From alessandro.corsi at unito.it Tue Apr 2 03:13:29 2019 From: alessandro.corsi at unito.it (Alessandro Corsi) Date: Mon, 1 Apr 2019 18:13:29 +0200 Subject: [Limdep Nlogit List] Crosstab command Message-ID: This might be a trivial question, but I just installed Nlogi6 and I was trying to get a table givng me the frequencies of a discrete variable. As I always did with previous versions, I gave the following command: CROSSTAB;Lhs=LEV_LET;Rhs=ONE$ but I received the message: Error??? 316: Crosstab: One of the variables is always <= 0 or >= 50. Though, my LEV_LET variable is neither <= 0 or >= 50, as shown by Dstat: DSTAT;Rhs=LEV_LET$ --------+--------------------------------------------------------------------- ??????? | Standard??????????????????????????????????? Missing Variable|???????? Mean??? Deviation????? Minimum Maximum??? Cases? Values --------+--------------------------------------------------------------------- ?LEV_LET|????? .706007????? .581714????????? 0.0 2.0??? 12752?????? 0 --------+--------------------------------------------------------------------- Descriptive Statistics for?? 1 variables DSTAT results are matrix LASTDSTA in current project. What is the problem? (BTW, I noticed that this version of Limdep does not seem to accept underscores in the variable names, even if they are reported with underscores in the variable list). Thnks for any help ??? Alessandro Alessandro Corsi Dept. of Economics and Statistics "Cognetti de Martiis" University of Torino, Italy Lungo Dora Siena, 100/A 10153 Torino Phone: +39-0116704409 From wgreene at stern.nyu.edu Tue Apr 2 03:16:56 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Mon, 1 Apr 2019 12:16:56 -0400 Subject: [Limdep Nlogit List] Crosstab command In-Reply-To: References: Message-ID: Alessandro. To get the frequencies for a discrete variable, use HISTOGRAM ; List ; Rhs = the variable $ Regards, Bill Greene On Mon, Apr 1, 2019 at 12:13 PM Alessandro Corsi wrote: > This might be a trivial question, but I just installed Nlogi6 and I was > trying to get a table givng me the frequencies of a discrete variable. > As I always did with previous versions, I gave the following command: > > CROSSTAB;Lhs=LEV_LET;Rhs=ONE$ > > but I received the message: Error 316: Crosstab: One of the variables > is always <= 0 or >= 50. > > Though, my LEV_LET variable is neither <= 0 or >= 50, as shown by Dstat: > > DSTAT;Rhs=LEV_LET$ > > --------+--------------------------------------------------------------------- > > | Standard Missing > Variable| Mean Deviation Minimum Maximum Cases Values > > --------+--------------------------------------------------------------------- > LEV_LET| .706007 .581714 0.0 2.0 12752 0 > > --------+--------------------------------------------------------------------- > > Descriptive Statistics for 1 variables > > DSTAT results are matrix LASTDSTA in current project. > > What is the problem? (BTW, I noticed that this version of Limdep does > not seem to accept underscores in the variable names, even if they are > reported with underscores in the variable list). > > Thnks for any help > > Alessandro > > > Alessandro Corsi > Dept. of Economics and Statistics "Cognetti de Martiis" > University of Torino, Italy > Lungo Dora Siena, 100/A > 10153 Torino > Phone: +39-0116704409 > > _______________________________________________ > 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/dZ23Cq7BKYtKYJ0YfZII0W?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 cathy.durham at oregonstate.edu Tue Apr 2 04:09:56 2019 From: cathy.durham at oregonstate.edu (Durham, Catherine Alison) Date: Mon, 1 Apr 2019 17:09:56 +0000 Subject: [Limdep Nlogit List] Crosstab command In-Reply-To: References: Message-ID: <00E811FB06246F43A379835C171DD4D90129F1F297@ex5.oregonstate.edu> Alessandro, I have tricked myself several times on the crosstab. Your dstat shows that your minimum is 0. The crosstab demands positive integers! -----Original Message----- From: Limdep On Behalf Of Alessandro Corsi Sent: Monday, April 1, 2019 9:13 AM To: limdep at mailman.sydney.edu.au Subject: [Limdep Nlogit List] Crosstab command This might be a trivial question, but I just installed Nlogi6 and I was trying to get a table givng me the frequencies of a discrete variable. As I always did with previous versions, I gave the following command: CROSSTAB;Lhs=LEV_LET;Rhs=ONE$ but I received the message: Error??? 316: Crosstab: One of the variables is always <= 0 or >= 50. Though, my LEV_LET variable is neither <= 0 or >= 50, as shown by Dstat: DSTAT;Rhs=LEV_LET$ --------+--------------------------------------------------------------- --------+------ ??????? | Standard??????????????????????????????????? Missing Variable|???????? Mean??? Deviation????? Minimum Maximum??? Cases? Variable|Values --------+--------------------------------------------------------------- --------+------ ?LEV_LET|????? .706007????? .581714????????? 0.0 2.0??? 12752?????? 0 --------+--------------------------------------------------------------- --------+------ Descriptive Statistics for?? 1 variables DSTAT results are matrix LASTDSTA in current project. What is the problem? (BTW, I noticed that this version of Limdep does not seem to accept underscores in the variable names, even if they are reported with underscores in the variable list). Thnks for any help ??? Alessandro Alessandro Corsi Dept. of Economics and Statistics "Cognetti de Martiis" University of Torino, Italy Lungo Dora Siena, 100/A 10153 Torino Phone: +39-0116704409 _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au http://limdep.itls.usyd.edu.au From d.petrolia at msstate.edu Tue Apr 9 00:03:05 2019 From: d.petrolia at msstate.edu (Petrolia, Daniel) Date: Mon, 8 Apr 2019 14:03:05 +0000 Subject: [Limdep Nlogit List] LCRPLOGIT with log-normal distribution In-Reply-To: References: Message-ID: Hi, my understanding is that log-normal distribution is not available for random parameter specifications in the LATENT CLASS RANDOM PARAMETER LOGIT (LCRPLOGIT) model of NLOGIT 6 (page N-638 of the NLOGIT 6 manual). I was interesting in applying the non-negative price coefficient constraint recommended in Carson and Czajkowski (JEEM 2019, https://protect-au.mimecast.com/s/SdnTCANZvPiLyDQxIGcW2R?domain=doi.org), but I don't think this is feasible, at least using NLOGIT, because it requires that the price coefficient be specified as log-normally distributed, which is available for RPLOGIT but not LCRPLOGIT. Is this correct? Thanks, Dan From wgreene at stern.nyu.edu Tue Apr 9 04:02:40 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Mon, 8 Apr 2019 14:02:40 -0400 Subject: [Limdep Nlogit List] LCRPLOGIT with log-normal distribution In-Reply-To: References: Message-ID: Dan. You are correct, NLRP does not support the lognormal model. In case it might serve the same purpose, it does support the one sided triangular, type 'O.' in the FCN definition. I do note, constraining parameters this way, generally doesn't actually do anything. For example, for a model in which you use the lognormal distribution, it is very likely that if you use the normal instead, you will observe the mean of the distribution to be more than 3 or 4 standard deviations from zero - that is, the parameter is essentially nonnegative anyway. Regards, Bill Greene On Mon, Apr 8, 2019 at 10:03 AM Petrolia, Daniel via Limdep < limdep at mailman.sydney.edu.au> wrote: > Hi, my understanding is that log-normal distribution is not available for > random parameter specifications in the LATENT CLASS RANDOM PARAMETER LOGIT > (LCRPLOGIT) model of NLOGIT 6 (page N-638 of the NLOGIT 6 manual). > > I was interesting in applying the non-negative price coefficient > constraint recommended in Carson and Czajkowski (JEEM 2019, > https://protect-au.mimecast.com/s/fGRGCK1qJZtzV9ErcM0uay?domain=doi.org), but I don't think this is > feasible, at least using NLOGIT, because it requires that the price > coefficient be specified as log-normally distributed, which is available > for RPLOGIT but not LCRPLOGIT. > > Is this correct? > > Thanks, > Dan > > > > > _______________________________________________ > 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/DhQhCL7rK8ty6XjYCqxNvV?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 nievesvaldes at gmail.com Thu Apr 18 02:07:51 2019 From: nievesvaldes at gmail.com (=?UTF-8?Q?Nieves_Vald=C3=A9s?=) Date: Wed, 17 Apr 2019 12:07:51 -0400 Subject: [Limdep Nlogit List] On the standard errors of the average partial effects after NLOGIT Message-ID: Hi there! I'm interested in using nlogit, specifically because I need to compute marginal effects and their standard errors. I'm looking at the Student Reference Guide of Nlogit 4.0 and it says "The standard deviations are not the asymptotic standard errors for the estimators of the marginal effects...Computing an appropriate standard error for that statistic is difficult to impossible...The hypothesis that a variable is not influential in the determination of the choices should be tested at the coefficient level." Is this correct? Is this also the case in Nlogit 6.0? Thanks in advance! Best Regards Nieves From wgreene at stern.nyu.edu Thu Apr 18 02:22:15 2019 From: wgreene at stern.nyu.edu (William Greene) Date: Wed, 17 Apr 2019 12:22:15 -0400 Subject: [Limdep Nlogit List] On the standard errors of the average partial effects after NLOGIT In-Reply-To: References: Message-ID: Nieves. NLOGIT has been extended since version 4. Regarding elasticities, there are two measures of variation. The first is the "standard deviation" you describe below. The second is the asymptotic variance that derives from the theory of the MLE. NLOGIT 4 did not compute these. NLOGIT 6 does. The meaning of the hypothesis that an elasticity equals zero in the context of the model remains ambiguous. But, you can obtain these asymptotic standard errors with NLOGIT 6. Regards Bill Greene On Wed, Apr 17, 2019 at 12:08 PM Nieves Vald?s via Limdep < limdep at mailman.sydney.edu.au> wrote: > Hi there! > I'm interested in using nlogit, specifically because I need to compute > marginal effects and their standard errors. > I'm looking at the Student Reference Guide of Nlogit 4.0 and it says "The > standard deviations are not the asymptotic standard errors for the > estimators of the marginal effects...Computing an appropriate standard > error for that statistic is difficult to impossible...The hypothesis that a > variable is not influential in the determination of the choices should be > tested at the coefficient level." > Is this correct? Is this also the case in Nlogit 6.0? > Thanks in advance! > Best Regards > Nieves > _______________________________________________ > 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/C3UwC2xZYvC1OGWRcntbcE?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