From zanoli at agrecon.univpm.it Sat Feb 2 00:27:09 2019 From: zanoli at agrecon.univpm.it (Raffaele Zanoli) Date: Fri, 1 Feb 2019 14:27:09 +0100 Subject: [Limdep Nlogit List] Inconisistent RPL results Message-ID: Hello Bill, I have the following issue with NLOGIT 6. I had strange results for this estimation, and I checked if something would change in I switch the baseline from the SWQ alternative to another one (SC in my case). While the MNL exhibit the same results apart, obviously, from ASC coefficinets, I get inconsistent results in the RPL model (I send here 100 iterations but with 1000 the problem worsen). The attributes are effects coded, but dummy coding will produce basically the same inconsistencies (though obviously with different parameter values). Could you give me a hint of what could be going on? Thanks Raffaele Prof. Dr. Raffaele Zanoli Professor of Food Marketing & Management Department of Agricultural, Food and Environmental Sciences Universit? Politecnica delle Marche Via Brecce Bianche 60129 Ancona- Italy tel. +39 071 220 4929 mobile +39 3483235654 |-> RPLOGIT ; Lhs=CHOICE ; CHOICES = CA,CO,SQ ; Model: U(CA)=ascCA+BIOD10*BIOD10+BIOD15*BIOD15+CLIMA*CLIRED+SCIE*SCIE+PRICE*PRICE/ U(CO)=ascCO+BIOD10*BIOD10+BIOD15*BIOD15+CLIMA*CLIRED+SCIE*SCIE+PRICE*PRICE/ U(SQ)=BIOD10*BIOD10+BIOD15*BIOD15+CLIMA*CLIRED+SCIE*SCIE+PRICE*PRICE ; Fcn = BIOD10(n),BIOD15(n),CLIMA(n),SCIE(n) ; Halton ; Pds=9 ; Pts=100 ; WTP=biod10/price,biod15/price,clima/price,scie/price ; Par$ Iterative procedure has converged Normal exit: 5 iterations. Status=0, F= .7317761D+04 ----------------------------------------------------------------------------- Start values obtained using MNL model Dependent variable Choice Log likelihood function -7317.76051 Estimation based on N = 7200, K = 7 Inf.Cr.AIC = 14649.5 AIC/N = 2.035 --------------------------------------- Log likelihood R-sqrd R2Adj Constants only -7566.7162 .0329 .0322 Note: R-sqrd = 1 - logL/Logl(constants) --------------------------------------- Chi-squared[ 5] = 497.91129 Prob [ chi squared > value ] = .00000 Response data are given as ind. choices Number of obs.= 7200, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHOICE| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- BIOD10| .24256*** .05384 4.51 .0000 .13704 .34809 BIOD15| .32428*** .04215 7.69 .0000 .24166 .40689 CLIMA| .04615 .03355 1.38 .1689 -.01960 .11190 SCIE| .19312*** .03403 5.68 .0000 .12643 .25981 ASCCA| -.94905*** .12115 -7.83 .0000 -1.18650 -.71160 PRICE| -.02365*** .00110 -21.46 .0000 -.02581 -.02149 ASCCO| -.41455*** .11970 -3.46 .0005 -.64915 -.17995 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Feb 01, 2019 at 00:34:30 PM ----------------------------------------------------------------------------- Iterative procedure has converged Normal exit: 23 iterations. Status=0, F= .5184372D+04 ----------------------------------------------------------------------------- Random Parameters Multinom. Logit Model Dependent variable CHOICE Log likelihood function -5184.37204 Restricted log likelihood -7910.00848 Chi squared [ 11](P= .000) 5451.27287 Significance level .00000 McFadden Pseudo R-squared .3445807 Estimation based on N = 7200, K = 11 Inf.Cr.AIC = 10390.7 AIC/N = 1.443 --------------------------------------- Log likelihood R-sqrd R2Adj No coefficients -7910.0085 .3446 .3441 Constants only -7566.7162 .3148 .3143 At start values -7317.7605 .2915 .2910 Note: R-sqrd = 1 - logL/Logl(constants) --------------------------------------- Response data are given as ind. choices Replications for simulated probs. = 100 Used Halton sequences in simulations. RPL model with panel has 800 groups Fixed number of obsrvs./group= 9 Number of obs.= 7200, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHOICE| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- |Random parameters in utility functions.......................... BIOD10| -.18622 .13056 -1.43 .1538 -.44211 .06968 BIOD15| -.24311* .14080 -1.73 .0842 -.51907 .03284 CLIMA| -.06382 .04788 -1.33 .1826 -.15766 .03003 SCIE| .25900*** .04937 5.25 .0000 .16223 .35576 |Nonrandom parameters in utility functions....................... ASCCA| .99982*** .17680 5.66 .0000 .65330 1.34634 PRICE| -.03730*** .00155 -24.13 .0000 -.04033 -.03427 ASCCO| 1.69001*** .17484 9.67 .0000 1.34733 2.03269 |Distns. of RPs. Std.Devs or limits of triangular................ NsBIOD10| 1.92473*** .21870 8.80 .0000 1.49609 2.35337 NsBIOD15| 1.58946*** .17158 9.26 .0000 1.25317 1.92574 NsCLIMA| .46827*** .06076 7.71 .0000 .34918 .58736 NsSCIE| .56041*** .06751 8.30 .0000 .42810 .69272 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Feb 01, 2019 at 00:39:17 PM ----------------------------------------------------------------------------- Saved Individual Estimates of WTP in matrix WTP_I [ 800x4] Alternative Attribute Income/Cost Chosen BIOD10 PRICE Chosen BIOD15 PRICE Chosen CLIMA PRICE Chosen SCIE PRICE (Saved absolute values. Check signs of coefficients.) |-> RPLOGIT ; Lhs=CHOICE ; CHOICES = CA,CO,SQ ; Model: U(CA)= +BIOD10*BIOD10+BIOD15*BIOD15+CLIMA*CLIRED+SCIE*SCIE+PRICE*PRICE/ U(CO)=ascCO+BIOD10*BIOD10+BIOD15*BIOD15+CLIMA*CLIRED+SCIE*SCIE+PRICE*PRICE/ U(SQ)=ascSQ ; Fcn = BIOD10(n),BIOD15(n),CLIMA(n),SCIE(n) ; Halton ; Pds=9 ; Pts=100 ; WTP=biod10/price,biod15/price,clima/price,scie/price ; Par$ Iterative procedure has converged Normal exit: 5 iterations. Status=0, F= .7317761D+04 ----------------------------------------------------------------------------- Start values obtained using MNL model Dependent variable Choice Log likelihood function -7317.76051 Estimation based on N = 7200, K = 7 Inf.Cr.AIC = 14649.5 AIC/N = 2.035 --------------------------------------- Log likelihood R-sqrd R2Adj Constants only -7566.7162 .0329 .0322 Note: R-sqrd = 1 - logL/Logl(constants) --------------------------------------- Chi-squared[ 5] = 497.91129 Prob [ chi squared > value ] = .00000 Response data are given as ind. choices Number of obs.= 7200, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHOICE| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- BIOD10| .24256*** .05384 4.51 .0000 .13704 .34809 BIOD15| .32428*** .04215 7.69 .0000 .24166 .40689 CLIMA| .04615 .03355 1.38 .1689 -.01960 .11190 SCIE| .19312*** .03403 5.68 .0000 .12643 .25981 PRICE| -.02365*** .00110 -21.46 .0000 -.02581 -.02149 ASCCO| .53450*** .03395 15.74 .0000 .46796 .60105 ASCSQ| .33606*** .05236 6.42 .0000 .23343 .43868 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Feb 01, 2019 at 00:58:50 PM ----------------------------------------------------------------------------- Iterative procedure has converged Normal exit: 23 iterations. Status=0, F= .6244799D+04 ----------------------------------------------------------------------------- Random Parameters Multinom. Logit Model Dependent variable CHOICE Log likelihood function -6244.79865 Restricted log likelihood -7910.00848 Chi squared [ 11](P= .000) 3330.41966 Significance level .00000 McFadden Pseudo R-squared .2105193 Estimation based on N = 7200, K = 11 Inf.Cr.AIC = 12511.6 AIC/N = 1.738 --------------------------------------- Log likelihood R-sqrd R2Adj No coefficients -7910.0085 .2105 .2099 Constants only -7566.7162 .1747 .1741 At start values -7317.7605 .1466 .1460 Note: R-sqrd = 1 - logL/Logl(constants) --------------------------------------- Response data are given as ind. choices Replications for simulated probs. = 100 Used Halton sequences in simulations. RPL model with panel has 800 groups Fixed number of obsrvs./group= 9 Number of obs.= 7200, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHOICE| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- |Random parameters in utility functions.......................... BIOD10| .18120* .10393 1.74 .0813 -.02251 .38491 BIOD15| -.11670 .09614 -1.21 .2248 -.30512 .07173 CLIMA| -.31503*** .06763 -4.66 .0000 -.44757 -.18248 SCIE| -.18281** .07674 -2.38 .0172 -.33322 -.03240 |Nonrandom parameters in utility functions....................... PRICE| -.03892*** .00160 -24.27 .0000 -.04206 -.03578 ASCCO| .77897*** .04635 16.81 .0000 .68813 .86981 ASCSQ| -.19761*** .06608 -2.99 .0028 -.32713 -.06809 |Distns. of RPs. Std.Devs or limits of triangular................ NsBIOD10| 2.03196*** .13903 14.61 .0000 1.75945 2.30446 NsBIOD15| 1.97433*** .10474 18.85 .0000 1.76904 2.17963 NsCLIMA| 1.22276*** .07530 16.24 .0000 1.07517 1.37035 NsSCIE| 1.62494*** .08083 20.10 .0000 1.46651 1.78336 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Feb 01, 2019 at 01:03:12 PM ----------------------------------------------------------------------------- Saved Individual Estimates of WTP in matrix WTP_I [ 800x4] Alternative Attribute Income/Cost Chosen BIOD10 PRICE Chosen BIOD15 PRICE Chosen CLIMA PRICE Chosen SCIE PRICE (Saved absolute values. Check signs of coefficients.) From jwb_bos at yahoo.com Tue Feb 19 00:58:49 2019 From: jwb_bos at yahoo.com (Bos J.W.B.) Date: Mon, 18 Feb 2019 14:58:49 +0100 Subject: [Limdep Nlogit List] bootstrap LCM Message-ID: Dear all, I?m trying to bootstrap a latent class model. As a basis for my code, I use the procedure described in the manual, section R15-3. Here?s my code: REGRESS;Lhs=klspe1;Rhs=one,klopen1,klopen1q,klywork$ namelist;x=klopen1,klopen1q,klywork$ calc;k=col(x)$ matrix;bb=init(k,1,0.0)$ PROC REGRESS;Lhs=klspe1;Rhs=one,klopen1,klopen1q, klywork;LCM=X;Pds=ni;Pts=2;algs=bfgs;maxit=2000; parameters;CProb=Cprob;Group=Group;keep=fitopen$ MATRIX;bb=b$ ENDPROC EXEC;Bootstrap=bb;N=1000$ The problem with this code is that it only bootstraps the coefficients for one class. Does anyone know the solution to this problem? I guess modifying the namelist command should do the trick, but I am not sure how?. gr, Jaap Bos From wgreene at stern.nyu.edu Wed Feb 20 08:19:09 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Tue, 19 Feb 2019 16:19:09 -0500 Subject: [Limdep Nlogit List] bootstrap LCM In-Reply-To: References: Message-ID: Jaap. In your regression spec in the procedure, you have RHS=one,x and LCM=x. This is surely overspecified - the same variables appear in the equation and in the class probabilities. In the specification of the model, you have ;PDS=ni, which means that the class specification is the same from period 1 to period n1. But, with ;LCM=x, you have the class probabilities changing over time. the result should be unpredictablle. In fact, I believe that limdep is ignoring periods 2 - ni of X in computing the class probabilities. But, again, the result seems to me like it might go awry. Also, shouldn't you be using a block bootstrap? Your data are a panel. You should have pds=ni in the EXEC command. Also, I don't think you want to repead CProb=Cprob;Group=Group;keep=fitopen 1,000 times. Do this with the original, non-bootstrapped regression. Finally, you might want to define BB to have 2*K cells. Best regards, Bill On Mon, Feb 18, 2019 at 8:59 AM Bos J.W.B. via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear all, > > I?m trying to bootstrap a latent class model. As a basis for my code, I > use the procedure described in the manual, section R15-3. > > Here?s my code: > > REGRESS;Lhs=klspe1;Rhs=one,klopen1,klopen1q,klywork$ > > namelist;x=klopen1,klopen1q,klywork$ > calc;k=col(x)$ > matrix;bb=init(k,1,0.0)$ > PROC > REGRESS;Lhs=klspe1;Rhs=one,klopen1,klopen1q, > klywork;LCM=X;Pds=ni;Pts=2;algs=bfgs;maxit=2000; > parameters;CProb=Cprob;Group=Group;keep=fitopen$ > MATRIX;bb=b$ > ENDPROC > EXEC;Bootstrap=bb;N=1000$ > > The problem with this code is that it only bootstraps the coefficients for > one class. Does anyone know the solution to this problem? I guess modifying > the namelist command should do the trick, but I am not sure how?. > > gr, Jaap Bos > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > -- William Greene Department of Economics Stern School of Business, New York University 44 West 4 St., 7-90 New York, NY, 10012 URL: https://protect-au.mimecast.com/s/y5vCCXLKZoiPZMoWc6wYVv?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.212.998.0876 Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics Associate Editor: Journal of Choice Modeling From avassilopoulos.aua at gmail.com Thu Feb 21 06:07:23 2019 From: avassilopoulos.aua at gmail.com (Achilleas' gmail) Date: Wed, 20 Feb 2019 21:07:23 +0200 Subject: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters Message-ID: <002201d4c94f$83bbf680$8b33e380$@gmail.com> Dear all, I'm estimating a Random Coefficients Ordered Logit Model with Y an ordered variable and X1 and X2 independent dummy vars: SETPANEL; GROUP=ID; PDS=ROUNDS $ ORDERED ; PANEL ; Lhs = Y ; Rhs = ONE, X1, X2 ; MODEL= LOGIT $ ORDERED ; PANEL; Lhs = Y ; Rhs = one, X1, X2 ; RPM ; par ; Fcn = one(n), X1(n), X2(n) ; HALTON ; MODEL= LOGIT $ My results are as follows: |Means for random parameters Constant| 8.33434*** .85987 9.69 .0000 6.64903 10.01966 X1 | .12776 .32251 .40 .6920 -.50436 .75988 X2 | -.03393 .33030 -.10 .9182 -.68131 .61344 |Scale parameters for dists. of random parameters Constant| 3.26244*** .22798 14.31 .0000 2.81560 3.70927 X1 | .62482*** .21928 2.85 .0044 .19503 1.05460 X2 | .78015*** .22758 3.43 .0006 .33409 1.22621 My intuition is that for the X's, the results of the unconditional estimates mean that their effect is non-significant (i.e. zero) on aggregate but with strong individual variability (i.e. showing that for some individuals it is different from zero). However, when I look at the individual specific (conditional) estimates, I find a strong indication of their effect being zero for all (i.e. confidence intervals for all individuals are way far from zero on both ends) Is my interpretation mistaken? And if not, how I can possibly explain the different conclusions I get from conditional vs unconditional estimates? Thank you, Achilleas From wgreene at stern.nyu.edu Thu Feb 21 15:11:04 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Wed, 20 Feb 2019 23:11:04 -0500 Subject: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters In-Reply-To: <002201d4c94f$83bbf680$8b33e380$@gmail.com> References: <002201d4c94f$83bbf680$8b33e380$@gmail.com> Message-ID: Achilleas: Considering X1, your results suggest that the population distribution of coefficients on X1 has a normal distribution with mean of 0.12776 and a standard deviation of 0.62482. Interpretations of "statistical significance" in this context are ambiguous. The statement However, when I look at the individual specific (conditional) estimates, I find a strong indication of their effect being zero for all (i.e. confidence intervals for all individuals are way far from zero on both ends) is contradictory. It seems to state that the effects are close to zero, but confidence intervals are far from zero. What is true is that the conditional estimates will vary around the mean of 0.12276. The mean of the conditional estimates should equal the unconditional estimate. The conditional standard deviations that are estimable will all be less than the 0.62482, and the average of these conditional standard deviations will also be strictly less than 0.62482. Characterizations of the distribution of random parameters should not be based on "confidence intervals." Some of the distribution may be near zero while more of the distribution may be quite far from zero. /Bill Greene On Wed, Feb 20, 2019 at 2:07 PM Achilleas' gmail via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear all, > > > > I'm estimating a Random Coefficients Ordered Logit Model with Y an ordered > variable and X1 and X2 independent dummy vars: > > > > SETPANEL; GROUP=ID; PDS=ROUNDS $ > > ORDERED ; PANEL ; Lhs = Y > > ; Rhs = ONE, X1, X2 > > ; MODEL= LOGIT $ > > ORDERED ; PANEL; Lhs = Y > > ; Rhs = one, X1, X2 > > ; RPM > > ; par > > ; Fcn = one(n), X1(n), X2(n) > > ; HALTON > > ; MODEL= LOGIT $ > > > > My results are as follows: > > > > |Means for random parameters > > Constant| 8.33434*** .85987 9.69 .0000 6.64903 10.01966 > > X1 | .12776 .32251 .40 .6920 -.50436 .75988 > > X2 | -.03393 .33030 -.10 .9182 -.68131 .61344 > > |Scale parameters for dists. of random parameters > > Constant| 3.26244*** .22798 14.31 .0000 2.81560 3.70927 > > X1 | .62482*** .21928 2.85 .0044 .19503 1.05460 > > X2 | .78015*** .22758 3.43 .0006 .33409 1.22621 > > > > My intuition is that for the X's, the results of the unconditional > estimates > mean that their effect is non-significant (i.e. zero) on aggregate but with > strong individual variability (i.e. showing that for some individuals it is > different from zero). > > > > However, when I look at the individual specific (conditional) estimates, I > find a strong indication of their effect being zero for all (i.e. > confidence > intervals for all individuals are way far from zero on both ends) > > > > Is my interpretation mistaken? And if not, how I can possibly explain the > different conclusions I get from conditional vs unconditional estimates? > > > > Thank you, > > > > Achilleas > > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > -- William Greene Department of Economics Stern School of Business, New York University 44 West 4 St., 7-90 New York, NY, 10012 URL: https://protect-au.mimecast.com/s/d4WfCL7rK8tDyPqxHBerlX?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.212.998.0876 Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics Associate Editor: Journal of Choice Modeling From avassilopoulos.aua at gmail.com Thu Feb 21 17:12:08 2019 From: avassilopoulos.aua at gmail.com (Achilleas' gmail) Date: Thu, 21 Feb 2019 08:12:08 +0200 Subject: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters In-Reply-To: References: <002201d4c94f$83bbf680$8b33e380$@gmail.com> Message-ID: <000101d4c9ac$617b1080$24713180$@gmail.com> Dear Prof. Green, Thank you for your reply. Indeed my statement was confusing the way I worded it. What I meant is that the conf. intervals of the conditional estimates include zero and the lower (upper) bounds are strictly negative (positive) for all individuals. So your last statement " Some of the distribution may be near zero while more of the distribution may be quite far from zero " (which was what I also expected) does not show itself in my data. When I plot the unconditional estimates with a kernel density plot, I see that only a small part of the tail is below zero as expected. Could this be an indication of random noise that is captured by the model as a distribution of population coefficients ? Best, Achilleas -----Original Message----- From: Limdep On Behalf Of William Greene via Limdep Sent: Thursday, February 21, 2019 06:11 To: Limdep and Nlogit Mailing List Cc: William Greene Subject: Re: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters Achilleas: Considering X1, your results suggest that the population distribution of coefficients on X1 has a normal distribution with mean of 0.12776 and a standard deviation of 0.62482. Interpretations of "statistical significance" in this context are ambiguous. The statement However, when I look at the individual specific (conditional) estimates, I find a strong indication of their effect being zero for all (i.e. confidence intervals for all individuals are way far from zero on both ends) is contradictory. It seems to state that the effects are close to zero, but confidence intervals are far from zero. What is true is that the conditional estimates will vary around the mean of 0.12276. The mean of the conditional estimates should equal the unconditional estimate. The conditional standard deviations that are estimable will all be less than the 0.62482, and the average of these conditional standard deviations will also be strictly less than 0.62482. Characterizations of the distribution of random parameters should not be based on "confidence intervals." Some of the distribution may be near zero while more of the distribution may be quite far from zero. /Bill Greene On Wed, Feb 20, 2019 at 2:07 PM Achilleas' gmail via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear all, > > > > I'm estimating a Random Coefficients Ordered Logit Model with Y an > ordered variable and X1 and X2 independent dummy vars: > > > > SETPANEL; GROUP=ID; PDS=ROUNDS $ > > ORDERED ; PANEL ; Lhs = Y > > ; Rhs = ONE, X1, X2 > > ; MODEL= LOGIT $ > > ORDERED ; PANEL; Lhs = Y > > ; Rhs = one, X1, X2 > > ; RPM > > ; par > > ; Fcn = one(n), X1(n), X2(n) > > ; HALTON > > ; MODEL= LOGIT $ > > > > My results are as follows: > > > > |Means for random parameters > > Constant| 8.33434*** .85987 9.69 .0000 6.64903 10.01966 > > X1 | .12776 .32251 .40 .6920 -.50436 .75988 > > X2 | -.03393 .33030 -.10 .9182 -.68131 .61344 > > |Scale parameters for dists. of random parameters > > Constant| 3.26244*** .22798 14.31 .0000 2.81560 3.70927 > > X1 | .62482*** .21928 2.85 .0044 .19503 1.05460 > > X2 | .78015*** .22758 3.43 .0006 .33409 1.22621 > > > > My intuition is that for the X's, the results of the unconditional > estimates mean that their effect is non-significant (i.e. zero) on > aggregate but with strong individual variability (i.e. showing that > for some individuals it is different from zero). > > > > However, when I look at the individual specific (conditional) > estimates, I find a strong indication of their effect being zero for all (i.e. > confidence > intervals for all individuals are way far from zero on both ends) > > > > Is my interpretation mistaken? And if not, how I can possibly explain > the different conclusions I get from conditional vs unconditional estimates? > > > > Thank you, > > > > Achilleas > > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > -- William Greene Department of Economics Stern School of Business, New York University 44 West 4 St., 7-90 New York, NY, 10012 URL: https://protect-au.mimecast.com/s/8kvwC3Q8Z2Fg60VXfgdF8d?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.212.998.0876 Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics Associate Editor: Journal of Choice Modeling _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au http://limdep.itls.usyd.edu.au From wgreene at stern.nyu.edu Thu Feb 21 22:04:08 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Thu, 21 Feb 2019 06:04:08 -0500 Subject: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters In-Reply-To: <000101d4c9ac$617b1080$24713180$@gmail.com> References: <002201d4c94f$83bbf680$8b33e380$@gmail.com> <000101d4c9ac$617b1080$24713180$@gmail.com> Message-ID: Achilleas. Your KDE is plotting the variation of the means of the conditional distributions across the individuals in the sample. From what you describe it sounds like you seek the range of variation of the within individual distributions (that is, the conditional distributions). There is a discussion of "centipede plots" in your manual that does this sort of thing. You might search for that material and see if it does what you are looking for. It uses the individual specific means and conditional standard deviations. Regards, Bill Greene On Thu, Feb 21, 2019 at 1:12 AM Achilleas' gmail via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear Prof. Green, > > Thank you for your reply. Indeed my statement was confusing the way I > worded > it. What I meant is that the conf. intervals of the conditional estimates > include zero and the lower (upper) bounds are strictly negative (positive) > for all individuals. > > So your last statement " Some of the distribution may be near zero while > more of the distribution may be quite far from zero " (which was what I > also > expected) does not show itself in my data. > > When I plot the unconditional estimates with a kernel density plot, I see > that only a small part of the tail is below zero as expected. > > Could this be an indication of random noise that is captured by the model > as > a distribution of population coefficients ? > > Best, > > Achilleas > > -----Original Message----- > From: Limdep On Behalf Of William > Greene via Limdep > Sent: Thursday, February 21, 2019 06:11 > To: Limdep and Nlogit Mailing List > Cc: William Greene > Subject: Re: [Limdep Nlogit List] Interpretation of Random coefficients > ordered logit parameters > > Achilleas: Considering X1, your results suggest that the population > distribution of coefficients on X1 has a normal distribution with mean of > 0.12776 and a standard deviation of 0.62482. > Interpretations of "statistical significance" in this context are > ambiguous. > The statement > > However, when I look at the individual specific (conditional) estimates, I > find a strong indication of their effect being zero for all (i.e. > confidence > intervals for all individuals are way far from zero on both ends) > > is contradictory. It seems to state that the effects are close to zero, > but > confidence intervals are far from zero. What is true is that the > conditional estimates will vary around the mean of 0.12276. The mean of > the > conditional estimates should equal the unconditional estimate. > The conditional standard deviations that are estimable will all be less > than > the 0.62482, and the average of these conditional standard deviations will > also be strictly less than 0.62482. > Characterizations of the distribution of random parameters should not be > based on "confidence intervals." Some of the distribution may be near zero > while more of the distribution may be quite far from zero. > > /Bill Greene > > > > On Wed, Feb 20, 2019 at 2:07 PM Achilleas' gmail via Limdep < > limdep at mailman.sydney.edu.au> wrote: > > > Dear all, > > > > > > > > I'm estimating a Random Coefficients Ordered Logit Model with Y an > > ordered variable and X1 and X2 independent dummy vars: > > > > > > > > SETPANEL; GROUP=ID; PDS=ROUNDS $ > > > > ORDERED ; PANEL ; Lhs = Y > > > > ; Rhs = ONE, X1, X2 > > > > ; MODEL= LOGIT $ > > > > ORDERED ; PANEL; Lhs = Y > > > > ; Rhs = one, X1, X2 > > > > ; RPM > > > > ; par > > > > ; Fcn = one(n), X1(n), X2(n) > > > > ; HALTON > > > > ; MODEL= LOGIT $ > > > > > > > > My results are as follows: > > > > > > > > |Means for random parameters > > > > Constant| 8.33434*** .85987 9.69 .0000 6.64903 10.01966 > > > > X1 | .12776 .32251 .40 .6920 -.50436 .75988 > > > > X2 | -.03393 .33030 -.10 .9182 -.68131 .61344 > > > > |Scale parameters for dists. of random parameters > > > > Constant| 3.26244*** .22798 14.31 .0000 2.81560 3.70927 > > > > X1 | .62482*** .21928 2.85 .0044 .19503 1.05460 > > > > X2 | .78015*** .22758 3.43 .0006 .33409 1.22621 > > > > > > > > My intuition is that for the X's, the results of the unconditional > > estimates mean that their effect is non-significant (i.e. zero) on > > aggregate but with strong individual variability (i.e. showing that > > for some individuals it is different from zero). > > > > > > > > However, when I look at the individual specific (conditional) > > estimates, I find a strong indication of their effect being zero for all > (i.e. > > confidence > > intervals for all individuals are way far from zero on both ends) > > > > > > > > Is my interpretation mistaken? And if not, how I can possibly explain > > the different conclusions I get from conditional vs unconditional > estimates? > > > > > > > > Thank you, > > > > > > > > Achilleas > > > > > > > > _______________________________________________ > > Limdep site list > > Limdep at mailman.sydney.edu.au > > http://limdep.itls.usyd.edu.au > > > > > > -- > William Greene > Department of Economics > Stern School of Business, New York University > 44 West 4 St., 7-90 > New York, NY, 10012 > URL: https://protect-au.mimecast.com/s/vNOnCJyp0qhW3Og9SVMBEc?domain=people.stern.nyu.edu > Email: wgreene at stern.nyu.edu > Ph. +1.212.998.0876 > Editor in Chief: Journal of Productivity Analysis Editor in Chief: > Foundations and Trends in Econometrics Associate Editor: Economics Letters > Associate Editor: Journal of Business and Economic Statistics Associate > Editor: Journal of Choice Modeling > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > -- William Greene Department of Economics Stern School of Business, New York University 44 West 4 St., 7-90 New York, NY, 10012 URL: https://protect-au.mimecast.com/s/vNOnCJyp0qhW3Og9SVMBEc?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.212.998.0876 Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics Associate Editor: Journal of Choice Modeling From avassilopoulos.aua at gmail.com Thu Feb 21 23:29:40 2019 From: avassilopoulos.aua at gmail.com (Achilleas' gmail) Date: Thu, 21 Feb 2019 14:29:40 +0200 Subject: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters In-Reply-To: References: <002201d4c94f$83bbf680$8b33e380$@gmail.com> <000101d4c9ac$617b1080$24713180$@gmail.com> Message-ID: <001e01d4c9e1$1ed754c0$5c85fe40$@gmail.com> Dear Prof. Greene, I agree about KDE but from its shape I (probably wrongly) expected that the confidence interval of some coefficients would be away from zero. The centipede plot is exactly what I need to show and actually it is the reason I started this thread. The plot shows observations around the horizontal axis and long lines (conf. intervals) that include zero in all observations. My worry was that this picture was not compatible with a stat.significant estimate unconditional standard deviation and might indicate random noise. I understand this is not the case so I will stick to individual-specific means and SD's to judge the effect of X1, X2. Thank you for your detailed reply. Best, Achilleas -----Original Message----- From: Limdep On Behalf Of William Greene via Limdep Sent: Thursday, February 21, 2019 13:04 To: Limdep and Nlogit Mailing List Cc: William Greene Subject: Re: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters Achilleas. Your KDE is plotting the variation of the means of the conditional distributions across the individuals in the sample. From what you describe it sounds like you seek the range of variation of the within individual distributions (that is, the conditional distributions). There is a discussion of "centipede plots" in your manual that does this sort of thing. You might search for that material and see if it does what you are looking for. It uses the individual specific means and conditional standard deviations. Regards, Bill Greene On Thu, Feb 21, 2019 at 1:12 AM Achilleas' gmail via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear Prof. Green, > > Thank you for your reply. Indeed my statement was confusing the way I > worded it. What I meant is that the conf. intervals of the conditional > estimates include zero and the lower (upper) bounds are strictly > negative (positive) for all individuals. > > So your last statement " Some of the distribution may be near zero > while more of the distribution may be quite far from zero " (which was > what I also > expected) does not show itself in my data. > > When I plot the unconditional estimates with a kernel density plot, I > see that only a small part of the tail is below zero as expected. > > Could this be an indication of random noise that is captured by the > model as a distribution of population coefficients ? > > Best, > > Achilleas > > -----Original Message----- > From: Limdep On Behalf Of > William Greene via Limdep > Sent: Thursday, February 21, 2019 06:11 > To: Limdep and Nlogit Mailing List > Cc: William Greene > Subject: Re: [Limdep Nlogit List] Interpretation of Random > coefficients ordered logit parameters > > Achilleas: Considering X1, your results suggest that the population > distribution of coefficients on X1 has a normal distribution with mean > of > 0.12776 and a standard deviation of 0.62482. > Interpretations of "statistical significance" in this context are > ambiguous. > The statement > > However, when I look at the individual specific (conditional) > estimates, I find a strong indication of their effect being zero for all (i.e. > confidence > intervals for all individuals are way far from zero on both ends) > > is contradictory. It seems to state that the effects are close to > zero, but confidence intervals are far from zero. What is true is > that the conditional estimates will vary around the mean of 0.12276. > The mean of the conditional estimates should equal the unconditional > estimate. > The conditional standard deviations that are estimable will all be > less than the 0.62482, and the average of these conditional standard > deviations will also be strictly less than 0.62482. > Characterizations of the distribution of random parameters should not > be based on "confidence intervals." Some of the distribution may be > near zero while more of the distribution may be quite far from zero. > > /Bill Greene > > > > On Wed, Feb 20, 2019 at 2:07 PM Achilleas' gmail via Limdep < > limdep at mailman.sydney.edu.au> wrote: > > > Dear all, > > > > > > > > I'm estimating a Random Coefficients Ordered Logit Model with Y an > > ordered variable and X1 and X2 independent dummy vars: > > > > > > > > SETPANEL; GROUP=ID; PDS=ROUNDS $ > > > > ORDERED ; PANEL ; Lhs = Y > > > > ; Rhs = ONE, X1, X2 > > > > ; MODEL= LOGIT $ > > > > ORDERED ; PANEL; Lhs = Y > > > > ; Rhs = one, X1, X2 > > > > ; RPM > > > > ; par > > > > ; Fcn = one(n), X1(n), X2(n) > > > > ; HALTON > > > > ; MODEL= LOGIT $ > > > > > > > > My results are as follows: > > > > > > > > |Means for random parameters > > > > Constant| 8.33434*** .85987 9.69 .0000 6.64903 10.01966 > > > > X1 | .12776 .32251 .40 .6920 -.50436 .75988 > > > > X2 | -.03393 .33030 -.10 .9182 -.68131 .61344 > > > > |Scale parameters for dists. of random parameters > > > > Constant| 3.26244*** .22798 14.31 .0000 2.81560 3.70927 > > > > X1 | .62482*** .21928 2.85 .0044 .19503 1.05460 > > > > X2 | .78015*** .22758 3.43 .0006 .33409 1.22621 > > > > > > > > My intuition is that for the X's, the results of the unconditional > > estimates mean that their effect is non-significant (i.e. zero) on > > aggregate but with strong individual variability (i.e. showing that > > for some individuals it is different from zero). > > > > > > > > However, when I look at the individual specific (conditional) > > estimates, I find a strong indication of their effect being zero for > > all > (i.e. > > confidence > > intervals for all individuals are way far from zero on both ends) > > > > > > > > Is my interpretation mistaken? And if not, how I can possibly > > explain the different conclusions I get from conditional vs > > unconditional > estimates? > > > > > > > > Thank you, > > > > > > > > Achilleas > > > > > > > > _______________________________________________ > > Limdep site list > > Limdep at mailman.sydney.edu.au > > http://limdep.itls.usyd.edu.au > > > > > > -- > William Greene > Department of Economics > Stern School of Business, New York University > 44 West 4 St., 7-90 > New York, NY, 10012 > URL: https://protect-au.mimecast.com/s/7NxuCr8DLRtRPgWOC7Ft1j?domain=people.stern.nyu.edu > Email: wgreene at stern.nyu.edu > Ph. +1.212.998.0876 > Editor in Chief: Journal of Productivity Analysis Editor in Chief: > Foundations and Trends in Econometrics Associate Editor: Economics > Letters Associate Editor: Journal of Business and Economic Statistics > Associate > Editor: Journal of Choice Modeling > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > -- William Greene Department of Economics Stern School of Business, New York University 44 West 4 St., 7-90 New York, NY, 10012 URL: https://protect-au.mimecast.com/s/7NxuCr8DLRtRPgWOC7Ft1j?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.212.998.0876 Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics Associate Editor: Journal of Choice Modeling _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au http://limdep.itls.usyd.edu.au From wgreene at stern.nyu.edu Thu Feb 21 23:51:45 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Thu, 21 Feb 2019 07:51:45 -0500 Subject: [Limdep Nlogit List] Interpretation of Random coefficients ordered logit parameters In-Reply-To: <001e01d4c9e1$1ed754c0$5c85fe40$@gmail.com> References: <002201d4c94f$83bbf680$8b33e380$@gmail.com> <000101d4c9ac$617b1080$24713180$@gmail.com> <001e01d4c9e1$1ed754c0$5c85fe40$@gmail.com> Message-ID: Achilleas. It's misleading to consider the intervals in the centipede plot "confidence intervals." The "confidence" aspect of a confidence interval relates to type 1 error induced by sampling variability. The intervals in the centipede plot are probability intervals from the conditional distribution. In a Bayesian context, they would be called HPD intervals. Regards Bill Greene On Thu, Feb 21, 2019 at 7:30 AM Achilleas' gmail via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear Prof. Greene, > > I agree about KDE but from its shape I (probably wrongly) expected that the > confidence interval of some coefficients would be away from zero. > > The centipede plot is exactly what I need to show and actually it is the > reason I started this thread. The plot shows observations around the > horizontal axis and long lines (conf. intervals) that include zero in all > observations. My worry was that this picture was not compatible with a > stat.significant estimate unconditional standard deviation and might > indicate random noise. > > I understand this is not the case so I will stick to individual-specific > means and SD's to judge the effect of X1, X2. > > Thank you for your detailed reply. > > Best, > Achilleas > > -----Original Message----- > From: Limdep On Behalf Of William > Greene via Limdep > Sent: Thursday, February 21, 2019 13:04 > To: Limdep and Nlogit Mailing List > Cc: William Greene > Subject: Re: [Limdep Nlogit List] Interpretation of Random coefficients > ordered logit parameters > > Achilleas. Your KDE is plotting the variation of the means of the > conditional distributions across the individuals in the sample. From what > you describe it sounds like you seek the range of variation of the within > individual distributions (that is, the conditional distributions). > There is a discussion of "centipede plots" in your manual that does this > sort of thing. You might search for that material and see if it does what > you are looking for. It uses the individual specific means and conditional > standard deviations. > Regards, > Bill Greene > > On Thu, Feb 21, 2019 at 1:12 AM Achilleas' gmail via Limdep < > limdep at mailman.sydney.edu.au> wrote: > > > Dear Prof. Green, > > > > Thank you for your reply. Indeed my statement was confusing the way I > > worded it. What I meant is that the conf. intervals of the conditional > > estimates include zero and the lower (upper) bounds are strictly > > negative (positive) for all individuals. > > > > So your last statement " Some of the distribution may be near zero > > while more of the distribution may be quite far from zero " (which was > > what I also > > expected) does not show itself in my data. > > > > When I plot the unconditional estimates with a kernel density plot, I > > see that only a small part of the tail is below zero as expected. > > > > Could this be an indication of random noise that is captured by the > > model as a distribution of population coefficients ? > > > > Best, > > > > Achilleas > > > > -----Original Message----- > > From: Limdep On Behalf Of > > William Greene via Limdep > > Sent: Thursday, February 21, 2019 06:11 > > To: Limdep and Nlogit Mailing List > > Cc: William Greene > > Subject: Re: [Limdep Nlogit List] Interpretation of Random > > coefficients ordered logit parameters > > > > Achilleas: Considering X1, your results suggest that the population > > distribution of coefficients on X1 has a normal distribution with mean > > of > > 0.12776 and a standard deviation of 0.62482. > > Interpretations of "statistical significance" in this context are > > ambiguous. > > The statement > > > > However, when I look at the individual specific (conditional) > > estimates, I find a strong indication of their effect being zero for all > (i.e. > > confidence > > intervals for all individuals are way far from zero on both ends) > > > > is contradictory. It seems to state that the effects are close to > > zero, but confidence intervals are far from zero. What is true is > > that the conditional estimates will vary around the mean of 0.12276. > > The mean of the conditional estimates should equal the unconditional > > estimate. > > The conditional standard deviations that are estimable will all be > > less than the 0.62482, and the average of these conditional standard > > deviations will also be strictly less than 0.62482. > > Characterizations of the distribution of random parameters should not > > be based on "confidence intervals." Some of the distribution may be > > near zero while more of the distribution may be quite far from zero. > > > > /Bill Greene > > > > > > > > On Wed, Feb 20, 2019 at 2:07 PM Achilleas' gmail via Limdep < > > limdep at mailman.sydney.edu.au> wrote: > > > > > Dear all, > > > > > > > > > > > > I'm estimating a Random Coefficients Ordered Logit Model with Y an > > > ordered variable and X1 and X2 independent dummy vars: > > > > > > > > > > > > SETPANEL; GROUP=ID; PDS=ROUNDS $ > > > > > > ORDERED ; PANEL ; Lhs = Y > > > > > > ; Rhs = ONE, X1, X2 > > > > > > ; MODEL= LOGIT $ > > > > > > ORDERED ; PANEL; Lhs = Y > > > > > > ; Rhs = one, X1, X2 > > > > > > ; RPM > > > > > > ; par > > > > > > ; Fcn = one(n), X1(n), X2(n) > > > > > > ; HALTON > > > > > > ; MODEL= LOGIT $ > > > > > > > > > > > > My results are as follows: > > > > > > > > > > > > |Means for random parameters > > > > > > Constant| 8.33434*** .85987 9.69 .0000 6.64903 > 10.01966 > > > > > > X1 | .12776 .32251 .40 .6920 -.50436 > .75988 > > > > > > X2 | -.03393 .33030 -.10 .9182 -.68131 > .61344 > > > > > > |Scale parameters for dists. of random parameters > > > > > > Constant| 3.26244*** .22798 14.31 .0000 2.81560 > 3.70927 > > > > > > X1 | .62482*** .21928 2.85 .0044 .19503 > 1.05460 > > > > > > X2 | .78015*** .22758 3.43 .0006 .33409 > 1.22621 > > > > > > > > > > > > My intuition is that for the X's, the results of the unconditional > > > estimates mean that their effect is non-significant (i.e. zero) on > > > aggregate but with strong individual variability (i.e. showing that > > > for some individuals it is different from zero). > > > > > > > > > > > > However, when I look at the individual specific (conditional) > > > estimates, I find a strong indication of their effect being zero for > > > all > > (i.e. > > > confidence > > > intervals for all individuals are way far from zero on both ends) > > > > > > > > > > > > Is my interpretation mistaken? And if not, how I can possibly > > > explain the different conclusions I get from conditional vs > > > unconditional > > estimates? > > > > > > > > > > > > Thank you, > > > > > > > > > > > > Achilleas > > > > > > > > > > > > _______________________________________________ > > > Limdep site list > > > Limdep at mailman.sydney.edu.au > > > http://limdep.itls.usyd.edu.au > > > > > > > > > > -- > > William Greene > > Department of Economics > > Stern School of Business, New York University > > 44 West 4 St., 7-90 > > New York, NY, 10012 > > URL: https://protect-au.mimecast.com/s/a3s5CQnzP0t1M2O8uxJ5Ie?domain=people.stern.nyu.edu > > Email: wgreene at stern.nyu.edu > > Ph. +1.212.998.0876 > > Editor in Chief: Journal of Productivity Analysis Editor in Chief: > > Foundations and Trends in Econometrics Associate Editor: Economics > > Letters Associate Editor: Journal of Business and Economic Statistics > > Associate > > Editor: Journal of Choice Modeling > > _______________________________________________ > > Limdep site list > > Limdep at mailman.sydney.edu.au > > http://limdep.itls.usyd.edu.au > > > > > > _______________________________________________ > > Limdep site list > > Limdep at mailman.sydney.edu.au > > http://limdep.itls.usyd.edu.au > > > > > > -- > William Greene > Department of Economics > Stern School of Business, New York University > 44 West 4 St., 7-90 > New York, NY, 10012 > URL: https://protect-au.mimecast.com/s/a3s5CQnzP0t1M2O8uxJ5Ie?domain=people.stern.nyu.edu > Email: wgreene at stern.nyu.edu > Ph. +1.212.998.0876 > Editor in Chief: Journal of Productivity Analysis Editor in Chief: > Foundations and Trends in Econometrics Associate Editor: Economics Letters > Associate Editor: Journal of Business and Economic Statistics Associate > Editor: Journal of Choice Modeling > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://limdep.itls.usyd.edu.au > > -- William Greene Department of Economics Stern School of Business, New York University 44 West 4 St., 7-90 New York, NY, 10012 URL: https://protect-au.mimecast.com/s/a3s5CQnzP0t1M2O8uxJ5Ie?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.212.998.0876 Editor in Chief: Journal of Productivity Analysis Editor in Chief: Foundations and Trends in Econometrics Associate Editor: Economics Letters Associate Editor: Journal of Business and Economic Statistics Associate Editor: Journal of Choice Modeling From kewan.mertens at kuleuven.be Fri Feb 22 19:31:36 2019 From: kewan.mertens at kuleuven.be (Kewan Mertens) Date: Fri, 22 Feb 2019 08:31:36 +0000 Subject: [Limdep Nlogit List] unsubsribe Message-ID: