Hi
I estimated this model in wtp-space:
V = list()
V[['alt1']] = Certain *(cost_B*(asc_B + torsk_B * KT1 + Cost1 + laks_B * VL1 + bunn_B * HB1 + land_B * KL1))+
(1-Certain)*(cost_T*(asc_T + torsk_T * KT1 + Cost1 + laks_T * VL1 + bunn_T * HB1 + land_T * KL1))
V[['alt2']] = Certain * (cost_B*(torsk_B * KT2 + Cost2 + laks_B * VL2 + bunn_B * HB2 + land_B * KL2))+
(1-Certain)* (cost_T*(torsk_T * KT2 + Cost2 + laks_T * VL2 + bunn_T * HB2 + land_T * KL2))
V[['alt3']] = Certain * (cost_B*(torsk_B * KT3 + Cost3 + laks_B * VL3 + bunn_B * HB3 + land_B * KL3))+
(1-Certain)* (cost_T*(torsk_T * KT3 + Cost3 + laks_T * VL3 + bunn_T * HB3 + land_T * KL3))
"Certain" is a dummy taking the value 1 for respondents in one information set in a split sample choice experiment, and the value 0 for respondents in the other information set. KT, VL, HB and KL are the attributes, and torsk, laks, bunn and land the parameters to be estimated. _B are parameters for baseline (information set 1) and _T are parameters for treatment (information set 2).
Estimating the model works fine, and the model converges. However, I get very high values for the ASC-parameters. This result is interpretable from an empiricle point of view (strong aversion against alternative 1, which is the SQ). I just wanted to ask whether there may be other, model technical reasons for this result, which I should take into consideration??
Model name : MXL_aqua_exp
Model description : MXL model with dummy coding
Model run at : 2022-06-30 11:24:53
Estimation method : bfgs
Model diagnosis : successful convergence
Number of individuals : 293
Number of rows in database : 2599
Number of modelled outcomes : 2599
Number of cores used : 3
Number of inter-individual draws : 1000 (SobolOwenFaureTezuka)
LL(start) : -2855.8
LL(0) : -2855.29
LL(C) : -2667.63
LL(final) : -1611.64
Rho-square (0) : 0.4356
Adj.Rho-square (0) : 0.4272
Rho-square (C) : 0.3959
Adj.Rho-square (C) : 0.3869
AIC : 3271.27
BIC : 3411.98
Estimated parameters : 24
Time taken (hh:mm:ss) : 01:22:6.62
pre-estimation : 00:05:55.97
estimation : 00:28:0.47
post-estimation : 00:48:10.18
Iterations : 118
Min abs eigenvalue of Hessian : 0.046956
Unconstrained optimisation.
Estimates:
Estimate s.e. t.rat.(0) Rob.s.e. Rob.t.rat.(0)
asc_B_mu 17.44502 3.09223 5.6416 8.277305 2.1076
asc_B_sig 19.82318 3.47555 5.7036 9.472019 2.0928
asc_T_mu 6.86215 1.26935 5.4060 1.851939 3.7054
asc_T_sig -14.42643 1.78324 -8.0900 2.934400 -4.9163
cost_B_mu -0.36412 0.16356 -2.2262 0.281769 -1.2923
cost_B_sig 1.10744 0.18507 5.9837 0.219688 5.0410
torsk_B_mu 0.38443 0.07701 4.9921 0.126627 3.0359
torsk_B_sig -0.55603 0.07488 -7.4259 0.148787 -3.7371
laks_B_mu 0.23840 0.11780 2.0238 0.244438 0.9753
laks_B_sig 0.74951 0.13155 5.6975 0.356692 2.1013
bunn_B_mu 0.72419 0.27200 2.6625 0.256366 2.8248
bunn_B_sig -3.6574e-04 8.5917e-04 -0.4257 0.001136 -0.3221
land_B_mu -0.01622 0.01423 -1.1402 0.013146 -1.2339
land_B_sig -0.06131 0.04961 -1.2358 0.152943 -0.4009
cost_T_mu -0.15527 0.16434 -0.9448 0.230286 -0.6743
cost_T_sig -0.91270 0.15905 -5.7385 0.163097 -5.5960
torsk_T_mu 0.25055 0.06783 3.6939 0.089304 2.8056
torsk_T_sig 0.57420 0.08993 6.3850 0.150160 3.8239
laks_T_mu 0.28932 0.08334 3.4714 0.073636 3.9291
laks_T_sig 0.60285 0.09112 6.6160 0.105155 5.7329
bunn_T_mu 0.66006 0.26217 2.5177 0.273440 2.4139
bunn_T_sig -0.09652 0.24403 -0.3955 0.114511 -0.8429
land_T_mu 0.02216 0.01593 1.3910 0.017167 1.2907
land_T_sig -0.09742 0.02474 -3.9381 0.048570 -2.0059
Thank you in advance for your reply.
best regards,
Margrethe
Important: Read this before posting to this forum
- This forum is for questions related to the use of Apollo. We will answer some general choice modelling questions too, where appropriate, and time permitting. We cannot answer questions about how to estimate choice models with other software packages.
- There is a very detailed manual for Apollo available at http://www.ApolloChoiceModelling.com/manual.html. This contains detailed descriptions of the various Apollo functions, and numerous examples are available at http://www.ApolloChoiceModelling.com/examples.html. In addition, help files are available for all functions, using e.g. ?apollo_mnl
- Before asking a question on the forum, users are kindly requested to follow these steps:
- Check that the same issue has not already been addressed in the forum - there is a search tool.
- Ensure that the correct syntax has been used. For any function, detailed instructions are available directly in Apollo, e.g. by using ?apollo_mnl for apollo_mnl
- Check the frequently asked questions section on the Apollo website, which discusses some common issues/failures. Please see http://www.apollochoicemodelling.com/faq.html
- Make sure that R is using the latest official release of Apollo.
- Users can check which version they are running by entering packageVersion("apollo").
- Then check what is the latest full release (not development version) at http://www.ApolloChoiceModelling.com/code.html.
- To update to the latest official version, just enter install.packages("apollo"). To update to a development version, download the appropriate binary file from http://www.ApolloChoiceModelling.com/code.html, and install the package from file
- If the above steps do not resolve the issue, then users should follow these steps when posting a question:
- provide full details on the issue, including the entire code and output, including any error messages
- posts will not immediately appear on the forum, but will be checked by a moderator first. This may take a day or two at busy times. There is no need to submit the post multiple times.
high ASC-coefficients in model in wtp-space
Re: high ASC-coefficients in model in wtp-space
Hi Margrethe,
I do not see any glaring problem with the model. But formal identifiability conditions for mixed logits are not simple, specially in the case of random ASC. So I would recommend you look at Walker (2002) for more details. Note that requirements are different for continuous and dummy-coded explanatory variables.
Furthermore, there may be "empirical identification" issues, meaning that there is not enough information in your data to support the specification you are trying to estimate. You may want to try simpler models (i.e. without so many random coefficients) and see if results are stable. If results change too much when using a simpler specification, it may be a sign that you have empirical identification issues, and I would recommend using a simpler model.
Best wishes
David
I do not see any glaring problem with the model. But formal identifiability conditions for mixed logits are not simple, specially in the case of random ASC. So I would recommend you look at Walker (2002) for more details. Note that requirements are different for continuous and dummy-coded explanatory variables.
Furthermore, there may be "empirical identification" issues, meaning that there is not enough information in your data to support the specification you are trying to estimate. You may want to try simpler models (i.e. without so many random coefficients) and see if results are stable. If results change too much when using a simpler specification, it may be a sign that you have empirical identification issues, and I would recommend using a simpler model.
Best wishes
David