 4.21 0.000 104.1935
>>285.8065
>> _Iclust_m~_2 | -1458.5 217.0275 -6.72 0.000 -1883.885
>>-1033.115
>> .
>> .
>> .
>> _Iclust_m~50 | 387.5 331.0245 1.17 0.242 -261.3241
>>1036.324
>> _cons | 5167.5 259.136 19.94 0.000 4659.581
>>5675.419
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q30 |
>> treatment | 152 48.88524 3.11 0.002 56.18255
>>247.8175
>> _Iclust_m~_2 | -1538 207.7434 -7.40 0.000 -1945.187
>>-1130.813
>> .
>> .
>> .
>> _Iclust_m~50 | 162 415.5631 0.39 0.697 -652.5239
>>976.5239
>> _cons | 6488 192.1978 33.76 0.000 6111.283
>>6864.717
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q40 |
>> treatment | 150 54.50545 2.75 0.006 43.16666
>>256.8333
>> _Iclust_m~_2 | -1771 348.4914 -5.08 0.000 -2454.06
>>-1087.94
>> .
>> .
>> .
>> _Iclust_m~50 | -181 321.9381 -0.56 0.574 -812.0144
>>450.0144
>> _cons | 7791 188.2547 41.39 0.000 7422.012
>>8159.988
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q50 |
>> treatment | 170 74.49461 2.28 0.022 23.98693
>>316.0131
>> _Iclust_m~_2 | -1535 332.7787 -4.61 0.000 -2187.263
>>-882.7375
>> .
>> .
>> .
>> _Iclust_m~50 | 40 392.3174 0.10 0.919 -728.9611
>>808.9611
>> _cons | 8915 143.1329 62.28 0.000 8634.453
>>9195.547
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q60 |
>> treatment | 106 81.76807 1.30 0.195 -54.2694
>>266.2694
>> _Iclust_m~_2 | -1751 366.1833 -4.78 0.000 -2468.737
>>-1033.263
>> .
>> .
>> .
>> _Iclust_m~50 | 156.5 307.9239 0.51 0.611 -447.0458
>>760.0458
>> _cons | 10308.5 309.545 33.30 0.000 9701.777
>>10915.22
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q70 |
>> treatment | 130 120.192 1.08 0.279 -105.5822
>>365.5822
>> _Iclust_m~_2 | -2505 412.422 -6.07 0.000 -3313.367
>>-1696.633
>> .
>> .
>> .
>> _Iclust_m~50 | -20 442.8143 -0.05 0.964 -887.9377
>>847.9377
>> _cons | 12345 405.513 30.44 0.000 11550.17
>>13139.83
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q80 |
>> treatment | 85 142.7379 0.60 0.552 -194.7733
>>364.7733
>> _Iclust_m~_2 | -2221 354.8877 -6.26 0.000 -2916.597
>>-1525.403
>> .
>> .
>> .
>> _cons | 14371 422.0801 34.05 0.000 13543.7
>>15198.3
>>
>>-------------+-----------------------------------------------------------
>>-----
>> q90 |
>> treatment | 70 255.9348 0.27 0.784 -431.6446
>>571.6446
>> _Iclust_m~_2 | -3300 584.2882 -5.65 0.000 -4445.233
>>-2154.767
>> .
>> .
>> .
>> _cons | 18330 448.8874 40.83 0.000 17450.16
>>19209.84
>>
>>-------------------------------------------------------------------------
>>-----
>>
>> . grqreg treatment, ci ols save(grtest) level(95)
>> file grtest.dta saved
>>
>> .
>> end of do-file
>>
>> . use grtest,clear
>>
>> . list qtile treatment*
>>
>> +----------------------------------------+
>> | qtile treatm~t treatm~i treatm~o |
>> |----------------------------------------|
>> 1. | .1 318.00 401.56 234.44 |
>> 2. | .2 320.00 422.07 217.93 |
>> 3. | .3 270.00 401.98 138.02 |
>> 4. | .4 304.00 461.43 146.57 |
>> 5. | .5 285.00 457.46 112.54 |
>> |----------------------------------------|
>> 6. | .6 294.00 481.63 106.37 |
>> 7. | .7 285.00 586.44 -16.44 |
>> 8. | .8 400.00 623.65 176.35 |
>> 9. | .9 1004.50 1489.70 519.30 |
>> 10. | .05 . . . |
>> |----------------------------------------|
>> 11. | .15 . . . |
>> 12. | .25 . . . |
>> 13. | .35 . . . |
>> 14. | .45 . . . |
>> 15. | .55 . . . |
>> |----------------------------------------|
>> 16. | .65 . . . |
>> 17. | .75 . . . |
>> 18. | .85 . . . |
>> --Break--
>> r(1);
>
>*
>* 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/
*
* 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/