Confusion about create matching pair id for conditional logistic regression

[Khoa Tran]

Dear Statalisters,

I use -ccmatch- creating match pair of fdz (0 1) base on condition of (size0 industry_id year) for conditional logistic but the result is different every time i resample my data with "ccmatch" . I don't know why ???

My purpose here is replicate an Empirical Research which estimate the following command: "clogit fdz prof fex re bo io so mo i.year, group(match pair)". The problem is if using match pair with above condition, year effect would be omitted (as my result). But the author's result still have year effect. Would it be more than 1:1 pair matching ???

Could it be another way to deal with my problem ??? Can anyone help me out ?

Thanks in advance,
Tran.


* Example generated by -dataex-. For more info, type help dataex
clear
input int year long id double(io mo so bo) float(size0 prof fex re fdz industry_id)
2010 1 0 .12433333333333334 0 .1125 27.19248 . . . 0 39
2011 1 .0591 .005909090909090909 0 .0591 27.428436 .17479824 .05401827 .1723082 0 39
2012 1 .1326 .014848484848484849 0 .1326 27.52621 .11754111 .035956357 .1552958 0 39
2013 1 .1326 .014848484848484849 0 .1326 27.77128 .09794068 .018998675 .1778895 1 39
2014 1 .1326 .00295454545454545 0 .1326 27.98284 .0646528 .014938015 .1543429 0 39
2015 1 .1326 .15749246242241516 0 .41659999999999997 28.30129 .052531 .016928075 .13223556 1 39
2016 1 .115 .2369147946623803 0 .5619999999999999 28.755177 .09918394 .013957932 .05956683 1 39
2017 1 .3969 .02033493114855471 0 .3969 29.15188 .13278016 .02618652 .05259333 0 39
2018 1 .5409999999999999 0 0 .5409999999999999 29.649805 .08303142 .02752975 .06201045 1 39
2019 1 .4808 0 0 .4808 29.70889 .10928367 .02993978 .035310697 0 39
2020 1 .5064 0 0 .5064 29.77922 .06510147 .02519532 .07545406 1 39
2010 2 .0676 .07658178263866305 0 .0676 26.5264 . . . 0 49
2011 2 0 .09501530876170586 0 .050199999999999995 26.52229 .2400468 .003060455 .1044411 0 49
2012 2 0 .06585358288047127 0 .0774 26.37842 .05504352 .0010283291 .1554744 0 49
2013 2 0 .07897480006695048 0 .060899999999999996 26.43393 .03954194 .001292198 .1150986 0 49
2014 2 0 .19658683368189117 0 .3501000000000001 26.41764 .03988202 .0011047537 .07581282 0 49
2015 2 0 .19658683368189117 0 .3501000000000001 26.49761 .01396251 .003519217 .07387918 0 49
2016 2 0 .19658683368189117 0 .3501000000000001 26.27585 .008206352 .0022916228 .07994727 0 49
2017 2 0 .1965958919254917 0 .3501000000000001 26.25095 .004557411 0 .08166952 0 49
2018 2 0 .24235944070094195 0 .3501000000000001 26.16117 .04734994 0 .06774298 0 49
2019 2 0 .48088886022652744 0 .4669 26.15045 .04529053 .0006513707 .06797796 0 49
2020 2 0 .0002841066206674039 0 .4669 26.074266 -.05177591 .0009489534 .067313254 0 49
2018 3 0 .2127 0 .39 26.689804 . . . 0 2
2019 3 0 .14419609652656898 0 .39 27.10124 .1124318 .01084829 .09985244 0 2
2020 3 0 .0014419609652656898 0 .39 27.121395 .03760833 .00742689 .06161167 0 2
2020 4 0 .000026709492310510776 0 .369 27.788107 . . . 0 2
2010 5 .3256 .03594462846049157 0 .43940000000000007 27.1234 . . . 0 49
2011 5 .4606 .04538701410899381 0 .639 26.893105 .1890139 .011581004 .00570925 0 49
2012 5 .1998 .03883473800392779 0 .4296 26.96963 .18843456 .008403406 .05922803 0 49
2013 5 .546 .06380888081703928 0 .5986 27.257446 .17536676 .007920567 .04169218 0 49
2014 5 .6325999999999999 .06349646483706868 0 .6853 27.21043 .13730961 .00591036 .04015971 0 49
2015 5 .7282 .03262586893552088 0 .7808999999999999 27.09429 .12323866 .004882585 .04185397 0 49
2016 5 .7282 .029374658668585036 0 .8083999999999999 27.199125 .1037067 .01109398 .03029851 0 49
2017 5 .7282 .0037511556017230897 0 .836 27.17586 .06413738 .013248065 .03781279 0 49
2018 5 .7833 .0039251101371396676 0 .8911 27.00655 .10141826 .014534102 .020440934 0 49
2019 5 .7833 .002008131156848977 0 .8911 27.01198 .10343378 .008755106 .05325864 0 49
2020 5 .7833 .00002008131156848977 0 .8911 27.02162 .0358118 .00541918 .04083129 0 49
2010 6 .7337 .0031091 .7337 .7337 26.20442 . . . 0 45
2011 6 .7337 .0042366999999999995 .7337 .7337 26.26069 .26046768 .000020908563 .18109457 0 45
2012 6 .7337 .0042366999999999995 .7337 .7337 26.347 .25834608 0 .20146476 0 45
2013 6 .7846000000000001 .0054691 .7337 .7846000000000001 26.234556 .13762811 0 .1944649 0 45
2014 6 .8106 .0054691 .7337 .8106 26.54046 .1779231 .00004135377 .15916114 0 45
2015 6 .8336 .0043416 .7337 .8336 26.559736 .1416133 .0010122565 .10875983 0 45
2016 6 .7337 .0030816 .7337 .7337 26.49794 .16146423 .003064701 .113166 0 45
2017 6 .7337 .0030816 .7337 .7337 26.531044 .10660448 .003179976 .08433916 0 45
2018 6 .7337 .0030816 0 .7337 26.62876 .12960026 .003237453 .08694538 0 45
2019 6 .249 .0007100000000000001 0 .7337 26.931896 .1618658 .005824494 .0949239 0 45
2020 6 .249 1.6666666666666667e-08 0 .7137 27.540007 .18501642 .026467754 .09798332 0 45
2010 7 0 .1626363636363636 0 .3072 27.31093 . . . 0 49
2011 7 0 .2206259621433531 0 .5081 27.399565 .2267009 .06066293 .11392295 0 49
2012 7 0 .22209332501796908 0 .39370000000000005 27.4468 .05644162 .03449485 .09445897 0 49
2013 7 0 .22209332501796908 0 .4509 27.30107 .03898083 .03241798 .06167464 0 49
2014 7 0 .274889366252393 0 .5609000000000001 27.453497 .04736602 .027770184 .06699489 1 49
2015 7 0 .30062351644798074 0 .5609000000000001 27.821413 .0815478 .04524821 .07493622 1 49
2016 7 0 .5672404979456944 0 .4991 27.80082 .066817105 .04697273 .06772606 1 49
2017 7 0 .5705428695803778 0 .5024000000000001 27.726 .06064828 .04106065 .07930972 1 49
2018 7 0 .5705107199992983 0 .5024000000000001 27.92621 .27369255 .03813255 .10509437 0 49
2019 7 0 .5450964980860473 0 .5024000000000001 28.01027 .1496834 .032147214 .22359025 0 49
2020 7 0 .0006811255618854906 0 .5859 28.137564 .05187015 .03059837 .2791422 1 49
2015 8 0 .19607843137254904 0 .22210000000000002 27.20748 . . . 0 24
2016 8 0 .12398039215686275 0 .22210000000000002 27.24907 .02744273 .012351288 .07165586 1 24
2017 8 0 .06182352941176471 0 .0981 27.19171 -.02871546 .011334904 .03981722 1 24
2018 8 0 .01223529411764706 0 .0981 27.06779 -.12854023 0 -.000021713144 1 24
2019 8 0 .01223529411764706 0 .0981 27.086737 .016250031 .015217934 -.14552215 1 24
2020 8 0 .0001223529411764706 0 .0981 27.00518 -.07953846 .0153993 -.1425846 1 24
2010 9 .51 .0276 .51 .51 23.990576 . . . 0 12
2011 9 .51 .0104 .51 .51 23.97393 .15687543 .017936027 .08350199 0 12
2012 9 .51 .030799999999999998 .51 .51 24.406 .1973363 .002888661 .10331335 0 12
2013 9 .374 .11000443337766712 .374 .6198 24.733335 .1534247 .0008204192 .11415566 0 12
2014 9 .374 .18040000000000003 .374 .47700000000000004 24.93854 .1462154 .00008892307 .08015082 0 12
2015 9 .374 .07736666666666667 .374 .5477 25.1292 .13142954 .000233931 .08878336 0 12
2016 9 .374 .08259542483660132 .374 .47700000000000004 25.24156 .1184007 .003376618 .12395971 0 12
2017 9 .374 .07736666666666667 .374 .5477 25.42183 .13232276 .001357133 .07915854 0 12
2018 9 .374 .0774 .374 .5477 25.490423 .12003741 .000011158285 .08575738 0 12
2019 9 .374 .07896797385620916 .374 .5477 25.582865 .12435027 0 .08504775 0 12
2020 9 .374 .0007896797385620915 .374 .5477 25.645763 .12546797 0 .08581303 0 12
2019 10 .33509999999999995 .32058 0 .6411 26.06456 . . . 0 34
2020 10 .4022 .00009031166256757585 0 .6681999999999999 26.821363 .26184383 .00981209 .20597233 0 34
2016 11 0 .28663854245605874 0 .4947 27.86626 . . . 1 47
2017 11 .08070000000000001 .227685371478522 0 .3639 28.07899 .0833243 .027177004 .05136853 1 47
2018 11 .08199999999999999 .23756083057470354 0 .355 28.169214 .06585487 .026555635 .05515828 1 47
2019 11 .0823 .23756083057470354 0 .3654 28.181154 .030353075 .0247267 .0482664 1 47
2020 11 .0593 .0014583652333765 0 .3565 28.23964 .04142813 .025217654 .02132138 1 47
2020 12 .4318 .00003991839122450788 0 .4318 29.908636 . . . 1 31
2012 13 .785 .00731868131868132 .2817 .785 27.79022 . . . 0 49
2013 13 .785 .00731868131868132 .2817 .785 27.64944 .05932568 .024616843 .07230325 0 49
2014 13 .785 .00784065934065934 .2817 .785 27.321533 .033324312 .027324757 .08279952 0 49
2015 13 .785 .00731868131868132 .2817 .785 27.39737 .11055572 .02867346 .063927285 0 49
2016 13 .7997 .0017032967032967034 .2817 .7997 27.331736 .02771803 .019304235 .10572298 0 49
2017 13 .7997 .003686813186813187 .2817 .7997 27.301195 .029648485 .01315408 .06156192 0 49
2018 13 .7997 .003038461538461539 .2817 .7997 27.03655 .0561363 .010007647 .073143125 0 49
2019 13 .7997 .003038461538461539 .2817 .7997 27.37233 .10501108 .01921686 .12642819 0 49
2020 13 .7997 .000013351648351648351 .2817 .7997 27.35475 .0591452 .019951297 .1729383 0 49
2010 14 .29 .14510005084954508 .1217 .42330000000000007 26.31561 . . . 1 39
2011 14 .29 .14510005084954508 .1217 .42330000000000007 26.22995 .030780187 .008306394 .024688063 0 39
2012 14 .29 .13329382780218557 .1217 .4397 26.1639 -.02230169 .005891602 .04843515 0 39
2013 14 .29 .16292381226071548 .1217 .4397 26.10627 .011895914 .0021995388 -.003341676 0 39
2014 14 .2922 .06202001209289091 .1226 .44300000000000006 26.124086 .008568563 .0010065596 .00627825 0 39
2015 14 .3157 .06228538962106743 .1325 .5301 26.18367 .029528776 .001195632 .01532987 0 39
2016 14 .3157 .06701716960020108 .1325 .5301 26.137606 .028306294 .0006648143 .05229416 0 39
end

[Felix Bittmann]

According to the helpfile, ccmatch relies on random matching so you will get new results each time you run it. If you want to avoid this, set a seed (help seed). Why is year omitted in your model? What is the Stata output?

[Khoa Tran]

Here is my result.
. ccmatch size industry_id year, id(id) cc(fdz)
. clogit fdz prof fex re bo io so mo i.year, group(match) nolog
note: 92 groups (92 obs) omitted because of all positive or
all negative outcomes.
note: 2012.year omitted because of no within-group variance.
note: 2013.year omitted because of no within-group variance.
note: 2014.year omitted because of no within-group variance.
note: 2015.year omitted because of no within-group variance.
note: 2016.year omitted because of no within-group variance.
note: 2017.year omitted because of no within-group variance.
note: 2018.year omitted because of no within-group variance.
note: 2019.year omitted because of no within-group variance.
note: 2020.year omitted because of no within-group variance.

Conditional (fixed-effects) logistic regression Number of obs = 1,986
LR chi2(7) = 739.41
Prob > chi2 = 0.0000
Log likelihood = -318.59142 Pseudo R2 = 0.5371

------------------------------------------------------------------------------
fdz | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
prof | -24.95285 1.982146 -12.59 0.000 -28.83778 -21.06791
fex | 74.81103 6.284775 11.90 0.000 62.4931 87.12896
re | -1.22927 .790169 -1.56 0.120 -2.777972 .319433
bo | .3445356 .544997 0.63 0.527 -.7236388 1.41271
io | -1.718957 .523544 -3.28 0.001 -2.745084 -.6928296
so | 1.239982 .4427962 2.80 0.005 .372117 2.107846
mo | -2.030151 .7658263 -2.65 0.008 -3.531143 -.5291588
|
year |
2012 | 0 (omitted)
2013 | 0 (omitted)
2014 | 0 (omitted)
2015 | 0 (omitted)
2016 | 0 (omitted)
2017 | 0 (omitted)
2018 | 0 (omitted)
2019 | 0 (omitted)
2020 | 0 (omitted)
------------------------------------------------------------------------------

[Clyde Schechter]

In your -ccmatch- command you told Stata to match on year. Having done that, year is constant within each matched pair, so -clogit- correctly discovers that the year effects are not estimable. If you need to estimate year effects, you can't match on it. (That's not a peculiarity of -clogit- or Stata, it's linear algebra and there is no way around it. More generally, in matched analyses, you cannot estimate the effects of the matching variables.)