From bergtold at ksu.edu Sun May 2 11:59:32 2021 From: bergtold at ksu.edu (Jason Bergtold) Date: Sun, 2 May 2021 01:59:32 +0000 Subject: [Limdep Nlogit List] Question on MVPROBIT Command Message-ID: Dear Group: I am estimating a multivariate probit model with six variables and 31 covariates. I run into a problem in that I get the following message: Error 376: Unable to store full estimated parameter vector. K > 150. Is there a way to increase the storage size to overcome this problem for this command? Thanks Jason From Thao.T.Thai at monash.edu Thu May 6 20:03:59 2021 From: Thao.T.Thai at monash.edu (Thao Thai) Date: Thu, 6 May 2021 20:03:59 +1000 Subject: [Limdep Nlogit List] MNL in WTP space Message-ID: Dear Nlogit users, I am estimating a MNL in WTP-space using the GMXL command in Nlogit (1) If I specify all of my 20 coefficients as constant, the model does not work. I specified ;tau = [0] as I think it would result in an MNL without scale. ;fcn= hos(c), pri(c), ind(c), gov(c), non(c), rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) *;tau = [0]* (2) If I specify some of my variables as non-random (which means I do not specify them in ;fcn), the model works. However, the non-random variables have coefficients in preference space (which means their coefficients are the same as those in MNL model in preference space) ;fcn= rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) *;tau = [0]* (3) When I used the same specification as in (2) but with ;gamma = [0], Nlogit reports "Error 805: Initial iterations cannot improve function.Status=3" ;fcn= rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) *;gamma = [0]* My questions are: 1. Why does Nlogit produce coefficients in preference space for non-random variables? Are there any ways to get an MNL in WTP space for all coefficients in Nlogit ? 2. Please advise why the model with ;gamma = [0] can not run? My detailed syntax is below. Any suggestions you can share would be much appreciated. Thank you! Best regards, Thao *(1) This did not work: * |-> CALC ; Ran(29531928) $ |-> sample;all$ |-> reject;ALLCHO#30$ |-> reject;FCPC=1$ |-> GMXlogit;userp ;lhs = cho, cset, alts ;choices = a,b,c,d,e,f ;checkdata ;fcn= hos(c), pri(c), ind(c), gov(c), non(c), rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) ;pts=100 ;pds=3 ;halton ;maxit=300 ;gmx ;par ;tau= [0] ;model: U(a) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(b) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(c) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(d) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(e) =hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(f) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N $ +----------------------------------------------------------+ | Inspecting the data set before estimation. | | These errors mark observations which will be skipped. | | Row Individual = 1st row then group number of data block | +----------------------------------------------------------+ No bad observations were found in the sample Iterative procedure has converged Normal exit: 5 iterations. Status=0, F= .3903337D+04 ----------------------------------------------------------------------------- Start values obtained using MNL model Dependent variable Choice Log likelihood function -3903.33652 Estimation based on N = 2370, K = 20 Inf.Cr.AIC = 7846.7 AIC/N = 3.311 --------------------------------------- Log likelihood R-sqrd R2Adj ASCs only model must be fit separately Use NLOGIT ;...;RHS=ONE$ Note: R-sqrd = 1 - logL/Logl(constants) Warning: Model does not contain a full set of ASCs. R-sqrd is problematic. Use model setup with ;RHS=one to get LogL0. --------------------------------------- Response data are given as ind. choices Number of obs.= 2370, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHO| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- HOS| .11397 .14452 .79 .4303 -.16929 .39722 PRI| .34043*** .12631 2.70 .0070 .09286 .58800 IND| -.94307*** .16462 -5.73 .0000 -1.26571 -.62042 GOV| .08533 .13264 .64 .5200 -.17464 .34529 NON| -.08990 .14032 -.64 .5217 -.36492 .18512 RLH1| .08068 .15904 .51 .6119 -.23102 .39239 RLH2| .19219 .13233 1.45 .1464 -.06717 .45155 FL| .20004*** .05424 3.69 .0002 .09373 .30634 CR1| .43243*** .05697 7.59 .0000 .32077 .54410 CR2| .15373* .08971 1.71 .0866 -.02210 .32957 LO1| -.56437*** .05770 -9.78 .0000 -.67746 -.45129 SA| .01263*** .00080 15.86 .0000 .01107 .01419 RLC1| .39288** .16670 2.36 .0184 .06614 .71961 RLC2| .30506** .15166 2.01 .0443 .00781 .60231 LO2| -.86848*** .08495 -10.22 .0000 -1.03499 -.70198 RLP1| -.01482 .12190 -.12 .9033 -.25373 .22410 RLI1| .65116*** .14499 4.49 .0000 .36699 .93534 RLI2| .77620*** .15085 5.15 .0000 .48054 1.07187 RLG1| -.34006*** .12687 -2.68 .0074 -.58873 -.09140 RLN1| -.08453 .13534 -.62 .5322 -.34980 .18073 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on May 06, 2021 at 03:04:54 PM ----------------------------------------------------------------------------- Error 805: Initial iterations cannot improve function.Status=3 Function value was .4246469942D+04 at entry ----------- .2835448956D+05 at exit ----------- Error 1025: Failed to fit model. See earlier diagnostic. *(2) This syntax worked and its results * |-> CALC ; Ran(29531928) $ |-> sample;all$ |-> reject;ALLCHO#30$ |-> reject;FCPC=1$ |-> GMXlogit;userp ;lhs = cho, cset, alts ;choices = a,b,c,d,e,f ;checkdata ;fcn= rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) ;pts=100 ;pds=3 ;halton ;maxit=300 ;gmx ;par ;tau= [0] ;model: U(a) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(b) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(c) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(d) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(e) =hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N / U(f) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON + rlh1 * RL_H1 + rlh2 * RL_H2 + fl * FL_H + cr1 * CR_H1 + cr2 * CR_H2 + lo1 * LO_H1 + sa * SA_H + rlc1 * RL_C1 + rlc2 * RL_C2 + fl * FL_C + cr1 * CR_C1 + cr2 * CR_C2 + lo1 * LO_C1 + lo2 * LO_C2 + sa * SA_C + rlp1 * RL_P1 + fl * FL_P + cr1 * CR_P1 + cr2 * CR_P2 + lo1 * LO_P1 + lo2 * LO_P2 + sa * SA_P + rli1 * RL_I1 + rli2 * RL_I2 + fl * FL_I + cr1 * CR_I1 + lo1 * LO_I1 + sa * SA_I + rlg1 * RL_G1 + fl * FL_G + cr1 * CR_G1 + lo1 * LO_G1 + sa * SA_G + rln1 * RL_N1 + fl * FL_N + cr1 * CR_N1 + lo1 * LO_N1 + lo2 * LO_N2 + sa * SA_N $ +----------------------------------------------------------+ | Inspecting the data set before estimation. | | These errors mark observations which will be skipped. | | Row Individual = 1st row then group number of data block | +----------------------------------------------------------+ No bad observations were found in the sample Iterative procedure has converged Normal exit: 5 iterations. Status=0, F= .3903337D+04 ----------------------------------------------------------------------------- Start values obtained using MNL model Dependent variable Choice Log likelihood function -3903.33652 Estimation based on N = 2370, K = 20 Inf.Cr.AIC = 7846.7 AIC/N = 3.311 --------------------------------------- Log likelihood R-sqrd R2Adj ASCs only model must be fit separately Use NLOGIT ;...;RHS=ONE$ Note: R-sqrd = 1 - logL/Logl(constants) Warning: Model does not contain a full set of ASCs. R-sqrd is problematic. Use model setup with ;RHS=one to get LogL0. --------------------------------------- Response data are given as ind. choices Number of obs.= 2370, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHO| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- RLH1| .08068 .15904 .51 .6119 -.23102 .39239 RLH2| .19219 .13233 1.45 .1464 -.06717 .45155 FL| .20004*** .05424 3.69 .0002 .09373 .30634 CR1| .43243*** .05697 7.59 .0000 .32077 .54410 CR2| .15373* .08971 1.71 .0866 -.02210 .32957 LO1| -.56437*** .05770 -9.78 .0000 -.67746 -.45129 SA| .01263*** .00080 15.86 .0000 .01107 .01419 RLC1| .39288** .16670 2.36 .0184 .06614 .71961 RLC2| .30506** .15166 2.01 .0443 .00781 .60231 LO2| -.86848*** .08495 -10.22 .0000 -1.03499 -.70198 RLP1| -.01482 .12190 -.12 .9033 -.25373 .22410 RLI1| .65116*** .14499 4.49 .0000 .36699 .93534 RLI2| .77620*** .15085 5.15 .0000 .48054 1.07187 RLG1| -.34006*** .12687 -2.68 .0074 -.58873 -.09140 RLN1| -.08453 .13534 -.62 .5322 -.34980 .18073 HOS| .11397 .14452 .79 .4303 -.16929 .39722 PRI| .34043*** .12631 2.70 .0070 .09286 .58800 IND| -.94307*** .16462 -5.73 .0000 -1.26571 -.62042 GOV| .08533 .13264 .64 .5200 -.17464 .34529 NON| -.08990 .14032 -.64 .5217 -.36492 .18512 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on May 06, 2021 at 02:26:45 PM ----------------------------------------------------------------------------- Iterative procedure has converged Normal exit: 48 iterations. Status=0, F= .3903337D+04 ----------------------------------------------------------------------------- Generalized Mixed (RP) Logit Model Dependent variable CHO Log likelihood function -3903.33652 Restricted log likelihood -4246.46994 Chi squared [ 20](P= .000) 686.26684 Significance level .00000 McFadden Pseudo R-squared .0808044 Estimation based on N = 2370, K = 20 Inf.Cr.AIC = 7846.7 AIC/N = 3.311 --------------------------------------- Log likelihood R-sqrd R2Adj No coefficients -4246.4699 .0808 .0793 Constants only can be computed directly Use NLOGIT ;...;RHS=ONE$ At start values ********** .6987 .6982 Note: R-sqrd = 1 - logL/Logl(constants) Warning: Model does not contain a full set of ASCs. R-sqrd is problematic. Use model setup with ;RHS=one to get LogL0. --------------------------------------- Response data are given as ind. choices Replications for simulated probs. = 100 Used Halton sequences in simulations. RPL model with panel has 790 groups Fixed number of obsrvs./group= 3 Number of obs.= 2370, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHO| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- |Random parameters in utility functions.............................. RLH1| 6.38639 12.59107 .51 .6120 -18.29166 31.06444 RLH2| 15.2124 10.50848 1.45 .1477 -5.3839 35.8086 FL| 15.8337*** 4.24480 3.73 .0002 7.5141 24.1534 CR1| 34.2284*** 4.99186 6.86 .0000 24.4446 44.0123 CR2| 12.1685* 7.13125 1.71 .0879 -1.8085 26.1455 LO1| -44.6720*** 5.35949 -8.34 .0000 -55.1764 -34.1676 SA| 1.0 .....(Fixed Parameter)..... RLC1| 31.0976** 13.94301 2.23 .0257 3.7698 58.4254 RLC2| 24.1465* 12.37301 1.95 .0510 -.1042 48.3971 LO2| -68.7436*** 8.40285 -8.18 .0000 -85.2129 -52.2744 RLP1| -1.17281 9.63099 -.12 .9031 -20.04920 17.70357 RLI1| 51.5419*** 11.86105 4.35 .0000 28.2946 74.7891 RLI2| 61.4391*** 11.65835 5.27 .0000 38.5891 84.2890 RLG1| -26.9173*** 9.86829 -2.73 .0064 -46.2588 -7.5758 RLN1| -6.69115 10.70810 -.62 .5321 -27.67865 14.29634 |Nonrandom parameters in utility functions........................... HOS| .11397 .14452 .79 .4303 -.16929 .39722 PRI| .34043*** .12631 2.70 .0070 .09286 .58800 IND| -.94307*** .16462 -5.73 .0000 -1.26571 -.62042 GOV| .08533 .13264 .64 .5200 -.17464 .34529 NON| -.08990 .14032 -.64 .5217 -.36492 .18512 |Distns. of RPs. Std.Devs or limits of triangular.................... CsRLH1| 0.0 .....(Fixed Parameter)..... CsRLH2| 0.0 .....(Fixed Parameter)..... CsFL| 0.0 .....(Fixed Parameter)..... CsCR1| 0.0 .....(Fixed Parameter)..... CsCR2| 0.0 .....(Fixed Parameter)..... CsLO1| 0.0 .....(Fixed Parameter)..... CsSA| 0.0 .....(Fixed Parameter)..... CsRLC1| 0.0 .....(Fixed Parameter)..... CsRLC2| 0.0 .....(Fixed Parameter)..... CsLO2| 0.0 .....(Fixed Parameter)..... CsRLP1| 0.0 .....(Fixed Parameter)..... CsRLI1| 0.0 .....(Fixed Parameter)..... CsRLI2| 0.0 .....(Fixed Parameter)..... CsRLG1| 0.0 .....(Fixed Parameter)..... CsRLN1| 0.0 .....(Fixed Parameter)..... |Variance parameter tau in GMX scale parameter....................... TauScale| 0.0 .....(Fixed Parameter)..... |Weighting parameter gamma in GMX model.............................. GammaMXL| 0.0 .....(Fixed Parameter)..... |Coefficient on SA in preference space form.................... Beta0WTP| .01263*** .00080 15.86 .0000 .01107 .01419 S_b0_WTP| 0.0 .....(Fixed Parameter)..... | Sample Mean Sample Std.Dev..................................... Sigma(i)| 1.0*** .2153D-05 ******** .0000 .10000D+01 .10000D+01 --------+-------------------------------------------------------------------- nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx. ***, **, * ==> Significance at 1%, 5%, 10% level. Fixed parameter ... is constrained to equal the value or had a nonpositive st.error because of an earlier problem. Model was estimated on May 06, 2021 at 02:57:53 PM ----------------------------------------------------------------------------- From wgreene at stern.nyu.edu Fri May 7 00:38:08 2021 From: wgreene at stern.nyu.edu (William Greene) Date: Thu, 6 May 2021 10:38:08 -0400 Subject: [Limdep Nlogit List] MNL in WTP space In-Reply-To: References: Message-ID: Thao. When there are no random coefficients in the model, and tau=gamma=0, then the resulting model is a simple MNL. Specifying this as an GMXL model makes the program act peculiarly as the GMXL is written specifically for the case when those are nonzero. It's hard to predict how it will behave. The right way to fit that model is as an MNL. But, it is also important that the MNL model in WTP space is a one to one transformation of the same model in preference space. That means that you get the identical answers in the two cases. This is a theoretical property of the MLE known as "invariance." The MLE is invariant to 1:1 transformations of the parameter space. /Bill Greene On Thu, May 6, 2021 at 6:05 AM Thao Thai via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear Nlogit users, > > I am estimating a MNL in WTP-space using the GMXL command in Nlogit > > (1) If I specify all of my 20 coefficients as constant, the model does not > work. I specified ;tau = [0] as I think it would result in an MNL without > scale. > > ;fcn= hos(c), pri(c), ind(c), gov(c), non(c), rlh1(c), rlh2(c), fl(c), > cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), > rli2(c), rlg1(c), rln1(c) > *;tau = [0]* > > (2) If I specify some of my variables as non-random (which means I do not > specify them in ;fcn), the model works. However, the non-random variables > have coefficients in preference space (which means their coefficients are > the same as those in MNL model in preference space) > > ;fcn= rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), > rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) > *;tau = [0]* > > (3) When I used the same specification as in (2) but with ;gamma = [0], > Nlogit reports "Error 805: Initial iterations cannot improve > function.Status=3" > ;fcn= rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), > rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) > *;gamma = [0]* > > My questions are: > 1. Why does Nlogit produce coefficients in preference space for non-random > variables? Are there any ways to get an MNL in WTP space for all > coefficients in Nlogit ? > 2. Please advise why the model with ;gamma = [0] can not run? > My detailed syntax is below. > > Any suggestions you can share would be much appreciated. Thank you! > Best regards, > Thao > > *(1) This did not work: * > > |-> CALC ; Ran(29531928) $ > |-> sample;all$ > |-> reject;ALLCHO#30$ > |-> reject;FCPC=1$ > |-> GMXlogit;userp > ;lhs = cho, cset, alts > ;choices = a,b,c,d,e,f > ;checkdata > ;fcn= hos(c), pri(c), ind(c), gov(c), non(c), rlh1(c), rlh2(c), fl(c), > cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), rlc2(c), lo2(c), rlp1(c), rli1(c), > rli2(c), rlg1(c), rln1(c) > ;pts=100 > ;pds=3 > ;halton > ;maxit=300 > ;gmx > ;par > ;tau= [0] > ;model: > U(a) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(b) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(c) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(d) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(e) =hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(f) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > $ > +----------------------------------------------------------+ > | Inspecting the data set before estimation. | > | These errors mark observations which will be skipped. | > | Row Individual = 1st row then group number of data block | > +----------------------------------------------------------+ > No bad observations were found in the sample > > Iterative procedure has converged > Normal exit: 5 iterations. Status=0, F= .3903337D+04 > > > ----------------------------------------------------------------------------- > Start values obtained using MNL model > Dependent variable Choice > Log likelihood function -3903.33652 > Estimation based on N = 2370, K = 20 > Inf.Cr.AIC = 7846.7 AIC/N = 3.311 > --------------------------------------- > Log likelihood R-sqrd R2Adj > ASCs only model must be fit separately > Use NLOGIT ;...;RHS=ONE$ > Note: R-sqrd = 1 - logL/Logl(constants) > Warning: Model does not contain a full > set of ASCs. R-sqrd is problematic. Use > model setup with ;RHS=one to get LogL0. > --------------------------------------- > Response data are given as ind. choices > Number of obs.= 2370, skipped 0 obs > > --------+-------------------------------------------------------------------- > | Standard Prob. 95% Confidence > CHO| Coefficient Error z |z|>Z* Interval > > --------+-------------------------------------------------------------------- > HOS| .11397 .14452 .79 .4303 -.16929 .39722 > PRI| .34043*** .12631 2.70 .0070 .09286 .58800 > IND| -.94307*** .16462 -5.73 .0000 -1.26571 -.62042 > GOV| .08533 .13264 .64 .5200 -.17464 .34529 > NON| -.08990 .14032 -.64 .5217 -.36492 .18512 > RLH1| .08068 .15904 .51 .6119 -.23102 .39239 > RLH2| .19219 .13233 1.45 .1464 -.06717 .45155 > FL| .20004*** .05424 3.69 .0002 .09373 .30634 > CR1| .43243*** .05697 7.59 .0000 .32077 .54410 > CR2| .15373* .08971 1.71 .0866 -.02210 .32957 > LO1| -.56437*** .05770 -9.78 .0000 -.67746 -.45129 > SA| .01263*** .00080 15.86 .0000 .01107 .01419 > RLC1| .39288** .16670 2.36 .0184 .06614 .71961 > RLC2| .30506** .15166 2.01 .0443 .00781 .60231 > LO2| -.86848*** .08495 -10.22 .0000 -1.03499 -.70198 > RLP1| -.01482 .12190 -.12 .9033 -.25373 .22410 > RLI1| .65116*** .14499 4.49 .0000 .36699 .93534 > RLI2| .77620*** .15085 5.15 .0000 .48054 1.07187 > RLG1| -.34006*** .12687 -2.68 .0074 -.58873 -.09140 > RLN1| -.08453 .13534 -.62 .5322 -.34980 .18073 > > --------+-------------------------------------------------------------------- > ***, **, * ==> Significance at 1%, 5%, 10% level. > Model was estimated on May 06, 2021 at 03:04:54 PM > > ----------------------------------------------------------------------------- > > Error 805: Initial iterations cannot improve function.Status=3 > Function value was .4246469942D+04 at entry ----------- > .2835448956D+05 at exit ----------- > Error 1025: Failed to fit model. See earlier diagnostic. > *(2) This syntax worked and its results * > > |-> CALC ; Ran(29531928) $ > |-> sample;all$ > |-> reject;ALLCHO#30$ > |-> reject;FCPC=1$ > |-> GMXlogit;userp > ;lhs = cho, cset, alts > ;choices = a,b,c,d,e,f > ;checkdata > ;fcn= rlh1(c), rlh2(c), fl(c), cr1(c), cr2(c), lo1(c), sa(*c), rlc1(c), > rlc2(c), lo2(c), rlp1(c), rli1(c), rli2(c), rlg1(c), rln1(c) > ;pts=100 > ;pds=3 > ;halton > ;maxit=300 > ;gmx > ;par > ;tau= [0] > ;model: > U(a) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(b) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(c) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(d) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(e) =hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > / > U(f) = hos*HOS + pri*PRI + ind*IND + gov*GOV + non*NON > + rlh1 * RL_H1 + rlh2 * RL_H2 > + fl * FL_H > + cr1 * CR_H1 + cr2 * CR_H2 > + lo1 * LO_H1 > + sa * SA_H > + rlc1 * RL_C1 + rlc2 * RL_C2 > + fl * FL_C > + cr1 * CR_C1 + cr2 * CR_C2 > + lo1 * LO_C1 + lo2 * LO_C2 > + sa * SA_C > + rlp1 * RL_P1 > + fl * FL_P > + cr1 * CR_P1 + cr2 * CR_P2 > + lo1 * LO_P1 + lo2 * LO_P2 > + sa * SA_P > + rli1 * RL_I1 + rli2 * RL_I2 > + fl * FL_I > + cr1 * CR_I1 > + lo1 * LO_I1 > + sa * SA_I > + rlg1 * RL_G1 > + fl * FL_G > + cr1 * CR_G1 > + lo1 * LO_G1 > + sa * SA_G > + rln1 * RL_N1 > + fl * FL_N > + cr1 * CR_N1 > + lo1 * LO_N1 + lo2 * LO_N2 > + sa * SA_N > $ > +----------------------------------------------------------+ > | Inspecting the data set before estimation. | > | These errors mark observations which will be skipped. | > | Row Individual = 1st row then group number of data block | > +----------------------------------------------------------+ > No bad observations were found in the sample > > Iterative procedure has converged > Normal exit: 5 iterations. Status=0, F= .3903337D+04 > > > ----------------------------------------------------------------------------- > Start values obtained using MNL model > Dependent variable Choice > Log likelihood function -3903.33652 > Estimation based on N = 2370, K = 20 > Inf.Cr.AIC = 7846.7 AIC/N = 3.311 > --------------------------------------- > Log likelihood R-sqrd R2Adj > ASCs only model must be fit separately > Use NLOGIT ;...;RHS=ONE$ > Note: R-sqrd = 1 - logL/Logl(constants) > Warning: Model does not contain a full > set of ASCs. R-sqrd is problematic. Use > model setup with ;RHS=one to get LogL0. > --------------------------------------- > Response data are given as ind. choices > Number of obs.= 2370, skipped 0 obs > > --------+-------------------------------------------------------------------- > | Standard Prob. 95% Confidence > CHO| Coefficient Error z |z|>Z* Interval > > --------+-------------------------------------------------------------------- > RLH1| .08068 .15904 .51 .6119 -.23102 .39239 > RLH2| .19219 .13233 1.45 .1464 -.06717 .45155 > FL| .20004*** .05424 3.69 .0002 .09373 .30634 > CR1| .43243*** .05697 7.59 .0000 .32077 .54410 > CR2| .15373* .08971 1.71 .0866 -.02210 .32957 > LO1| -.56437*** .05770 -9.78 .0000 -.67746 -.45129 > SA| .01263*** .00080 15.86 .0000 .01107 .01419 > RLC1| .39288** .16670 2.36 .0184 .06614 .71961 > RLC2| .30506** .15166 2.01 .0443 .00781 .60231 > LO2| -.86848*** .08495 -10.22 .0000 -1.03499 -.70198 > RLP1| -.01482 .12190 -.12 .9033 -.25373 .22410 > RLI1| .65116*** .14499 4.49 .0000 .36699 .93534 > RLI2| .77620*** .15085 5.15 .0000 .48054 1.07187 > RLG1| -.34006*** .12687 -2.68 .0074 -.58873 -.09140 > RLN1| -.08453 .13534 -.62 .5322 -.34980 .18073 > HOS| .11397 .14452 .79 .4303 -.16929 .39722 > PRI| .34043*** .12631 2.70 .0070 .09286 .58800 > IND| -.94307*** .16462 -5.73 .0000 -1.26571 -.62042 > GOV| .08533 .13264 .64 .5200 -.17464 .34529 > NON| -.08990 .14032 -.64 .5217 -.36492 .18512 > > --------+-------------------------------------------------------------------- > ***, **, * ==> Significance at 1%, 5%, 10% level. > Model was estimated on May 06, 2021 at 02:26:45 PM > > ----------------------------------------------------------------------------- > > Iterative procedure has converged > Normal exit: 48 iterations. Status=0, F= .3903337D+04 > > > ----------------------------------------------------------------------------- > Generalized Mixed (RP) Logit Model > Dependent variable CHO > Log likelihood function -3903.33652 > Restricted log likelihood -4246.46994 > Chi squared [ 20](P= .000) 686.26684 > Significance level .00000 > McFadden Pseudo R-squared .0808044 > Estimation based on N = 2370, K = 20 > Inf.Cr.AIC = 7846.7 AIC/N = 3.311 > --------------------------------------- > Log likelihood R-sqrd R2Adj > No coefficients -4246.4699 .0808 .0793 > Constants only can be computed directly > Use NLOGIT ;...;RHS=ONE$ > At start values ********** .6987 .6982 > Note: R-sqrd = 1 - logL/Logl(constants) > Warning: Model does not contain a full > set of ASCs. R-sqrd is problematic. Use > model setup with ;RHS=one to get LogL0. > --------------------------------------- > Response data are given as ind. choices > Replications for simulated probs. = 100 > Used Halton sequences in simulations. > RPL model with panel has 790 groups > Fixed number of obsrvs./group= 3 > Number of obs.= 2370, skipped 0 obs > > --------+-------------------------------------------------------------------- > | Standard Prob. 95% Confidence > CHO| Coefficient Error z |z|>Z* Interval > > --------+-------------------------------------------------------------------- > |Random parameters in utility > functions.............................. > RLH1| 6.38639 12.59107 .51 .6120 -18.29166 31.06444 > RLH2| 15.2124 10.50848 1.45 .1477 -5.3839 35.8086 > FL| 15.8337*** 4.24480 3.73 .0002 7.5141 24.1534 > CR1| 34.2284*** 4.99186 6.86 .0000 24.4446 44.0123 > CR2| 12.1685* 7.13125 1.71 .0879 -1.8085 26.1455 > LO1| -44.6720*** 5.35949 -8.34 .0000 -55.1764 -34.1676 > SA| 1.0 .....(Fixed Parameter)..... > RLC1| 31.0976** 13.94301 2.23 .0257 3.7698 58.4254 > RLC2| 24.1465* 12.37301 1.95 .0510 -.1042 48.3971 > LO2| -68.7436*** 8.40285 -8.18 .0000 -85.2129 -52.2744 > RLP1| -1.17281 9.63099 -.12 .9031 -20.04920 17.70357 > RLI1| 51.5419*** 11.86105 4.35 .0000 28.2946 74.7891 > RLI2| 61.4391*** 11.65835 5.27 .0000 38.5891 84.2890 > RLG1| -26.9173*** 9.86829 -2.73 .0064 -46.2588 -7.5758 > RLN1| -6.69115 10.70810 -.62 .5321 -27.67865 14.29634 > |Nonrandom parameters in utility > functions........................... > HOS| .11397 .14452 .79 .4303 -.16929 .39722 > PRI| .34043*** .12631 2.70 .0070 .09286 .58800 > IND| -.94307*** .16462 -5.73 .0000 -1.26571 -.62042 > GOV| .08533 .13264 .64 .5200 -.17464 .34529 > NON| -.08990 .14032 -.64 .5217 -.36492 .18512 > |Distns. of RPs. Std.Devs or limits of > triangular.................... > CsRLH1| 0.0 .....(Fixed Parameter)..... > CsRLH2| 0.0 .....(Fixed Parameter)..... > CsFL| 0.0 .....(Fixed Parameter)..... > CsCR1| 0.0 .....(Fixed Parameter)..... > CsCR2| 0.0 .....(Fixed Parameter)..... > CsLO1| 0.0 .....(Fixed Parameter)..... > CsSA| 0.0 .....(Fixed Parameter)..... > CsRLC1| 0.0 .....(Fixed Parameter)..... > CsRLC2| 0.0 .....(Fixed Parameter)..... > CsLO2| 0.0 .....(Fixed Parameter)..... > CsRLP1| 0.0 .....(Fixed Parameter)..... > CsRLI1| 0.0 .....(Fixed Parameter)..... > CsRLI2| 0.0 .....(Fixed Parameter)..... > CsRLG1| 0.0 .....(Fixed Parameter)..... > CsRLN1| 0.0 .....(Fixed Parameter)..... > |Variance parameter tau in GMX scale > parameter....................... > TauScale| 0.0 .....(Fixed Parameter)..... > |Weighting parameter gamma in GMX > model.............................. > GammaMXL| 0.0 .....(Fixed Parameter)..... > |Coefficient on SA in preference space > form.................... > Beta0WTP| .01263*** .00080 15.86 .0000 .01107 .01419 > S_b0_WTP| 0.0 .....(Fixed Parameter)..... > | Sample Mean Sample > Std.Dev..................................... > Sigma(i)| 1.0*** .2153D-05 ******** .0000 .10000D+01 .10000D+01 > > --------+-------------------------------------------------------------------- > nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx. > ***, **, * ==> Significance at 1%, 5%, 10% level. > Fixed parameter ... is constrained to equal the value or > had a nonpositive st.error because of an earlier problem. > Model was estimated on May 06, 2021 at 02:57:53 PM > > ----------------------------------------------------------------------------- > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/_P-bClx1Nji2AD49zTGHxbu?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/8pTZCmO5glujAG40RFOi6g5?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 From bergtold at ksu.edu Sat May 8 08:32:32 2021 From: bergtold at ksu.edu (Jason Bergtold) Date: Fri, 7 May 2021 22:32:32 +0000 Subject: [Limdep Nlogit List] Question on MVPROBIT Command In-Reply-To: References: Message-ID: Dear Group: I am trying to run a Random Valuations Model following Wang and in Alberni et al (2003) (Analysis of contingent valuation data with multiple bids and response options allowing respondents to express uncertainty." I have a ordered response variable with 7 ordered categories from definitely no to definitely yes. I am modeling this using an ordered Probit model. The base model would estimate 7 threshold values with the first and last being set to -inf and inf, respectively. The middle 5 threshold values for the model, call them mu1 < mu2 < mu 3 References: Message-ID: Dear Group: I am trying to run a Random Valuations Model following Wang and in Alberni et al (2003) (Analysis of contingent valuation data with multiple bids and response options allowing respondents to express uncertainty." I have a ordered response variable with 7 ordered categories from definitely no to definitely yes. I am modeling this using an ordered Probit model. The base model would estimate 7 threshold values with the first and last being set to -inf and inf, respectively. The middle 5 threshold values for the model, call them mu1 < mu2 < mu 3 References: Message-ID: Jason. mu5 = 1 and mu2 = mu4 do not accomplish what you are trying to do. These constraints will force certain probabilities to equal zero or be negative. I can't tell what you are restricting, but in order for the model to be kosher, you must have -inf < 0 < mu1 < mu2 < mu3 1, then your model produces zero probabilities. Likewise, forcing mu2 = mu4 will make a mess of things. You can't have mu2 = mu4 and have mu3 < mu4 and mu3 > mu2. Regards, Bill Greene On Fri, May 7, 2021 at 6:32 PM Jason Bergtold wrote: > Dear Group: > > I am trying to run a Random Valuations Model following Wang and in Alberni > et al (2003) (Analysis of contingent valuation data with multiple bids and > response options allowing respondents to express uncertainty." I have a > ordered response variable with 7 ordered categories from definitely no to > definitely yes. I am modeling this using an ordered Probit model. The base > model would estimate 7 threshold values with the first and last being set > to -inf and inf, respectively. The middle 5 threshold values for the model, > call them mu1 < mu2 < mu 3 the model that mu 1 - mu5 = 0 and mu2 - mu 4 = 0, to make the estimation of > the threshold parameters "symmetric". I am wondering how to impose this > restriction using the Ordered command. > > Regards, > > Jason > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/HXQfCVARKgCx5DRNPUGFDYC?domain=limdep.itls.usyd.edu.au > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/HXQfCVARKgCx5DRNPUGFDYC?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/5BwuCWLVXkU5PAWRyfxMva_?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 From bergtold at ksu.edu Sat May 8 09:32:37 2021 From: bergtold at ksu.edu (Jason Bergtold) Date: Fri, 7 May 2021 23:32:37 +0000 Subject: [Limdep Nlogit List] Question on MVPROBIT Command In-Reply-To: References: Message-ID: Bill, I do want -inf < mu1 < mu2 < mu3 On Behalf Of William Greene via Limdep Sent: Friday, May 7, 2021 6:27 PM To: Limdep and Nlogit Mailing List Cc: William Greene Subject: Re: [Limdep Nlogit List] Question on MVPROBIT Command This email originated from outside of K-State. Jason. mu5 = 1 and mu2 = mu4 do not accomplish what you are trying to do. These constraints will force certain probabilities to equal zero or be negative. I can't tell what you are restricting, but in order for the model to be kosher, you must have -inf < 0 < mu1 < mu2 < mu3 1, then your model produces zero probabilities. Likewise, forcing mu2 = mu4 will make a mess of things. You can't have mu2 = mu4 and have mu3 < mu4 and mu3 > mu2. Regards, Bill Greene On Fri, May 7, 2021 at 6:32 PM Jason Bergtold wrote: > Dear Group: > > I am trying to run a Random Valuations Model following Wang and in > Alberni et al (2003) (Analysis of contingent valuation data with > multiple bids and response options allowing respondents to express > uncertainty." I have a ordered response variable with 7 ordered > categories from definitely no to definitely yes. I am modeling this > using an ordered Probit model. The base model would estimate 7 > threshold values with the first and last being set to -inf and inf, > respectively. The middle 5 threshold values for the model, call them > mu1 < mu2 < mu 3 model that mu 1 - mu5 = 0 and mu2 - mu 4 = 0, to make the estimation > of the threshold parameters "symmetric". I am wondering how to impose this restriction using the Ordered command. > > Regards, > > Jason > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/_gvgCGv0oyC10VMD3tKblkY?domain=limdep.itls.usyd.edu.au > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/_gvgCGv0oyC10VMD3tKblkY?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/mkAqCJyBrGfq17Y69UGwT4q?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au https://protect-au.mimecast.com/s/_gvgCGv0oyC10VMD3tKblkY?domain=limdep.itls.usyd.edu.au From christoph.buschmann at thuenen.de Wed May 19 17:40:49 2021 From: christoph.buschmann at thuenen.de (Christoph Buschmann) Date: Wed, 19 May 2021 09:40:49 +0200 (CEST) Subject: [Limdep Nlogit List] Question on missing values in N-logit Message-ID: <813758758.2526827.1621410049880.JavaMail.zimbra@thuenen.de> Dear group, I have a question on how N-logit deals with missing data in the covariates. My choice experi-ments were carried out completely by all respondents, but 8 respondents did not answer all questions about the covariates. I have ten choice cards per respondent and 3 alternatives in each choice situation. So, my dataset comprises three rows for each choice situation. N-Logit reports that, for the 8 respondents in question, per choice situation one missing value has been found and refers to the first row per choice situation, i.e. the row that gives information about the choice of alternative 1. I assume that, for the 8 respondents in question, N-logit skips one row per choice situation (the one for alternative 1), including the information about covariates, and keeps the other two rows (for alternatives 2 and 3). This would mean that for the 8 respondents in question, some of the information is retained and so it makes sense to keep them in the data set. In fact, if I remove the 8 respondents from the dataset by hand, the model deteriorates (AIC). What is your experience with missing data? Is my interpretation correct? Thank you very much for your help. Kind regards, Christoph Buschmann -- Christoph Buschmann, M.Sc. Stabsstelle Klima/ Coordination Unit Climate Th?nen-Institut Bundesallee 49 D-38116 Braunschweig Germany Im Home-Office erreichbar per e-mail E-mail: christoph.buschmann at thuenen.de Homepage: [ https://protect-au.mimecast.com/s/0m2OC5QPXJiZRK8v6hzW5He?domain=thuenen.de | https://protect-au.mimecast.com/s/0m2OC5QPXJiZRK8v6hzW5He?domain=thuenen.de ] Twitter: @ThuenenClimSoil From david.hensher at sydney.edu.au Wed May 19 18:04:26 2021 From: david.hensher at sydney.edu.au (David Hensher) Date: Wed, 19 May 2021 08:04:26 +0000 Subject: [Limdep Nlogit List] Question on missing values in N-logit In-Reply-To: <813758758.2526827.1621410049880.JavaMail.zimbra@thuenen.de> References: <813758758.2526827.1621410049880.JavaMail.zimbra@thuenen.de> Message-ID: <2F625673-B1DE-4981-84BE-05346D280FD2@sydney.edu.au> Missing data can exist for a number of reasons. While not being sure what is happening here there are two codes of value. -999 will remove an entire choice set and -888 will ignore an attribute level associated with an alternative or individual you have in the data. So if a covariant is say income, code as -888 to retain the choice set but note this amounts to it being missing which can be problematic if you too many of these are missing and it is retained in the model. Some people either replace it with the mean or mode or run an auxiliary regression to try and predict it based on other covariates. David Sent from my iPhone 0418 433 057 David A Hensher Choice modellers prefer Nlogit: See https://protect-au.mimecast.com/s/mWd8CP7LAXfKZmYN6SzzuT_?domain=limdep.com ITLS celebrating 30 years Note: hgroup at hensher.com.au David.hensher at bigpond.com David.hensher at sydney.edu.au are linked so use one only On 19 May 2021, at 5:51 pm, Christoph Buschmann wrote: ?Dear group, I have a question on how N-logit deals with missing data in the covariates. My choice experi-ments were carried out completely by all respondents, but 8 respondents did not answer all questions about the covariates. I have ten choice cards per respondent and 3 alternatives in each choice situation. So, my dataset comprises three rows for each choice situation. N-Logit reports that, for the 8 respondents in question, per choice situation one missing value has been found and refers to the first row per choice situation, i.e. the row that gives information about the choice of alternative 1. I assume that, for the 8 respondents in question, N-logit skips one row per choice situation (the one for alternative 1), including the information about covariates, and keeps the other two rows (for alternatives 2 and 3). This would mean that for the 8 respondents in question, some of the information is retained and so it makes sense to keep them in the data set. In fact, if I remove the 8 respondents from the dataset by hand, the model deteriorates (AIC). What is your experience with missing data? Is my interpretation correct? Thank you very much for your help. Kind regards, Christoph Buschmann -- Christoph Buschmann, M.Sc. Stabsstelle Klima/ Coordination Unit Climate Th?nen-Institut Bundesallee 49 D-38116 Braunschweig Germany Im Home-Office erreichbar per e-mail E-mail: christoph.buschmann at thuenen.de Homepage: [ https://protect-au.mimecast.com/s/OGMdCQnMBZfk9qGlRtPHnQf?domain=thuenen.de | https://protect-au.mimecast.com/s/OGMdCQnMBZfk9qGlRtPHnQf?domain=thuenen.de ] Twitter: @ThuenenClimSoil _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au https://protect-au.mimecast.com/s/stc4CROND2uvQl90OCPwOo3?domain=limdep.itls.usyd.edu.au From wgreene at stern.nyu.edu Wed May 19 22:07:58 2021 From: wgreene at stern.nyu.edu (William Greene) Date: Wed, 19 May 2021 08:07:58 -0400 Subject: [Limdep Nlogit List] Question on missing values in N-logit In-Reply-To: <2F625673-B1DE-4981-84BE-05346D280FD2@sydney.edu.au> References: <813758758.2526827.1621410049880.JavaMail.zimbra@thuenen.de> <2F625673-B1DE-4981-84BE-05346D280FD2@sydney.edu.au> Message-ID: David and all. Be careful in using the -888 form for missing data. To reiterate, if one of the 3 alts has missing values for an attribute (-999), the entire choice task has to be removed from the sample. Christoph, your interpretation is not correct. NLOGIT does not create a new choice task by removing rows from the old one. -888 does not operate on the data. It operates on the coefficient vector. For observations with x(j,k) = -888, where j is the alt and k is the characteristic, when computing beta'x(j,k) for that alt j, beta(k) is set equal to zero - not x(j,k). This matters. You don't want to set the x(j,k) to zero, especially if it is a price. Nonattendance (-888) means the marginal utility is zero, not that the attribute is zero. Cheers Bill Greene On Wed, May 19, 2021 at 4:04 AM David Hensher via Limdep < limdep at mailman.sydney.edu.au> wrote: > Missing data can exist for a number of reasons. While not being sure what > is happening here there are two codes of value. -999 will remove an entire > choice set and -888 will ignore an attribute level associated with an > alternative or individual you have in the data. So if a covariant is say > income, code as -888 to retain the choice set but note this amounts to it > being missing which can be problematic if you too many of these are missing > and it is retained in the model. Some people either replace it with the > mean or mode or run an auxiliary regression to try and predict it based on > other covariates. > David > > Sent from my iPhone > 0418 433 057 > David A Hensher > Choice modellers prefer Nlogit: See https://protect-au.mimecast.com/s/FiLdCP7LAXfKZ28v4hz_nfd?domain=limdep.com > ITLS celebrating 30 years > > > Note: > hgroup at hensher.com.au David.hensher at bigpond.com > David.hensher at sydney.edu.au > are linked so use one only > > > On 19 May 2021, at 5:51 pm, Christoph Buschmann < > christoph.buschmann at thuenen.de> wrote: > > ?Dear group, > > I have a question on how N-logit deals with missing data in the > covariates. My choice experi-ments were carried out completely by all > respondents, but 8 respondents did not answer all questions about the > covariates. > > I have ten choice cards per respondent and 3 alternatives in each choice > situation. So, my dataset comprises three rows for each choice situation. > N-Logit reports that, for the 8 respondents in question, per choice > situation one missing value has been found and refers to the first row per > choice situation, i.e. the row that gives information about the choice of > alternative 1. > > I assume that, for the 8 respondents in question, N-logit skips one row > per choice situation (the one for alternative 1), including the information > about covariates, and keeps the other two rows (for alternatives 2 and 3). > > This would mean that for the 8 respondents in question, some of the > information is retained and so it makes sense to keep them in the data set. > In fact, if I remove the 8 respondents from the dataset by hand, the model > deteriorates (AIC). > > What is your experience with missing data? Is my interpretation correct? > > Thank you very much for your help. > > > Kind regards, > > Christoph Buschmann > > -- > Christoph Buschmann, M.Sc. > > Stabsstelle Klima/ Coordination Unit Climate > Th?nen-Institut > Bundesallee 49 > D-38116 Braunschweig > Germany > > Im Home-Office erreichbar per e-mail > E-mail: christoph.buschmann at thuenen.de > Homepage: [ https://protect-au.mimecast.com/s/x086CQnMBZfk9KMB6sP7Xui?domain=thuenen.de | https://protect-au.mimecast.com/s/x086CQnMBZfk9KMB6sP7Xui?domain=thuenen.de ] > Twitter: @ThuenenClimSoil > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/F5DvCROND2uvQpjnriP87jD?domain=limdep.itls.usyd.edu.au > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/F5DvCROND2uvQpjnriP87jD?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/96YvCVARKgCxXEN0lfJflBu?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 From christoph.buschmann at thuenen.de Thu May 20 18:03:33 2021 From: christoph.buschmann at thuenen.de (Christoph Buschmann) Date: Thu, 20 May 2021 10:03:33 +0200 (CEST) Subject: [Limdep Nlogit List] Question on missing values in N-logit In-Reply-To: References: <813758758.2526827.1621410049880.JavaMail.zimbra@thuenen.de> <2F625673-B1DE-4981-84BE-05346D280FD2@sydney.edu.au> Message-ID: <1350047485.2702441.1621497813590.JavaMail.zimbra@thuenen.de> Dear David and Bill, thank you very much for your answers. Just to be on the safe side: I understand that I should either remove the eight respondents or fill their missing data on covariates (characteristics) with the mean, median or so, because: - Coding missing covariate data with -888 is problematic, as you described Bill. - Coding missing covariate data with -999 is also problematic: that is the situation I described in my first e-mail: Per choice task one row is removed, but ?NLOGIT does not create a new choice task by removing rows from the old one? as you said Bill. Also, to be on the safe side, I send you attached a screenshot of NLogit?s report and the dataset. For example, the report says that row 1561 and row 1564 are skipped. In the dataset we see that row 1561 and 1564 describe Alternative 1 out of three Alternatives (Alternative three is Opt-out). So, it is problematic that the first alternative is skipped, but the information on the other two alternatives are still in the data set. But Nlogit does not create a new choice task out of the remaining two alternatives. Thank you again for your help. Kind regards, Christoph Buschmann ----- Urspr?ngliche Mail ----- Von: "Limdep and Nlogit Mailing List" An: "Limdep and Nlogit Mailing List" CC: "William Greene" Gesendet: Mittwoch, 19. Mai 2021 14:07:58 Betreff: Re: [Limdep Nlogit List] Question on missing values in N-logit David and all. Be careful in using the -888 form for missing data. To reiterate, if one of the 3 alts has missing values for an attribute (-999), the entire choice task has to be removed from the sample. Christoph, your interpretation is not correct. NLOGIT does not create a new choice task by removing rows from the old one. -888 does not operate on the data. It operates on the coefficient vector. For observations with x(j,k) = -888, where j is the alt and k is the characteristic, when computing beta'x(j,k) for that alt j, beta(k) is set equal to zero - not x(j,k). This matters. You don't want to set the x(j,k) to zero, especially if it is a price. Nonattendance (-888) means the marginal utility is zero, not that the attribute is zero. Cheers Bill Greene On Wed, May 19, 2021 at 4:04 AM David Hensher via Limdep < limdep at mailman.sydney.edu.au> wrote: > Missing data can exist for a number of reasons. While not being sure what > is happening here there are two codes of value. -999 will remove an entire > choice set and -888 will ignore an attribute level associated with an > alternative or individual you have in the data. So if a covariant is say > income, code as -888 to retain the choice set but note this amounts to it > being missing which can be problematic if you too many of these are missing > and it is retained in the model. Some people either replace it with the > mean or mode or run an auxiliary regression to try and predict it based on > other covariates. > David > > Sent from my iPhone > 0418 433 057 > David A Hensher > Choice modellers prefer Nlogit: See https://protect-au.mimecast.com/s/uMhOC4QOPEiB6118RhOsFga?domain=limdep.com > ITLS celebrating 30 years > > > Note: > hgroup at hensher.com.au David.hensher at bigpond.com > David.hensher at sydney.edu.au > are linked so use one only > > > On 19 May 2021, at 5:51 pm, Christoph Buschmann < > christoph.buschmann at thuenen.de> wrote: > > ?Dear group, > > I have a question on how N-logit deals with missing data in the > covariates. My choice experi-ments were carried out completely by all > respondents, but 8 respondents did not answer all questions about the > covariates. > > I have ten choice cards per respondent and 3 alternatives in each choice > situation. So, my dataset comprises three rows for each choice situation. > N-Logit reports that, for the 8 respondents in question, per choice > situation one missing value has been found and refers to the first row per > choice situation, i.e. the row that gives information about the choice of > alternative 1. > > I assume that, for the 8 respondents in question, N-logit skips one row > per choice situation (the one for alternative 1), including the information > about covariates, and keeps the other two rows (for alternatives 2 and 3). > > This would mean that for the 8 respondents in question, some of the > information is retained and so it makes sense to keep them in the data set. > In fact, if I remove the 8 respondents from the dataset by hand, the model > deteriorates (AIC). > > What is your experience with missing data? Is my interpretation correct? > > Thank you very much for your help. > > > Kind regards, > > Christoph Buschmann > > -- > Christoph Buschmann, M.Sc. > > Stabsstelle Klima/ Coordination Unit Climate > Th?nen-Institut > Bundesallee 49 > D-38116 Braunschweig > Germany > > Im Home-Office erreichbar per e-mail > E-mail: christoph.buschmann at thuenen.de > Homepage: [ https://protect-au.mimecast.com/s/7CwcC5QPXJiZRGGlwCO9o2L?domain=thuenen.de | https://protect-au.mimecast.com/s/7CwcC5QPXJiZRGGlwCO9o2L?domain=thuenen.de ] > Twitter: @ThuenenClimSoil > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/VUG_C6XQ4LfrMKKWZCmq4jK?domain=limdep.itls.usyd.edu.au > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/VUG_C6XQ4LfrMKKWZCmq4jK?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/nAXXC71R2NTAykkRnIN6c5y?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au https://protect-au.mimecast.com/s/VUG_C6XQ4LfrMKKWZCmq4jK?domain=limdep.itls.usyd.edu.au From James.Tang at health.nsw.gov.au Thu May 27 14:43:57 2021 From: James.Tang at health.nsw.gov.au (James Tang) Date: Thu, 27 May 2021 04:43:57 +0000 Subject: [Limdep Nlogit List] Nlogit and calculating choice probabilities Message-ID: <1622090638408.50708@health.nsw.gov.au> Dear Nlogit expert, I have been using Nlogit for MNL model and 2 way interactions for a model building about preferences. I wanted to know how I can get percentages of a choice of a specific attribute's level. For example, attribute mode of screen (4 levels) frequency of screen (4 levels) cost (3 levels) sharing information (3 levels) I wanted to know how many participants (percentage) picked a specific mode of screen (i.e. percentages of each of the 4 levels). How would I code this in nlogit coding? Thank you James? ? This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of NSW Health or any of its entities. From wgreene at stern.nyu.edu Fri May 28 02:00:03 2021 From: wgreene at stern.nyu.edu (William Greene) Date: Thu, 27 May 2021 12:00:03 -0400 Subject: [Limdep Nlogit List] Nlogit and calculating choice probabilities In-Reply-To: <1622090638408.50708@health.nsw.gov.au> References: <1622090638408.50708@health.nsw.gov.au> Message-ID: James. Just fit your MNL model and add ;Describe to the command. This should get you what you are looking for. /B. Greene On Thu, May 27, 2021 at 12:44 AM James Tang via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear Nlogit expert, > > I have been using Nlogit for MNL model and 2 way interactions for a model > building about preferences. > > I wanted to know how I can get percentages of a choice of a specific > attribute's level. > > For example, > > attribute > mode of screen (4 levels) > frequency of screen (4 levels) > cost (3 levels) > sharing information (3 levels) > > I wanted to know how many participants (percentage) picked a specific mode > of screen (i.e. percentages of each of the 4 levels). > > How would I code this in nlogit coding? > > Thank you > > James? > > ? > > > This message is intended for the addressee named and may contain > confidential information. If you are not the intended recipient, please > delete it and notify the sender. > > Views expressed in this message are those of the individual sender, and > are not necessarily the views of NSW Health or any of its entities. > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/GVeQC2xMQzip0GJkrcnbhgm?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/LPLyC3QNPBipWMJm3cqxT5M?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 From James.Tang at health.nsw.gov.au Fri May 28 08:16:21 2021 From: James.Tang at health.nsw.gov.au (James Tang) Date: Thu, 27 May 2021 22:16:21 +0000 Subject: [Limdep Nlogit List] Nlogit and calculating choice probabilities In-Reply-To: References: <1622090638408.50708@health.nsw.gov.au>, Message-ID: <1622153782218.36328@health.nsw.gov.au> Thanks, Prof Greene for quick reply. The output that I get is below. I'm I correct to interpret --- the percentage of mode_0 is the average of the mean of all 259 observations. (0.228+0.236)/2 x100 or would it be the chose A mode_0 +chose B mode_0 = (0.252+0.242)/2 x100? NLOGIT ;lhs = choice, cset, altij ;choices =A, B ;describe ;model: U(A) = MODE_0*MODE_0 + MODE_1*MODE_1 + MODE_2*MODE_2 + freq_0*FREQ_0 + freq_1*FREQ_1 + freq_2*FREQ_2 +cost*COST + share_0*SHARE_0 + share_1*SHARE_1/ U(B) = MODE_0*MODE_0 + MODE_1*MODE_1 + MODE_2*MODE_2 + freq_0*FREQ_0 + freq_1*FREQ_1 + freq_2*FREQ_2 +cost*COST + share_0*SHARE_0 + share_1*SHARE_1$ +-------------------------------------------------------------------------+ | Descriptive Statistics for Alternative A | | Utility Function | (CBS wt = 1.00000) | 131.0 observs. | | Coefficient | All 259.0 obs.|that chose A | | Name Value Variable | Mean Std. Dev.|Mean Std. Dev. | | ------------------- -------- | -------------------+------------------- | | MODE_0 .0149 MODE_0 | .228 .420| .252 .436 | | MODE_1 -.3864 MODE_1 | .255 .437| .267 .444 | | MODE_2 .4979 MODE_2 | .282 .451| .275 .448 | | FREQ_0 -.3217 FREQ_0 | .282 .451| .313 .465 | | FREQ_1 -.2259 FREQ_1 | .232 .423| .229 .422 | | FREQ_2 -.3758 FREQ_2 | .232 .423| .206 .406 | | COST -.1906 COST | 4.938 3.189| 3.565 2.810 | | SHARE_0 .6877 SHARE_0 | .309 .463| .359 .481 | | SHARE_1 .5938 SHARE_1 | .367 .483| .466 .501 | +-------------------------------------------------------------------------+ +-------------------------------------------------------------------------+ | Descriptive Statistics for Alternative B | | Utility Function | (CBS wt = 1.00000) | 128.0 observs. | | Coefficient | All 259.0 obs.|that chose B | | Name Value Variable | Mean Std. Dev.|Mean Std. Dev. | | ------------------- -------- | -------------------+------------------- | | MODE_0 .0149 MODE_0 | .236 .425| .242 .430 | | MODE_1 -.3864 MODE_1 | .251 .434| .195 .398 | | MODE_2 .4979 MODE_2 | .232 .423| .289 .455 | | FREQ_0 -.3217 FREQ_0 | .216 .412| .227 .420 | | FREQ_1 -.2259 FREQ_1 | .255 .437| .273 .447 | | FREQ_2 -.3758 FREQ_2 | .278 .449| .188 .392 | | COST -.1906 COST | 4.969 3.208| 3.563 2.783 | | SHARE_0 .6877 SHARE_0 | .344 .476| .375 .486 | | SHARE_1 .5938 SHARE_1 | .317 .466| .375 .486 | +-------------------------------------------------------------------------+ ________________________________________ From: Limdep on behalf of William Greene via Limdep Sent: Friday, 28 May 2021 02:00 To: Limdep and Nlogit Mailing List Cc: William Greene Subject: Re: [Limdep Nlogit List] Nlogit and calculating choice probabilities James. Just fit your MNL model and add ;Describe to the command. This should get you what you are looking for. /B. Greene On Thu, May 27, 2021 at 12:44 AM James Tang via Limdep < limdep at mailman.sydney.edu.au> wrote: > Dear Nlogit expert, > > I have been using Nlogit for MNL model and 2 way interactions for a model > building about preferences. > > I wanted to know how I can get percentages of a choice of a specific > attribute's level. > > For example, > > attribute > mode of screen (4 levels) > frequency of screen (4 levels) > cost (3 levels) > sharing information (3 levels) > > I wanted to know how many participants (percentage) picked a specific mode > of screen (i.e. percentages of each of the 4 levels). > > How would I code this in nlogit coding? > > Thank you > > James? > > ? > > > This message is intended for the addressee named and may contain > confidential information. If you are not the intended recipient, please > delete it and notify the sender. > > Views expressed in this message are those of the individual sender, and > are not necessarily the views of NSW Health or any of its entities. > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > https://protect-au.mimecast.com/s/Ty_cC6XQ4LfrZgM8KSp1uZp?domain=limdep.itls.usyd.edu.au > > -- William Greene Department of Economics, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/o2VuC71R2NTAn0yokSWbKmy?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu 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 _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au https://protect-au.mimecast.com/s/Ty_cC6XQ4LfrZgM8KSp1uZp?domain=limdep.itls.usyd.edu.au This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of NSW Health or any of its entities.