From doctorjasonong at gmail.com Thu Apr 2 13:29:03 2020 From: doctorjasonong at gmail.com (Jason Ong) Date: Thu, 2 Apr 2020 13:29:03 +1100 Subject: [Limdep Nlogit List] structure for testing interaction effects in opt out alternative Message-ID: Hi, I would like to explore the effect of sociodemographic characteristics for people choosing the opt out alternative in my DCE. When I put the interaction terms for my opt out alternative in the main model, is there a preference of just adding the sociodemographic terms (Option 1 - see text in red below) OR creating an interaction term with the opt out alternative (Option 2 - see text in red below) they actually give quite different results. *OPTION 1* NLOGIT ; Lhs = choicev, cset, altij ; Choices = A,B,C ; rpl ?; Correlated ; fcn=cost1(n), cost2(n), cost3(n),loc1(n), loc2(n), loc3(n), loc4(n), loc5(n), loc6(n), loc7(n), mode1(n), mode2(n), pers1(n), pers2(n), pers3(n), hivmed(n) ; pds=12 ; pts=10 ? 10 for tests, between 500-1000 iterations for final model (let it run overnight) ; halton ; Model: U(A,B)=cost1*cost1+cost2*cost2+cost3*cost3+loc1*loc1+loc2*loc2+loc3*loc3+loc4*loc4+loc5*loc5+loc6*loc6+loc7*loc7 +mode1*mode1+mode2*mode2+pers1*pers1+pers2*pers2+pers3*pers3+hivmed*hivmed/ U(C)=neither + young*young + male*male + second*second + eversex*eversex + evertest*evertest$ ----------------------------------------------------------------------------- Random Parameters Multinom. Logit Model Dependent variable CHOICEV Log likelihood function -2750.06883 Restricted log likelihood -3559.50382 Chi squared [ 38](P= .000) 1618.86998 Significance level .00000 McFadden Pseudo R-squared .2274011 Estimation based on N = 3240, K = 38 Inf.Cr.AIC = 5576.1 AIC/N = 1.721 --------------------------------------- Log likelihood R-sqrd R2Adj No coefficients -3559.5038 .2274 .2228 Constants only can be computed directly Use NLOGIT ;...;RHS=ONE$ At start values -2781.2815 .0112 .0054 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. = 10 Used Halton sequences in simulations. RPL model with panel has 270 groups Fixed number of obsrvs./group= 12 Number of obs.= 3240, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHOICEV| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- |Random parameters in utility functions.......................... COST1| .14649*** .04458 3.29 .0010 .05912 .23386 COST2| -.15473*** .04625 -3.35 .0008 -.24537 -.06409 COST3| -.59762*** .05661 -10.56 .0000 -.70857 -.48666 LOC1| .10425 .07731 1.35 .1775 -.04726 .25577 LOC2| .09492 .07367 1.29 .1976 -.04947 .23931 LOC3| -.22262*** .07107 -3.13 .0017 -.36191 -.08333 LOC4| -.16815** .07468 -2.25 .0244 -.31453 -.02177 LOC5| .11153 .07643 1.46 .1445 -.03827 .26132 LOC6| -.16419** .07699 -2.13 .0329 -.31508 -.01330 LOC7| .15955** .07811 2.04 .0411 .00645 .31265 MODE1| .20164*** .03570 5.65 .0000 .13166 .27161 MODE2| -.06031 .03723 -1.62 .1052 -.13328 .01265 PERS1| -.07449* .04442 -1.68 .0935 -.16154 .01256 PERS2| -.00084 .04654 -.02 .9857 -.09205 .09038 PERS3| -.06296 .06329 -.99 .3198 -.18700 .06108 HIVMED| -.16880*** .02146 -7.87 .0000 -.21086 -.12675 |Nonrandom parameters in utility functions....................... NEITHER| -1.63089*** .08338 -19.56 .0000 -1.79431 -1.46747 YOUNG| .03880 .07502 .52 .6050 -.10824 .18583 MALE| .05301 .06767 .78 .4335 -.07963 .18564 SECOND| .14195* .07599 1.87 .0617 -.00698 .29088 EVERSEX| .19307*** .06955 2.78 .0055 .05676 .32939 EVERTEST| -.05922 .07121 -.83 .4057 -.19879 .08036 |Distns. of RPs. Std.Devs or limits of triangular................ NsCOST1| .19758** .09131 2.16 .0305 .01862 .37655 NsCOST2| .21214* .11753 1.81 .0711 -.01820 .44249 NsCOST3| .49293*** .07104 6.94 .0000 .35369 .63216 NsLOC1| .24263** .11400 2.13 .0333 .01920 .46606 NsLOC2| .16908* .09025 1.87 .0610 -.00780 .34596 NsLOC3| .07511 .10010 .75 .4531 -.12109 .27130 NsLOC4| .00998 .10294 .10 .9227 -.19177 .21174 NsLOC5| .07192 .12174 .59 .5547 -.16669 .31052 NsLOC6| .20690** .10337 2.00 .0453 .00430 .40949 NsLOC7| .07074 .09646 .73 .4633 -.11831 .25979 NsMODE1| .08671* .04435 1.96 .0505 -.00021 .17363 NsMODE2| .14239*** .04517 3.15 .0016 .05386 .23092 NsPERS1| .03078 .04556 .68 .4993 -.05852 .12007 NsPERS2| .05790 .04805 1.20 .2282 -.03628 .15209 NsPERS3| .04128 .06577 .63 .5302 -.08762 .17018 NsHIVMED| .03298 .03951 .83 .4040 -.04447 .11042 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Mar 27, 2020 at 00:45:24 PM ----------------------------------------------------------------------------- or *OPTION 2* I created interaction terms here e.g. you_nei = young*neither NLOGIT ; Lhs = choicev, cset, altij ; Choices = A,B,C ; rpl ?; Correlated ; fcn=cost1(n), cost2(n), cost3(n),loc1(n), loc2(n), loc3(n), loc4(n), loc5(n), loc6(n), loc7(n), mode1(n), mode2(n), pers1(n), pers2(n), pers3(n), hivmed(n) ; pds=12 ; pts=10 ? 10 for tests, between 500-1000 iterations for final model (let it run overnight) ; halton ; Model: U(A,B)=cost1*cost1+cost2*cost2+cost3*cost3+loc1*loc1+loc2*loc2+loc3*loc3+loc4*loc4+loc5*loc5+loc6*loc6+loc7*loc7 +mode1*mode1+mode2*mode2+pers1*pers1+pers2*pers2+pers3*pers3+hivmed*hivmed/ U(C)=neither + you_nei*you_nei+ men_nei*men_nei + sec_nei*sec_nei + sex_nei*sex_nei + tes_nei*tes_nei$ ----------------------------------------------------------------------------- Random Parameters Multinom. Logit Model Dependent variable CHOICEV Log likelihood function -2750.06883 Restricted log likelihood -3559.50382 Chi squared [ 38](P= .000) 1618.86998 Significance level .00000 McFadden Pseudo R-squared .2274011 Estimation based on N = 3240, K = 38 Inf.Cr.AIC = 5576.1 AIC/N = 1.721 --------------------------------------- Log likelihood R-sqrd R2Adj No coefficients -3559.5038 .2274 .2228 Constants only can be computed directly Use NLOGIT ;...;RHS=ONE$ At start values -2781.2815 .0112 .0054 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. = 10 Used Halton sequences in simulations. RPL model with panel has 270 groups Fixed number of obsrvs./group= 12 Number of obs.= 3240, skipped 0 obs --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence CHOICEV| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- |Random parameters in utility functions.......................... COST1| .14649*** .04458 3.29 .0010 .05912 .23386 COST2| -.15473*** .04625 -3.35 .0008 -.24537 -.06409 COST3| -.59762*** .05661 -10.56 .0000 -.70857 -.48666 LOC1| .10425 .07731 1.35 .1775 -.04726 .25577 LOC2| .09492 .07367 1.29 .1976 -.04947 .23931 LOC3| -.22262*** .07107 -3.13 .0017 -.36191 -.08333 LOC4| -.16815** .07468 -2.25 .0244 -.31453 -.02177 LOC5| .11153 .07643 1.46 .1445 -.03827 .26132 LOC6| -.16419** .07699 -2.13 .0329 -.31508 -.01330 LOC7| .15955** .07811 2.04 .0411 .00645 .31265 MODE1| .20164*** .03570 5.65 .0000 .13166 .27161 MODE2| -.06031 .03723 -1.62 .1052 -.13328 .01265 PERS1| -.07449* .04442 -1.68 .0935 -.16154 .01256 PERS2| -.00084 .04654 -.02 .9857 -.09205 .09038 PERS3| -.06296 .06329 -.99 .3198 -.18700 .06108 HIVMED| -.16880*** .02146 -7.87 .0000 -.21086 -.12675 |Nonrandom parameters in utility functions....................... NEITHER| -1.63089*** .08338 -19.56 .0000 -1.79431 -1.46747 YOU_NEI| -.02335 .04515 -.52 .6050 -.11185 .06515 MEN_NEI| -.03191 .04073 -.78 .4335 -.11174 .04793 SEC_NEI| -.08544* .04574 -1.87 .0617 -.17508 .00420 SEX_NEI| -.11621*** .04186 -2.78 .0055 -.19826 -.03417 TES_NEI| .03564 .04286 .83 .4057 -.04837 .11965 |Distns. of RPs. Std.Devs or limits of triangular................ NsCOST1| .19758** .09131 2.16 .0305 .01862 .37655 NsCOST2| .21214* .11753 1.81 .0711 -.01820 .44249 NsCOST3| .49293*** .07104 6.94 .0000 .35369 .63216 NsLOC1| .24263** .11400 2.13 .0333 .01920 .46606 NsLOC2| .16908* .09025 1.87 .0610 -.00780 .34596 NsLOC3| .07511 .10010 .75 .4531 -.12109 .27130 NsLOC4| .00998 .10294 .10 .9227 -.19177 .21174 NsLOC5| .07192 .12174 .59 .5547 -.16669 .31052 NsLOC6| .20690** .10337 2.00 .0453 .00430 .40949 NsLOC7| .07074 .09646 .73 .4633 -.11831 .25979 NsMODE1| .08671* .04435 1.96 .0505 -.00021 .17363 NsMODE2| .14239*** .04517 3.15 .0016 .05386 .23092 NsPERS1| .03078 .04556 .68 .4993 -.05852 .12007 NsPERS2| .05790 .04805 1.20 .2282 -.03628 .15209 NsPERS3| .04128 .06577 .63 .5302 -.08762 .17018 NsHIVMED| .03298 .03951 .83 .4040 -.04447 .11042 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Mar 27, 2020 at 00:55:10 PM ----------------------------------------------------------------------------- thank you for your help Best, *Jason Ong* Twitter: @DrJasonJOng PhD, MMed, MBBS, FAChSHM, FRACGP Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health Associate Professor (Hon), London School of Hygiene and Tropical Medicine, UK Central Clinical School, Monash University, Australia Melbourne School of Population and Global Health, University of Melbourne, Australia Associate Editor, Sexually Transmitted Infections Special Issues Editor, Sexual Health Board Director, ASHM (www.ashm.org.au) https://protect-au.mimecast.com/s/k35-CjZ1N7iGE2E6SWRL9Z?domain=lshtm.ac.uk https://protect-au.mimecast.com/s/7TjeCk81N9tXl7lLcVYPnY?domain=researchgate.net From megoded at us.es Sun Apr 5 02:17:21 2020 From: megoded at us.es (=?iso-8859-1?Q?MAR=CDA_DEL_PILAR_ESPINOSA_GODED?=) Date: Sat, 4 Apr 2020 16:17:21 +0000 Subject: [Limdep Nlogit List] GMXL in WTP space (optimization) Message-ID: Hello, i am trying to run a GMXL model in the WTP space and i get the following message Line search at iteration 1 does not improve the function Exiting optimization With < 4 iterations, this may not be a good solution to the optimization. The log-likelihood is flat. Try refit- ting with ;Output=3 and examining the derivatives." Below the model. (when i run the model without fixing the payment atribute to 1 - i do not get the message). Thanks!!!!! |-> GMXLOGIT; Lhs = Choice_L ; Choices=1,2; ;fcn=AUSED(N|#0001000),AUSEA(N|#1000000), ANAT(N|#0001111), AFAFA(N|#0011000),AFAFO(N|#0001000),AINF(N|#0101010) ,AWOD(N|#0010000), AWOA(N|#0010000), ABUFA (N|#0000101),ABUFO(*N|#0000000) ;rpl=region1, etud_b, female, ageC,seniorz0, seniorz1, seniorz2 ; model: U(1,2) = AUSED *Cuse_d_A + AUSEA*Cuse_a_A + ANAT*Cnatu_A + AFAFA *Cfa_fa_A + AFAFO*Cfa_fo_A + AINF*Cinfo_A + AWOD* Cwo_d_A + AWOA* Cwo_a_A +ABUFA* Cbu_fa_A+ ABUFO* Cbu_fo_A ; Pds = 9 ; Draws=75;parameters; maxit=400$ From elin.spegel at ltu.se Mon Apr 6 16:22:10 2020 From: elin.spegel at ltu.se (Elin Spegel) Date: Mon, 6 Apr 2020 06:22:10 +0000 Subject: [Limdep Nlogit List] unsubscription Message-ID: Heloo. I would like to unsubscribe from this list. Kind regards/ Elin Spegel, mail: elin.spegel at ltu.se From mariasosadios at yahoo.com Tue Apr 7 03:04:36 2020 From: mariasosadios at yahoo.com (=?UTF-8?Q?Mar=C3=ADa_Sosa_Dios?=) Date: Mon, 6 Apr 2020 17:04:36 +0000 (UTC) Subject: [Limdep Nlogit List] unsubscription In-Reply-To: References: Message-ID: <272235589.1350214.1586192676148@mail.yahoo.com> Please, I would like to unsubscribe as well Thank you very much Kind Regards, Maria On Monday, April 6, 2020, 02:26:24 AM GMT-4, Elin Spegel via Limdep wrote: Heloo. I would like to unsubscribe from this list. Kind regards/ Elin Spegel, mail: elin.spegel at ltu.se _______________________________________________ Limdep site list Limdep at mailman.sydney.edu.au http://limdep.itls.usyd.edu.au From ext.ioanna.grammatikopoulou at luke.fi Wed Apr 8 19:12:45 2020 From: ext.ioanna.grammatikopoulou at luke.fi (Grammatikopoulou Ioanna (Luke)) Date: Wed, 8 Apr 2020 09:12:45 +0000 Subject: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par Message-ID: Hi! I am trying to impose starting values on a mixed logit with correlated parameters models. As starting values I would like to use the values I get from the mixed logit model. What I get though when I try to run the model is the following errors: Error 71: Variable list contains a name not in the expected table. Error 539: Variable list: The unidentifiable string is NF Error 1085: Unidentified name found in NF) Why is that? Here is the syntax I am using: ? NLOGIT;Lhs=CHOICE ;Choices=NF,SQ,MF ;pds=8 ;Labels=b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15 ;Start=-3.5,0.07,0.09,0.04,0.09,-0.009,0.08,0.02,0.19,-0.06,0.54,-0.04,0.15,-0.07,0.46,-0.001 ;rpl ;corr ;pts=10 ;Fcn=b0(N),b1(N),b2(N),b3(N),b4(N),b5(N),b6(N),b7(N),b8(N),b9(N),b10(N),b11(N),b12(N),b13(N),b14(N),b15(N) ;parameters ;Halton ;Model: U(NF)=b0*BAU+ b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ U(SQ)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ U(MF)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE$ Thank you in advance. Ioanna Grammatikopoulou Ioanna Grammatikopoulou Post-doc fellow Natural Resources Institute Finland (LUKE) Latokartanonkaari 9 FI-00790, Helsinki FINLAND E-mail: ioanna.grammatikopoulou at luke.fi Ioanna Grammatikopoulou Post-doc fellow Natural Resource Institute Finland (LUKE) Latokartanonkaari 9 FI-00790, Helsinki Finland E-mail: ioanna.grammatikopoulou at luke.fi From wgreene at stern.nyu.edu Thu Apr 9 03:43:07 2020 From: wgreene at stern.nyu.edu (William Greene) Date: Wed, 8 Apr 2020 13:43:07 -0400 Subject: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par In-Reply-To: References: Message-ID: You can provide starting values for the named parameters in the choice model by adding them to the first occurrences in the definitions of the model. For example, ;Model: U(NF)=b0(-3.5)*BAU (Delete the ;labels setting) However, I'm not sure what this implies for the associated variance parameters. (1) 15 random parameters in a model like this will be very optimistic. This is a difficult log likelihood to maximize in any event. This specification may be excessive. (2) You might want to try fitting the model with your desired starting values and fixed (nonrandom) parameters, just to find a setup that works initially. Regards, William Greene On Wed, Apr 8, 2020 at 5:13 AM Grammatikopoulou Ioanna (Luke) < ext.ioanna.grammatikopoulou at luke.fi> wrote: > Hi! > I am trying to impose starting values on a mixed logit with correlated > parameters models. As starting values I would like to use the values I get > from the mixed logit model. What I get though when I try to run the model > is the following errors: > > Error 71: Variable list contains a name not in the expected table. > Error 539: Variable list: The unidentifiable string is NF > Error 1085: Unidentified name found in NF) > > Why is that? > > Here is the syntax I am using: > > NLOGIT;Lhs=CHOICE > ;Choices=NF,SQ,MF > ;pds=8 > ;Labels=b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15 > > ;Start=-3.5,0.07,0.09,0.04,0.09,-0.009,0.08,0.02,0.19,-0.06,0.54,-0.04,0.15,-0.07,0.46,-0.001 > ;rpl > ;corr > ;pts=10 > > ;Fcn=b0(N),b1(N),b2(N),b3(N),b4(N),b5(N),b6(N),b7(N),b8(N),b9(N),b10(N),b11(N),b12(N),b13(N),b14(N),b15(N) > ;parameters > ;Halton > ;Model: > U(NF)=b0*BAU+ > b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > U(SQ)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > U(MF)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE$ > > > Thank you in advance. > > Ioanna Grammatikopoulou > > > Ioanna Grammatikopoulou > Post-doc fellow > Natural Resources Institute Finland (LUKE) > Latokartanonkaari 9 > FI-00790, Helsinki > FINLAND > E-mail: ioanna.grammatikopoulou at luke.fi ioanna.grammatikopoulou at mtt.fi> > > > Ioanna Grammatikopoulou > Post-doc fellow > Natural Resource Institute Finland (LUKE) > Latokartanonkaari 9 > FI-00790, Helsinki > Finland > E-mail: ioanna.grammatikopoulou at luke.fi ioanna.grammatikopoulou at luke.fi> > > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://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/97VtC5QPXJixy2lguz8mqC?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 doctorjasonong at gmail.com Thu Apr 9 09:11:03 2020 From: doctorjasonong at gmail.com (Jason Ong) Date: Thu, 9 Apr 2020 09:11:03 +1000 Subject: [Limdep Nlogit List] structure for testing interaction effects in opt out alternative In-Reply-To: References: Message-ID: hello, I am not sure if my post below got through? thanks for your help Best, *Jason Ong* Twitter: @DrJasonJOng PhD, MMed, MBBS, FAChSHM, FRACGP Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health Associate Professor (Hon), London School of Hygiene and Tropical Medicine, UK Central Clinical School, Monash University, Australia Melbourne School of Population and Global Health, University of Melbourne, Australia Associate Editor, Sexually Transmitted Infections Special Issues Editor, Sexual Health Board Director, ASHM (https://protect-au.mimecast.com/s/93FXClx1NjiA0QwjhGnh4B?domain=ashm.org.au) https://protect-au.mimecast.com/s/tJsoCmO5gluABOo9hOQWmB?domain=lshtm.ac.uk https://protect-au.mimecast.com/s/wWzCCnx1jniKA8ZkhNgHP9?domain=researchgate.net On Thu, Apr 2, 2020 at 1:29 PM Jason Ong wrote: > Hi, > > I would like to explore the effect of sociodemographic characteristics for > people choosing the opt out alternative in my DCE. > When I put the interaction terms for my opt out alternative in the main > model, is there a preference of just adding the sociodemographic terms > (Option 1 - see text in red below) OR creating an interaction term with the > opt out alternative (Option 2 - see text in red below) > they actually give quite different results. > > *OPTION 1* > NLOGIT > ; Lhs = choicev, cset, altij > ; Choices = A,B,C > ; rpl > ?; Correlated > ; fcn=cost1(n), cost2(n), cost3(n),loc1(n), loc2(n), loc3(n), loc4(n), > loc5(n), loc6(n), loc7(n), mode1(n), mode2(n), pers1(n), pers2(n), > pers3(n), hivmed(n) > ; pds=12 > ; pts=10 > ? 10 for tests, between 500-1000 iterations for final model (let it run > overnight) > ; halton > ; Model: > > U(A,B)=cost1*cost1+cost2*cost2+cost3*cost3+loc1*loc1+loc2*loc2+loc3*loc3+loc4*loc4+loc5*loc5+loc6*loc6+loc7*loc7 > +mode1*mode1+mode2*mode2+pers1*pers1+pers2*pers2+pers3*pers3+hivmed*hivmed/ > U(C)=neither + young*young + male*male + second*second + eversex*eversex + > evertest*evertest$ > > > ----------------------------------------------------------------------------- > Random Parameters Multinom. Logit Model > Dependent variable CHOICEV > Log likelihood function -2750.06883 > Restricted log likelihood -3559.50382 > Chi squared [ 38](P= .000) 1618.86998 > Significance level .00000 > McFadden Pseudo R-squared .2274011 > Estimation based on N = 3240, K = 38 > Inf.Cr.AIC = 5576.1 AIC/N = 1.721 > --------------------------------------- > Log likelihood R-sqrd R2Adj > No coefficients -3559.5038 .2274 .2228 > Constants only can be computed directly > Use NLOGIT ;...;RHS=ONE$ > At start values -2781.2815 .0112 .0054 > 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. = 10 > Used Halton sequences in simulations. > RPL model with panel has 270 groups > Fixed number of obsrvs./group= 12 > Number of obs.= 3240, skipped 0 obs > > --------+-------------------------------------------------------------------- > | Standard Prob. 95% Confidence > CHOICEV| Coefficient Error z |z|>Z* Interval > > --------+-------------------------------------------------------------------- > |Random parameters in utility functions.......................... > COST1| .14649*** .04458 3.29 .0010 .05912 .23386 > COST2| -.15473*** .04625 -3.35 .0008 -.24537 -.06409 > COST3| -.59762*** .05661 -10.56 .0000 -.70857 -.48666 > LOC1| .10425 .07731 1.35 .1775 -.04726 .25577 > LOC2| .09492 .07367 1.29 .1976 -.04947 .23931 > LOC3| -.22262*** .07107 -3.13 .0017 -.36191 -.08333 > LOC4| -.16815** .07468 -2.25 .0244 -.31453 -.02177 > LOC5| .11153 .07643 1.46 .1445 -.03827 .26132 > LOC6| -.16419** .07699 -2.13 .0329 -.31508 -.01330 > LOC7| .15955** .07811 2.04 .0411 .00645 .31265 > MODE1| .20164*** .03570 5.65 .0000 .13166 .27161 > MODE2| -.06031 .03723 -1.62 .1052 -.13328 .01265 > PERS1| -.07449* .04442 -1.68 .0935 -.16154 .01256 > PERS2| -.00084 .04654 -.02 .9857 -.09205 .09038 > PERS3| -.06296 .06329 -.99 .3198 -.18700 .06108 > HIVMED| -.16880*** .02146 -7.87 .0000 -.21086 -.12675 > |Nonrandom parameters in utility functions....................... > NEITHER| -1.63089*** .08338 -19.56 .0000 -1.79431 -1.46747 > YOUNG| .03880 .07502 .52 .6050 -.10824 .18583 > MALE| .05301 .06767 .78 .4335 -.07963 .18564 > SECOND| .14195* .07599 1.87 .0617 -.00698 .29088 > EVERSEX| .19307*** .06955 2.78 .0055 .05676 .32939 > EVERTEST| -.05922 .07121 -.83 .4057 -.19879 .08036 > |Distns. of RPs. Std.Devs or limits of triangular................ > NsCOST1| .19758** .09131 2.16 .0305 .01862 .37655 > NsCOST2| .21214* .11753 1.81 .0711 -.01820 .44249 > NsCOST3| .49293*** .07104 6.94 .0000 .35369 .63216 > NsLOC1| .24263** .11400 2.13 .0333 .01920 .46606 > NsLOC2| .16908* .09025 1.87 .0610 -.00780 .34596 > NsLOC3| .07511 .10010 .75 .4531 -.12109 .27130 > NsLOC4| .00998 .10294 .10 .9227 -.19177 .21174 > NsLOC5| .07192 .12174 .59 .5547 -.16669 .31052 > NsLOC6| .20690** .10337 2.00 .0453 .00430 .40949 > NsLOC7| .07074 .09646 .73 .4633 -.11831 .25979 > NsMODE1| .08671* .04435 1.96 .0505 -.00021 .17363 > NsMODE2| .14239*** .04517 3.15 .0016 .05386 .23092 > NsPERS1| .03078 .04556 .68 .4993 -.05852 .12007 > NsPERS2| .05790 .04805 1.20 .2282 -.03628 .15209 > NsPERS3| .04128 .06577 .63 .5302 -.08762 .17018 > NsHIVMED| .03298 .03951 .83 .4040 -.04447 .11042 > > --------+-------------------------------------------------------------------- > ***, **, * ==> Significance at 1%, 5%, 10% level. > Model was estimated on Mar 27, 2020 at 00:45:24 PM > > ----------------------------------------------------------------------------- > > or > > *OPTION 2* > I created interaction terms here > e.g. you_nei = young*neither > > NLOGIT > ; Lhs = choicev, cset, altij > ; Choices = A,B,C > ; rpl > ?; Correlated > ; fcn=cost1(n), cost2(n), cost3(n),loc1(n), loc2(n), loc3(n), loc4(n), > loc5(n), loc6(n), loc7(n), mode1(n), mode2(n), pers1(n), pers2(n), > pers3(n), hivmed(n) > ; pds=12 > ; pts=10 > ? 10 for tests, between 500-1000 iterations for final model (let it run > overnight) > ; halton > ; Model: > > U(A,B)=cost1*cost1+cost2*cost2+cost3*cost3+loc1*loc1+loc2*loc2+loc3*loc3+loc4*loc4+loc5*loc5+loc6*loc6+loc7*loc7 > +mode1*mode1+mode2*mode2+pers1*pers1+pers2*pers2+pers3*pers3+hivmed*hivmed/ > U(C)=neither + you_nei*you_nei+ men_nei*men_nei + sec_nei*sec_nei + > sex_nei*sex_nei + tes_nei*tes_nei$ > > > ----------------------------------------------------------------------------- > Random Parameters Multinom. Logit Model > Dependent variable CHOICEV > Log likelihood function -2750.06883 > Restricted log likelihood -3559.50382 > Chi squared [ 38](P= .000) 1618.86998 > Significance level .00000 > McFadden Pseudo R-squared .2274011 > Estimation based on N = 3240, K = 38 > Inf.Cr.AIC = 5576.1 AIC/N = 1.721 > --------------------------------------- > Log likelihood R-sqrd R2Adj > No coefficients -3559.5038 .2274 .2228 > Constants only can be computed directly > Use NLOGIT ;...;RHS=ONE$ > At start values -2781.2815 .0112 .0054 > 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. = 10 > Used Halton sequences in simulations. > RPL model with panel has 270 groups > Fixed number of obsrvs./group= 12 > Number of obs.= 3240, skipped 0 obs > > --------+-------------------------------------------------------------------- > | Standard Prob. 95% Confidence > CHOICEV| Coefficient Error z |z|>Z* Interval > > --------+-------------------------------------------------------------------- > |Random parameters in utility functions.......................... > COST1| .14649*** .04458 3.29 .0010 .05912 .23386 > COST2| -.15473*** .04625 -3.35 .0008 -.24537 -.06409 > COST3| -.59762*** .05661 -10.56 .0000 -.70857 -.48666 > LOC1| .10425 .07731 1.35 .1775 -.04726 .25577 > LOC2| .09492 .07367 1.29 .1976 -.04947 .23931 > LOC3| -.22262*** .07107 -3.13 .0017 -.36191 -.08333 > LOC4| -.16815** .07468 -2.25 .0244 -.31453 -.02177 > LOC5| .11153 .07643 1.46 .1445 -.03827 .26132 > LOC6| -.16419** .07699 -2.13 .0329 -.31508 -.01330 > LOC7| .15955** .07811 2.04 .0411 .00645 .31265 > MODE1| .20164*** .03570 5.65 .0000 .13166 .27161 > MODE2| -.06031 .03723 -1.62 .1052 -.13328 .01265 > PERS1| -.07449* .04442 -1.68 .0935 -.16154 .01256 > PERS2| -.00084 .04654 -.02 .9857 -.09205 .09038 > PERS3| -.06296 .06329 -.99 .3198 -.18700 .06108 > HIVMED| -.16880*** .02146 -7.87 .0000 -.21086 -.12675 > |Nonrandom parameters in utility functions....................... > NEITHER| -1.63089*** .08338 -19.56 .0000 -1.79431 -1.46747 > YOU_NEI| -.02335 .04515 -.52 .6050 -.11185 .06515 > MEN_NEI| -.03191 .04073 -.78 .4335 -.11174 .04793 > SEC_NEI| -.08544* .04574 -1.87 .0617 -.17508 .00420 > SEX_NEI| -.11621*** .04186 -2.78 .0055 -.19826 -.03417 > TES_NEI| .03564 .04286 .83 .4057 -.04837 .11965 > |Distns. of RPs. Std.Devs or limits of triangular................ > NsCOST1| .19758** .09131 2.16 .0305 .01862 .37655 > NsCOST2| .21214* .11753 1.81 .0711 -.01820 .44249 > NsCOST3| .49293*** .07104 6.94 .0000 .35369 .63216 > NsLOC1| .24263** .11400 2.13 .0333 .01920 .46606 > NsLOC2| .16908* .09025 1.87 .0610 -.00780 .34596 > NsLOC3| .07511 .10010 .75 .4531 -.12109 .27130 > NsLOC4| .00998 .10294 .10 .9227 -.19177 .21174 > NsLOC5| .07192 .12174 .59 .5547 -.16669 .31052 > NsLOC6| .20690** .10337 2.00 .0453 .00430 .40949 > NsLOC7| .07074 .09646 .73 .4633 -.11831 .25979 > NsMODE1| .08671* .04435 1.96 .0505 -.00021 .17363 > NsMODE2| .14239*** .04517 3.15 .0016 .05386 .23092 > NsPERS1| .03078 .04556 .68 .4993 -.05852 .12007 > NsPERS2| .05790 .04805 1.20 .2282 -.03628 .15209 > NsPERS3| .04128 .06577 .63 .5302 -.08762 .17018 > NsHIVMED| .03298 .03951 .83 .4040 -.04447 .11042 > > --------+-------------------------------------------------------------------- > ***, **, * ==> Significance at 1%, 5%, 10% level. > Model was estimated on Mar 27, 2020 at 00:55:10 PM > > ----------------------------------------------------------------------------- > > thank you for your help > > Best, > > *Jason Ong* > Twitter: @DrJasonJOng > PhD, MMed, MBBS, FAChSHM, FRACGP > > Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health > Associate Professor (Hon), London School of Hygiene and Tropical Medicine, > UK > Central Clinical School, Monash University, Australia > Melbourne School of Population and Global Health, University of Melbourne, > Australia > Associate Editor, Sexually Transmitted Infections > Special Issues Editor, Sexual Health > Board Director, ASHM (https://protect-au.mimecast.com/s/93FXClx1NjiA0QwjhGnh4B?domain=ashm.org.au) > https://protect-au.mimecast.com/s/tJsoCmO5gluABOo9hOQWmB?domain=lshtm.ac.uk > https://protect-au.mimecast.com/s/wWzCCnx1jniKA8ZkhNgHP9?domain=researchgate.net > > > From ext.ioanna.grammatikopoulou at luke.fi Thu Apr 9 19:12:20 2020 From: ext.ioanna.grammatikopoulou at luke.fi (Grammatikopoulou Ioanna (Luke)) Date: Thu, 9 Apr 2020 09:12:20 +0000 Subject: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par In-Reply-To: References: , Message-ID: <1586423541535.75002@luke.fi> Dear William I tried your suggestion but the model doesnt run (I get error 1085). Thank you so much for your suggestions. Ioanna Ioanna Grammatikopoulou Post-doc fellow Natural Resources Institute Finland (LUKE) Latokartanonkaari 9 FI-00790, Helsinki FINLAND E-mail: ioanna.grammatikopoulou at luke.fi ________________________________________ From: Limdep on behalf of William Greene via Limdep Sent: Wednesday, April 8, 2020 8:43 PM To: Limdep and Nlogit Mailing List Cc: William Greene Subject: Re: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par You can provide starting values for the named parameters in the choice model by adding them to the first occurrences in the definitions of the model. For example, ;Model: U(NF)=b0(-3.5)*BAU (Delete the ;labels setting) However, I'm not sure what this implies for the associated variance parameters. (1) 15 random parameters in a model like this will be very optimistic. This is a difficult log likelihood to maximize in any event. This specification may be excessive. (2) You might want to try fitting the model with your desired starting values and fixed (nonrandom) parameters, just to find a setup that works initially. Regards, William Greene On Wed, Apr 8, 2020 at 5:13 AM Grammatikopoulou Ioanna (Luke) < ext.ioanna.grammatikopoulou at luke.fi> wrote: > Hi! > I am trying to impose starting values on a mixed logit with correlated > parameters models. As starting values I would like to use the values I get > from the mixed logit model. What I get though when I try to run the model > is the following errors: > > Error 71: Variable list contains a name not in the expected table. > Error 539: Variable list: The unidentifiable string is NF > Error 1085: Unidentified name found in NF) > > Why is that? > > Here is the syntax I am using: > > NLOGIT;Lhs=CHOICE > ;Choices=NF,SQ,MF > ;pds=8 > ;Labels=b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15 > > ;Start=-3.5,0.07,0.09,0.04,0.09,-0.009,0.08,0.02,0.19,-0.06,0.54,-0.04,0.15,-0.07,0.46,-0.001 > ;rpl > ;corr > ;pts=10 > > ;Fcn=b0(N),b1(N),b2(N),b3(N),b4(N),b5(N),b6(N),b7(N),b8(N),b9(N),b10(N),b11(N),b12(N),b13(N),b14(N),b15(N) > ;parameters > ;Halton > ;Model: > U(NF)=b0*BAU+ > b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > U(SQ)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > U(MF)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE$ > > > Thank you in advance. > > Ioanna Grammatikopoulou > > > Ioanna Grammatikopoulou > Post-doc fellow > Natural Resources Institute Finland (LUKE) > Latokartanonkaari 9 > FI-00790, Helsinki > FINLAND > E-mail: ioanna.grammatikopoulou at luke.fi ioanna.grammatikopoulou at mtt.fi> > > > Ioanna Grammatikopoulou > Post-doc fellow > Natural Resource Institute Finland (LUKE) > Latokartanonkaari 9 > FI-00790, Helsinki > Finland > E-mail: ioanna.grammatikopoulou at luke.fi ioanna.grammatikopoulou at luke.fi> > > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://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/oLn4CwV1vMfAxWPOiVLVpR?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 http://limdep.itls.usyd.edu.au From wgreene at stern.nyu.edu Thu Apr 9 23:40:38 2020 From: wgreene at stern.nyu.edu (William Greene) Date: Thu, 9 Apr 2020 09:40:38 -0400 Subject: [Limdep Nlogit List] structure for testing interaction effects in opt out alternative In-Reply-To: References: Message-ID: Jason. It's not clear to me how the models you show below are revealing information about the effect of the demographics on the probability of "OptOut." That is, what specific results show what you are looking for? That said, however, given the way you have phrased the question, I would have built this as a binary choice model in which the dependent variable equals 1 if choose OptOut and 0 if something else. The partial effects from that model would address what are looking for. /Bill Greene On Wed, Apr 8, 2020 at 7:11 PM Jason Ong via Limdep < limdep at mailman.sydney.edu.au> wrote: > hello, > > I am not sure if my post below got through? > thanks for your help > > Best, > > > *Jason Ong* > Twitter: @DrJasonJOng > PhD, MMed, MBBS, FAChSHM, FRACGP > > Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health > Associate Professor (Hon), London School of Hygiene and Tropical Medicine, > UK > Central Clinical School, Monash University, Australia > Melbourne School of Population and Global Health, University of Melbourne, > Australia > Associate Editor, Sexually Transmitted Infections > Special Issues Editor, Sexual Health > Board Director, ASHM (https://protect-au.mimecast.com/s/novECQnMBZf3q1O5HxvjgM?domain=ashm.org.au) > https://protect-au.mimecast.com/s/D_P9CROND2uRlDEAfNc7N9?domain=lshtm.ac.uk > https://protect-au.mimecast.com/s/JoYICVARKgC5Kn3Dfy0wxD?domain=researchgate.net > > > > > On Thu, Apr 2, 2020 at 1:29 PM Jason Ong wrote: > > > Hi, > > > > I would like to explore the effect of sociodemographic characteristics > for > > people choosing the opt out alternative in my DCE. > > When I put the interaction terms for my opt out alternative in the main > > model, is there a preference of just adding the sociodemographic terms > > (Option 1 - see text in red below) OR creating an interaction term with > the > > opt out alternative (Option 2 - see text in red below) > > they actually give quite different results. > > > > *OPTION 1* > > NLOGIT > > ; Lhs = choicev, cset, altij > > ; Choices = A,B,C > > ; rpl > > ?; Correlated > > ; fcn=cost1(n), cost2(n), cost3(n),loc1(n), loc2(n), loc3(n), loc4(n), > > loc5(n), loc6(n), loc7(n), mode1(n), mode2(n), pers1(n), pers2(n), > > pers3(n), hivmed(n) > > ; pds=12 > > ; pts=10 > > ? 10 for tests, between 500-1000 iterations for final model (let it run > > overnight) > > ; halton > > ; Model: > > > > > U(A,B)=cost1*cost1+cost2*cost2+cost3*cost3+loc1*loc1+loc2*loc2+loc3*loc3+loc4*loc4+loc5*loc5+loc6*loc6+loc7*loc7 > > > +mode1*mode1+mode2*mode2+pers1*pers1+pers2*pers2+pers3*pers3+hivmed*hivmed/ > > U(C)=neither + young*young + male*male + second*second + eversex*eversex > + > > evertest*evertest$ > > > > > > > ----------------------------------------------------------------------------- > > Random Parameters Multinom. Logit Model > > Dependent variable CHOICEV > > Log likelihood function -2750.06883 > > Restricted log likelihood -3559.50382 > > Chi squared [ 38](P= .000) 1618.86998 > > Significance level .00000 > > McFadden Pseudo R-squared .2274011 > > Estimation based on N = 3240, K = 38 > > Inf.Cr.AIC = 5576.1 AIC/N = 1.721 > > --------------------------------------- > > Log likelihood R-sqrd R2Adj > > No coefficients -3559.5038 .2274 .2228 > > Constants only can be computed directly > > Use NLOGIT ;...;RHS=ONE$ > > At start values -2781.2815 .0112 .0054 > > 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. = 10 > > Used Halton sequences in simulations. > > RPL model with panel has 270 groups > > Fixed number of obsrvs./group= 12 > > Number of obs.= 3240, skipped 0 obs > > > > > --------+-------------------------------------------------------------------- > > | Standard Prob. 95% Confidence > > CHOICEV| Coefficient Error z |z|>Z* Interval > > > > > --------+-------------------------------------------------------------------- > > |Random parameters in utility functions.......................... > > COST1| .14649*** .04458 3.29 .0010 .05912 .23386 > > COST2| -.15473*** .04625 -3.35 .0008 -.24537 -.06409 > > COST3| -.59762*** .05661 -10.56 .0000 -.70857 -.48666 > > LOC1| .10425 .07731 1.35 .1775 -.04726 .25577 > > LOC2| .09492 .07367 1.29 .1976 -.04947 .23931 > > LOC3| -.22262*** .07107 -3.13 .0017 -.36191 -.08333 > > LOC4| -.16815** .07468 -2.25 .0244 -.31453 -.02177 > > LOC5| .11153 .07643 1.46 .1445 -.03827 .26132 > > LOC6| -.16419** .07699 -2.13 .0329 -.31508 -.01330 > > LOC7| .15955** .07811 2.04 .0411 .00645 .31265 > > MODE1| .20164*** .03570 5.65 .0000 .13166 .27161 > > MODE2| -.06031 .03723 -1.62 .1052 -.13328 .01265 > > PERS1| -.07449* .04442 -1.68 .0935 -.16154 .01256 > > PERS2| -.00084 .04654 -.02 .9857 -.09205 .09038 > > PERS3| -.06296 .06329 -.99 .3198 -.18700 .06108 > > HIVMED| -.16880*** .02146 -7.87 .0000 -.21086 -.12675 > > |Nonrandom parameters in utility functions....................... > > NEITHER| -1.63089*** .08338 -19.56 .0000 -1.79431 -1.46747 > > YOUNG| .03880 .07502 .52 .6050 -.10824 .18583 > > MALE| .05301 .06767 .78 .4335 -.07963 .18564 > > SECOND| .14195* .07599 1.87 .0617 -.00698 .29088 > > EVERSEX| .19307*** .06955 2.78 .0055 .05676 .32939 > > EVERTEST| -.05922 .07121 -.83 .4057 -.19879 .08036 > > |Distns. of RPs. Std.Devs or limits of triangular................ > > NsCOST1| .19758** .09131 2.16 .0305 .01862 .37655 > > NsCOST2| .21214* .11753 1.81 .0711 -.01820 .44249 > > NsCOST3| .49293*** .07104 6.94 .0000 .35369 .63216 > > NsLOC1| .24263** .11400 2.13 .0333 .01920 .46606 > > NsLOC2| .16908* .09025 1.87 .0610 -.00780 .34596 > > NsLOC3| .07511 .10010 .75 .4531 -.12109 .27130 > > NsLOC4| .00998 .10294 .10 .9227 -.19177 .21174 > > NsLOC5| .07192 .12174 .59 .5547 -.16669 .31052 > > NsLOC6| .20690** .10337 2.00 .0453 .00430 .40949 > > NsLOC7| .07074 .09646 .73 .4633 -.11831 .25979 > > NsMODE1| .08671* .04435 1.96 .0505 -.00021 .17363 > > NsMODE2| .14239*** .04517 3.15 .0016 .05386 .23092 > > NsPERS1| .03078 .04556 .68 .4993 -.05852 .12007 > > NsPERS2| .05790 .04805 1.20 .2282 -.03628 .15209 > > NsPERS3| .04128 .06577 .63 .5302 -.08762 .17018 > > NsHIVMED| .03298 .03951 .83 .4040 -.04447 .11042 > > > > > --------+-------------------------------------------------------------------- > > ***, **, * ==> Significance at 1%, 5%, 10% level. > > Model was estimated on Mar 27, 2020 at 00:45:24 PM > > > > > ----------------------------------------------------------------------------- > > > > or > > > > *OPTION 2* > > I created interaction terms here > > e.g. you_nei = young*neither > > > > NLOGIT > > ; Lhs = choicev, cset, altij > > ; Choices = A,B,C > > ; rpl > > ?; Correlated > > ; fcn=cost1(n), cost2(n), cost3(n),loc1(n), loc2(n), loc3(n), loc4(n), > > loc5(n), loc6(n), loc7(n), mode1(n), mode2(n), pers1(n), pers2(n), > > pers3(n), hivmed(n) > > ; pds=12 > > ; pts=10 > > ? 10 for tests, between 500-1000 iterations for final model (let it run > > overnight) > > ; halton > > ; Model: > > > > > U(A,B)=cost1*cost1+cost2*cost2+cost3*cost3+loc1*loc1+loc2*loc2+loc3*loc3+loc4*loc4+loc5*loc5+loc6*loc6+loc7*loc7 > > > +mode1*mode1+mode2*mode2+pers1*pers1+pers2*pers2+pers3*pers3+hivmed*hivmed/ > > U(C)=neither + you_nei*you_nei+ men_nei*men_nei + sec_nei*sec_nei + > > sex_nei*sex_nei + tes_nei*tes_nei$ > > > > > > > ----------------------------------------------------------------------------- > > Random Parameters Multinom. Logit Model > > Dependent variable CHOICEV > > Log likelihood function -2750.06883 > > Restricted log likelihood -3559.50382 > > Chi squared [ 38](P= .000) 1618.86998 > > Significance level .00000 > > McFadden Pseudo R-squared .2274011 > > Estimation based on N = 3240, K = 38 > > Inf.Cr.AIC = 5576.1 AIC/N = 1.721 > > --------------------------------------- > > Log likelihood R-sqrd R2Adj > > No coefficients -3559.5038 .2274 .2228 > > Constants only can be computed directly > > Use NLOGIT ;...;RHS=ONE$ > > At start values -2781.2815 .0112 .0054 > > 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. = 10 > > Used Halton sequences in simulations. > > RPL model with panel has 270 groups > > Fixed number of obsrvs./group= 12 > > Number of obs.= 3240, skipped 0 obs > > > > > --------+-------------------------------------------------------------------- > > | Standard Prob. 95% Confidence > > CHOICEV| Coefficient Error z |z|>Z* Interval > > > > > --------+-------------------------------------------------------------------- > > |Random parameters in utility functions.......................... > > COST1| .14649*** .04458 3.29 .0010 .05912 .23386 > > COST2| -.15473*** .04625 -3.35 .0008 -.24537 -.06409 > > COST3| -.59762*** .05661 -10.56 .0000 -.70857 -.48666 > > LOC1| .10425 .07731 1.35 .1775 -.04726 .25577 > > LOC2| .09492 .07367 1.29 .1976 -.04947 .23931 > > LOC3| -.22262*** .07107 -3.13 .0017 -.36191 -.08333 > > LOC4| -.16815** .07468 -2.25 .0244 -.31453 -.02177 > > LOC5| .11153 .07643 1.46 .1445 -.03827 .26132 > > LOC6| -.16419** .07699 -2.13 .0329 -.31508 -.01330 > > LOC7| .15955** .07811 2.04 .0411 .00645 .31265 > > MODE1| .20164*** .03570 5.65 .0000 .13166 .27161 > > MODE2| -.06031 .03723 -1.62 .1052 -.13328 .01265 > > PERS1| -.07449* .04442 -1.68 .0935 -.16154 .01256 > > PERS2| -.00084 .04654 -.02 .9857 -.09205 .09038 > > PERS3| -.06296 .06329 -.99 .3198 -.18700 .06108 > > HIVMED| -.16880*** .02146 -7.87 .0000 -.21086 -.12675 > > |Nonrandom parameters in utility functions....................... > > NEITHER| -1.63089*** .08338 -19.56 .0000 -1.79431 -1.46747 > > YOU_NEI| -.02335 .04515 -.52 .6050 -.11185 .06515 > > MEN_NEI| -.03191 .04073 -.78 .4335 -.11174 .04793 > > SEC_NEI| -.08544* .04574 -1.87 .0617 -.17508 .00420 > > SEX_NEI| -.11621*** .04186 -2.78 .0055 -.19826 -.03417 > > TES_NEI| .03564 .04286 .83 .4057 -.04837 .11965 > > |Distns. of RPs. Std.Devs or limits of triangular................ > > NsCOST1| .19758** .09131 2.16 .0305 .01862 .37655 > > NsCOST2| .21214* .11753 1.81 .0711 -.01820 .44249 > > NsCOST3| .49293*** .07104 6.94 .0000 .35369 .63216 > > NsLOC1| .24263** .11400 2.13 .0333 .01920 .46606 > > NsLOC2| .16908* .09025 1.87 .0610 -.00780 .34596 > > NsLOC3| .07511 .10010 .75 .4531 -.12109 .27130 > > NsLOC4| .00998 .10294 .10 .9227 -.19177 .21174 > > NsLOC5| .07192 .12174 .59 .5547 -.16669 .31052 > > NsLOC6| .20690** .10337 2.00 .0453 .00430 .40949 > > NsLOC7| .07074 .09646 .73 .4633 -.11831 .25979 > > NsMODE1| .08671* .04435 1.96 .0505 -.00021 .17363 > > NsMODE2| .14239*** .04517 3.15 .0016 .05386 .23092 > > NsPERS1| .03078 .04556 .68 .4993 -.05852 .12007 > > NsPERS2| .05790 .04805 1.20 .2282 -.03628 .15209 > > NsPERS3| .04128 .06577 .63 .5302 -.08762 .17018 > > NsHIVMED| .03298 .03951 .83 .4040 -.04447 .11042 > > > > > --------+-------------------------------------------------------------------- > > ***, **, * ==> Significance at 1%, 5%, 10% level. > > Model was estimated on Mar 27, 2020 at 00:55:10 PM > > > > > ----------------------------------------------------------------------------- > > > > thank you for your help > > > > Best, > > > > *Jason Ong* > > Twitter: @DrJasonJOng > > PhD, MMed, MBBS, FAChSHM, FRACGP > > > > Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health > > Associate Professor (Hon), London School of Hygiene and Tropical > Medicine, > > UK > > Central Clinical School, Monash University, Australia > > Melbourne School of Population and Global Health, University of > Melbourne, > > Australia > > Associate Editor, Sexually Transmitted Infections > > Special Issues Editor, Sexual Health > > Board Director, ASHM (https://protect-au.mimecast.com/s/novECQnMBZf3q1O5HxvjgM?domain=ashm.org.au) > > https://protect-au.mimecast.com/s/D_P9CROND2uRlDEAfNc7N9?domain=lshtm.ac.uk > > https://protect-au.mimecast.com/s/JoYICVARKgC5Kn3Dfy0wxD?domain=researchgate.net > > > > > > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://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/YU5CCWLVXkUPYLOAfn1j4S?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 wgreene at stern.nyu.edu Thu Apr 9 23:44:11 2020 From: wgreene at stern.nyu.edu (William Greene) Date: Thu, 9 Apr 2020 09:44:11 -0400 Subject: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par In-Reply-To: <1586423541535.75002@luke.fi> References: <1586423541535.75002@luke.fi> Message-ID: Ioanna. It looks like perhaps you have used the name NF for some other purpose in your commands. Is that correct? /Bill Greene On Thu, Apr 9, 2020 at 5:12 AM Grammatikopoulou Ioanna (Luke) < ext.ioanna.grammatikopoulou at luke.fi> wrote: > Dear William > I tried your suggestion but the model doesnt run (I get error 1085). > Thank you so much for your suggestions. > > Ioanna > > Ioanna Grammatikopoulou > Post-doc fellow > Natural Resources Institute Finland (LUKE) > Latokartanonkaari 9 > FI-00790, Helsinki > FINLAND > E-mail: ioanna.grammatikopoulou at luke.fi > > ________________________________________ > From: Limdep on behalf of William > Greene via Limdep > Sent: Wednesday, April 8, 2020 8:43 PM > To: Limdep and Nlogit Mailing List > Cc: William Greene > Subject: Re: [Limdep Nlogit List] FW: starting values in mixed logit with > correlated par > > You can provide starting values for the named parameters in the choice > model by adding them > to the first occurrences in the definitions of the model. For example, > ;Model: U(NF)=b0(-3.5)*BAU (Delete the ;labels setting) > However, I'm not sure what this implies for the associated variance > parameters. > (1) 15 random parameters in a model like this will be very optimistic. > This is a difficult log likelihood > to maximize in any event. This specification may be excessive. > (2) You might want to try fitting the model with your desired starting > values and fixed (nonrandom) > parameters, just to find a setup that works initially. > Regards, > William Greene > > On Wed, Apr 8, 2020 at 5:13 AM Grammatikopoulou Ioanna (Luke) < > ext.ioanna.grammatikopoulou at luke.fi> wrote: > > > Hi! > > I am trying to impose starting values on a mixed logit with correlated > > parameters models. As starting values I would like to use the values I > get > > from the mixed logit model. What I get though when I try to run the model > > is the following errors: > > > > Error 71: Variable list contains a name not in the expected table. > > Error 539: Variable list: The unidentifiable string is NF > > Error 1085: Unidentified name found in NF) > > > > Why is that? > > > > Here is the syntax I am using: > > > > NLOGIT;Lhs=CHOICE > > ;Choices=NF,SQ,MF > > ;pds=8 > > ;Labels=b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15 > > > > > ;Start=-3.5,0.07,0.09,0.04,0.09,-0.009,0.08,0.02,0.19,-0.06,0.54,-0.04,0.15,-0.07,0.46,-0.001 > > ;rpl > > ;corr > > ;pts=10 > > > > > ;Fcn=b0(N),b1(N),b2(N),b3(N),b4(N),b5(N),b6(N),b7(N),b8(N),b9(N),b10(N),b11(N),b12(N),b13(N),b14(N),b15(N) > > ;parameters > > ;Halton > > ;Model: > > U(NF)=b0*BAU+ > > > b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > > > > U(SQ)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > > > > U(MF)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE$ > > > > > > Thank you in advance. > > > > Ioanna Grammatikopoulou > > > > > > Ioanna Grammatikopoulou > > Post-doc fellow > > Natural Resources Institute Finland (LUKE) > > Latokartanonkaari 9 > > FI-00790, Helsinki > > FINLAND > > E-mail: ioanna.grammatikopoulou at luke.fi > ioanna.grammatikopoulou at mtt.fi> > > > > > > Ioanna Grammatikopoulou > > Post-doc fellow > > Natural Resource Institute Finland (LUKE) > > Latokartanonkaari 9 > > FI-00790, Helsinki > > Finland > > E-mail: ioanna.grammatikopoulou at luke.fi > ioanna.grammatikopoulou at luke.fi> > > > > > > > > _______________________________________________ > > Limdep site list > > Limdep at mailman.sydney.edu.au > > http://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/O_-nCNLJyQUExYjZSmo3f4?domain=people.stern.nyu.edu > Email: wgreene at stern.nyu.edu > Ph. +1.646.596.3296 > 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 > 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, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/O_-nCNLJyQUExYjZSmo3f4?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 dgiantsios at uom.gr Fri Apr 10 04:50:12 2020 From: dgiantsios at uom.gr (dgiantsios at uom.gr) Date: Thu, 09 Apr 2020 21:50:12 +0300 Subject: [Limdep Nlogit List] QUERIES Message-ID: <0553d10f6632453c3e90ee8448ebb4fc@uom.gr> Dear All I HAVE RUN A STOCHASTIC FRONTIER ANALYSIS MODEL WITH A SAMPLE OF 7,500 FIRMS. DATA EDITOR SHOWS ME THE EFFICIENCY SCORE ONLY FOR THE FIRST 5,000 FIRMS. HOW CAN I HAVE THE SCORE FOR THE OTHER 2,500 FIRMS? THANKS IN ADVANCE From ext.ioanna.grammatikopoulou at luke.fi Fri Apr 10 16:48:41 2020 From: ext.ioanna.grammatikopoulou at luke.fi (Grammatikopoulou Ioanna (Luke)) Date: Fri, 10 Apr 2020 06:48:41 +0000 Subject: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par In-Reply-To: References: <1586423541535.75002@luke.fi>, Message-ID: <1586501322373.53360@luke.fi> Dear Bill I havent used NF for some other purpose. The syntax works fine if I dont include ;corr. I also tried to restrict the correlations but again it collapses. I believe it is the number of attributes. Because if I use half of the parameters it works fine. Still though i cant resolve the problem with the starting values. Thanks for your consultation. Ioanna Ioanna Grammatikopoulou Post-doc fellow Natural Resources Institute Finland (LUKE) Latokartanonkaari 9 FI-00790, Helsinki FINLAND E-mail: ioanna.grammatikopoulou at luke.fi ________________________________________ From: Limdep on behalf of William Greene via Limdep Sent: Thursday, April 9, 2020 4:44 PM To: Limdep and Nlogit Mailing List Cc: William Greene Subject: Re: [Limdep Nlogit List] FW: starting values in mixed logit with correlated par Ioanna. It looks like perhaps you have used the name NF for some other purpose in your commands. Is that correct? /Bill Greene On Thu, Apr 9, 2020 at 5:12 AM Grammatikopoulou Ioanna (Luke) < ext.ioanna.grammatikopoulou at luke.fi> wrote: > Dear William > I tried your suggestion but the model doesnt run (I get error 1085). > Thank you so much for your suggestions. > > Ioanna > > Ioanna Grammatikopoulou > Post-doc fellow > Natural Resources Institute Finland (LUKE) > Latokartanonkaari 9 > FI-00790, Helsinki > FINLAND > E-mail: ioanna.grammatikopoulou at luke.fi > > ________________________________________ > From: Limdep on behalf of William > Greene via Limdep > Sent: Wednesday, April 8, 2020 8:43 PM > To: Limdep and Nlogit Mailing List > Cc: William Greene > Subject: Re: [Limdep Nlogit List] FW: starting values in mixed logit with > correlated par > > You can provide starting values for the named parameters in the choice > model by adding them > to the first occurrences in the definitions of the model. For example, > ;Model: U(NF)=b0(-3.5)*BAU (Delete the ;labels setting) > However, I'm not sure what this implies for the associated variance > parameters. > (1) 15 random parameters in a model like this will be very optimistic. > This is a difficult log likelihood > to maximize in any event. This specification may be excessive. > (2) You might want to try fitting the model with your desired starting > values and fixed (nonrandom) > parameters, just to find a setup that works initially. > Regards, > William Greene > > On Wed, Apr 8, 2020 at 5:13 AM Grammatikopoulou Ioanna (Luke) < > ext.ioanna.grammatikopoulou at luke.fi> wrote: > > > Hi! > > I am trying to impose starting values on a mixed logit with correlated > > parameters models. As starting values I would like to use the values I > get > > from the mixed logit model. What I get though when I try to run the model > > is the following errors: > > > > Error 71: Variable list contains a name not in the expected table. > > Error 539: Variable list: The unidentifiable string is NF > > Error 1085: Unidentified name found in NF) > > > > Why is that? > > > > Here is the syntax I am using: > > > > NLOGIT;Lhs=CHOICE > > ;Choices=NF,SQ,MF > > ;pds=8 > > ;Labels=b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15 > > > > > ;Start=-3.5,0.07,0.09,0.04,0.09,-0.009,0.08,0.02,0.19,-0.06,0.54,-0.04,0.15,-0.07,0.46,-0.001 > > ;rpl > > ;corr > > ;pts=10 > > > > > ;Fcn=b0(N),b1(N),b2(N),b3(N),b4(N),b5(N),b6(N),b7(N),b8(N),b9(N),b10(N),b11(N),b12(N),b13(N),b14(N),b15(N) > > ;parameters > > ;Halton > > ;Model: > > U(NF)=b0*BAU+ > > > b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > > > > U(SQ)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE/ > > > > > U(MF)=b1*Plant2+b2*Plant3+b3*BUTERF2+b4*BUTERF3+b5*BIRD2+b6*BIRD3+b7*EROSION2+b8*EROSION3+b9*WATER2+b10*WATER3+b11*CLIMATE2+b12*CLIMATE3+b13*FOOD2+b14*FOOD3+b15*PRICE$ > > > > > > Thank you in advance. > > > > Ioanna Grammatikopoulou > > > > > > Ioanna Grammatikopoulou > > Post-doc fellow > > Natural Resources Institute Finland (LUKE) > > Latokartanonkaari 9 > > FI-00790, Helsinki > > FINLAND > > E-mail: ioanna.grammatikopoulou at luke.fi > ioanna.grammatikopoulou at mtt.fi> > > > > > > Ioanna Grammatikopoulou > > Post-doc fellow > > Natural Resource Institute Finland (LUKE) > > Latokartanonkaari 9 > > FI-00790, Helsinki > > Finland > > E-mail: ioanna.grammatikopoulou at luke.fi > ioanna.grammatikopoulou at luke.fi> > > > > > > > > _______________________________________________ > > Limdep site list > > Limdep at mailman.sydney.edu.au > > http://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/4euwCVARKgC5J5n6SG4WHK?domain=people.stern.nyu.edu > Email: wgreene at stern.nyu.edu > Ph. +1.646.596.3296 > 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 > 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, emeritus Stern School of Business, New York University 44 West 4 St. New York, NY, 10012 URL: https://protect-au.mimecast.com/s/4euwCVARKgC5J5n6SG4WHK?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 http://limdep.itls.usyd.edu.au From doctorjasonong at gmail.com Mon Apr 13 14:12:32 2020 From: doctorjasonong at gmail.com (Jason Ong) Date: Mon, 13 Apr 2020 14:12:32 +1000 Subject: [Limdep Nlogit List] Statistical significance value for the omitted/reference level in an effects coded model Message-ID: hi NLOGIT team, I am running an RPL model which I have effects coded. The model output doesn't display the reference level's coefficient so because this is effects coded, to get the coefficient of the reference level, I use the negative sum of the other level's coefficients. currently to calculate the standard error of the coefficient of the reference level, I have to use the covariance matrix for each level of the attribute to calculate the standard error for the coefficient of the reference level. Is there an easier way in the current NLOGIT output for me to do this or do I just need to keep manually calculating this? thanks for your help. *Jason Ong* Twitter: @DrJasonJOng PhD, MMed, MBBS, FAChSHM, FRACGP Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health Associate Professor (Hon), London School of Hygiene and Tropical Medicine, UK Central Clinical School, Monash University, Australia Melbourne School of Population and Global Health, University of Melbourne, Australia Associate Editor, Sexually Transmitted Infections Special Issues Editor, Sexual Health Board Director, ASHM (https://protect-au.mimecast.com/s/n2TVClx1NjiAYPO5FGO3kK?domain=ashm.org.au) https://protect-au.mimecast.com/s/tNLbCmO5gluAQP1KFOyrkq?domain=lshtm.ac.uk https://protect-au.mimecast.com/s/fJC_Cnx1jniKR3lquNIaOG?domain=researchgate.net From wgreene at stern.nyu.edu Tue Apr 14 03:13:38 2020 From: wgreene at stern.nyu.edu (William Greene) Date: Mon, 13 Apr 2020 13:13:38 -0400 Subject: [Limdep Nlogit List] Statistical significance value for the omitted/reference level in an effects coded model In-Reply-To: References: Message-ID: Jason. The coefficient on the "reference" level is zero, not the negative of the sum of the other coefficients. To make this calculation, you are assuming that the sum of the coefficients equals zero. I don't think that is appropriate. On the other hand, if you do want to compute a standard error for that function of several coefficients, you can use the WALD command after your NLOGIT command. It's simple. Bill Greene On Mon, Apr 13, 2020 at 12:13 AM Jason Ong via Limdep < limdep at mailman.sydney.edu.au> wrote: > hi NLOGIT team, > > I am running an RPL model which I have effects coded. > The model output doesn't display the reference level's coefficient > so because this is effects coded, to get the coefficient of the reference > level, I use the negative sum of the other level's coefficients. > currently to calculate the standard error of the coefficient of the > reference level, I have to use the covariance matrix for each level of the > attribute to calculate the standard error for the coefficient of the > reference level. > Is there an easier way in the current NLOGIT output for me to do this or do > I just need to keep manually calculating this? > thanks for your help. > > *Jason Ong* > Twitter: @DrJasonJOng > PhD, MMed, MBBS, FAChSHM, FRACGP > > Sexual Health Physician, Melbourne Sexual Health Centre, Alfred Health > Associate Professor (Hon), London School of Hygiene and Tropical Medicine, > UK > Central Clinical School, Monash University, Australia > Melbourne School of Population and Global Health, University of Melbourne, > Australia > Associate Editor, Sexually Transmitted Infections > Special Issues Editor, Sexual Health > Board Director, ASHM (https://protect-au.mimecast.com/s/O3fdCWLVXkUxgmO3t6kPBo?domain=ashm.org.au) > https://protect-au.mimecast.com/s/i64BCXLW2mUGyo2AfVdCT4?domain=lshtm.ac.uk > https://protect-au.mimecast.com/s/2JR5CYW8NocgJ2MYi9T9Y9?domain=researchgate.net > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://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/iHuCCZY1NqixV9lvtyVfgI?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 bills at gra-inc.com Tue Apr 21 03:59:05 2020 From: bills at gra-inc.com (William Spitz) Date: Mon, 20 Apr 2020 13:59:05 -0400 Subject: [Limdep Nlogit List] F-test question Message-ID: <4b08abd7-ab02-c648-605f-c0685425b6f1@gra-inc.com> I am doing an F-test for equality of coefficients in a simple linear model. Base model is y = a + b X I am testing whether the observations should be split into Groups 1 and 2: y = a1 + b1 X + a2 + b2 X But when I run the restricted model (using CLS: b(1)=b(3), b(2)=b(4)), I don't get the same restricted coefficient values as I do when running the Base model. They're pretty close, but not the same -- and it seems to be off by more than just rounding error. Is this a "feature" of imposing restrictions via CLS? Thanks in advance. /Bill Spitz From wgreene at stern.nyu.edu Tue Apr 21 04:52:16 2020 From: wgreene at stern.nyu.edu (William Greene) Date: Mon, 20 Apr 2020 14:52:16 -0400 Subject: [Limdep Nlogit List] F-test question In-Reply-To: <4b08abd7-ab02-c648-605f-c0685425b6f1@gra-inc.com> References: <4b08abd7-ab02-c648-605f-c0685425b6f1@gra-inc.com> Message-ID: Bill. It is not a feature. Try something like this: create;fed=female*educ ; med=educ-fed ; male=1-female$$ regr;lhs=income;rhs=female,fed,male,med ;cls:b(1)-b(3)=0,b(2)-b(4)=0$$ regr;lhs=income;rhs=one,educ$ Please see below for results. /Bill Greene ----------------------------------------------------------------------------- Restricted least squares regression ............ LHS=INCOME Mean = .44476 Standard deviation = .21659 ---------- No. of observations = 3377 DegFreedom Mean square Regression Sum of Squares = 11.3623 1 11.36229 Residual Sum of Squares = 147.004 3375 .04356 Total Sum of Squares = 158.367 3376 .04691 ---------- Standard error of e = .20870 Root MSE .20864 Fit R-squared = .07175 R-bar squared .07147 Model test F[ 1, 3375] = 260.86136 Prob F > F* .00000 Restrictions F[ 2, 3373] = 1.74597 Prob F > F* .17463 --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence INCOME| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- FEMALE| .16692*** .01757 9.50 .0000 .13248 .20136 FED| .02415*** .00150 16.15 .0000 .02122 .02708 MALE| .16692*** .01757 9.50 .0000 .13248 .20136 MED| .02415*** .00150 16.15 .0000 .02122 .02708 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Apr 20, 2020 at 02:47:40 PM ----------------------------------------------------------------------------- |-> regr;lhs=income;rhs=one,educ$ ----------------------------------------------------------------------------- Ordinary least squares regression ............ LHS=INCOME Mean = .44476 Standard deviation = .21659 ---------- No. of observations = 3377 DegFreedom Mean square Regression Sum of Squares = 11.3623 1 11.36229 Residual Sum of Squares = 147.004 3375 .04356 Total Sum of Squares = 158.367 3376 .04691 ---------- Standard error of e = .20870 Root MSE .20864 Fit R-squared = .07175 R-bar squared .07147 Model test F[ 1, 3375] = 260.86136 Prob F > F* .00000 --------+-------------------------------------------------------------------- | Standard Prob. 95% Confidence INCOME| Coefficient Error z |z|>Z* Interval --------+-------------------------------------------------------------------- Constant| .16692*** .01757 9.50 .0000 .13248 .20136 EDUC| .02415*** .00150 16.15 .0000 .02122 .02708 --------+-------------------------------------------------------------------- ***, **, * ==> Significance at 1%, 5%, 10% level. Model was estimated on Apr 20, 2020 at 02:49:10 PM ----------------------------------------------------------------------------- On Mon, Apr 20, 2020 at 2:36 PM William Spitz wrote: > I am doing an F-test for equality of coefficients in a simple linear model. > > Base model is y = a + b X > I am testing whether the observations should be split into Groups 1 and 2: > y = a1 + b1 X + a2 + b2 X > > But when I run the restricted model (using CLS: b(1)=b(3), b(2)=b(4)), I > don't get the same restricted coefficient values as I do when running > the Base model. > They're pretty close, but not the same -- and it seems to be off by more > than just rounding error. > > Is this a "feature" of imposing restrictions via CLS? > > Thanks in advance. > /Bill Spitz > > _______________________________________________ > Limdep site list > Limdep at mailman.sydney.edu.au > http://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/_m2lCmO5glu6YjJvfGdl1N?domain=people.stern.nyu.edu Email: wgreene at stern.nyu.edu Ph. +1.646.596.3296 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 bills at gra-inc.com Tue Apr 21 05:54:27 2020 From: bills at gra-inc.com (William Spitz) Date: Mon, 20 Apr 2020 15:54:27 -0400 Subject: [Limdep Nlogit List] F-test question In-Reply-To: <4b08abd7-ab02-c648-605f-c0685425b6f1@gra-inc.com> References: <4b08abd7-ab02-c648-605f-c0685425b6f1@gra-inc.com> Message-ID: <45cdad4d-13de-3b3f-ddc9-5a74994656d6@gra-inc.com> Sorry, never mind. I figured this out. On 4/20/2020 1:59 PM, William Spitz wrote: > I am doing an F-test for equality of coefficients in a simple linear > model. > > Base model is y = a + b X > I am testing whether the observations should be split into Groups 1 and 2: > y = a1 + b1 X + a2 + b2 X > > But when I run the restricted model (using CLS: b(1)=b(3), b(2)=b(4)), > I don't get the same restricted coefficient values as I do when > running the Base model. > They're pretty close, but not the same -- and it seems to be off by > more than just rounding error. > > Is this a "feature" of imposing restrictions via CLS? > > Thanks in advance. > /Bill Spitz >