Important: Read this before posting to this forum

  1. 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.
  2. 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
  3. Before asking a question on the forum, users are kindly requested to follow these steps:
    1. Check that the same issue has not already been addressed in the forum - there is a search tool.
    2. 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
    3. Check the frequently asked questions section on the Apollo website, which discusses some common issues/failures. Please see http://www.apollochoicemodelling.com/faq.html
    4. Make sure that R is using the latest official release of Apollo.
  4. If the above steps do not resolve the issue, then users should follow these steps when posting a question:
    1. provide full details on the issue, including the entire code and output, including any error messages
    2. 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.

Person attributes in unlabelled choices

Ask general questions about model specification and estimation that are not Apollo specific but relevant to Apollo users.
Post Reply
Anna
Posts: 5
Joined: 01 Feb 2021, 20:47

Person attributes in unlabelled choices

Post by Anna »

Hi,

I'm trying to estimate a simple MNL using data from an unlabelled stated choice survey. I'm fine with the alternative specific attributes but would also like to include socio-demographic data. Is that possible?
I have only ever estimated models using labelled data so I'm a little off at the moment. Could you point me in the right direction?

I'm sorry if this isn't the right place to ask this question.

Best wishes
Anna
stephanehess
Site Admin
Posts: 974
Joined: 24 Apr 2020, 16:29

Re: Person attributes in unlabelled choices

Post by stephanehess »

Hi Anna

if you have an attribute that varies across alternatives, then you can interact that attribute with socio-demographics, whether the data is labelled or not. So for example, if you have price_1 and price_2, you could have

V[[1]] = ... b_price * price_1 * (income/av_income)^lambda_inc
V[[2]] = ... b_price * price_2 * (income/av_income)^lambda_inc

where lambda_inc is then an income elasticity

Stephane
--------------------------------
Stephane Hess
www.stephanehess.me.uk
Anna
Posts: 5
Joined: 01 Feb 2021, 20:47

Re: Person attributes in unlabelled choices

Post by Anna »

Hi Stephane,

thanks for the quick response! That makes sense. I've already tried it with my data and it works like a charm.

Thanks!
Anna
Sutsch
Posts: 8
Joined: 13 Sep 2021, 12:51

Person attributes in unlabelled choices

Post by Sutsch »

Hi together,

I don't know if this is the right place to adress my issue, however, it seems close to the question of Anna (kind of).

I have data from an unlabelled choice experiment with four attributes (e.g. service and interaction channels). Before taking the choice tasks I asked the respondents to evalue the different attributes (and levels) based on a three-stage maturity model (MM). This means I have maturity perceptions of each attribute level for every respondent (= Person attributes).

I want to estimate two models:
1) Base attributes that interact with the maturity model estimations, e.g. V = ... (b_service1 * b_service1_x_MM * MMservice1==1) * service1==1... (service1 beeing one attribute level of service)
2) We had the idea to substitute the attribute codings in the utility function with the individual MM scores, in order to obtain the preference for the MM level, e.g. V = ...b_MM1_service1 * MM1_service1==1 + b_MM2_service1 * MM1_service1==2 + ...

I know that the 2nd model impacts my choice design, but I was wondering if you have done something like this before, or/and if you have thoughts on that procedure.

Thank you very much in advance for your thoughts!

Best
Sutsch
stephanehess
Site Admin
Posts: 974
Joined: 24 Apr 2020, 16:29

Re: Person attributes in unlabelled choices

Post by stephanehess »

Hi

could you be more specific and also define what your attributes are in your utility function, not just the interaction terms?

Another thing to note is that you should probably not treat these maturity perceptions as error free explanatory variables

Stephane
--------------------------------
Stephane Hess
www.stephanehess.me.uk
Sutsch
Posts: 8
Joined: 13 Sep 2021, 12:51

Re: Person attributes in unlabelled choices

Post by Sutsch »

Hi Stephane,

again: Thank you very much for your quick reply! That helps a lot.

My base attributes and the corresponding attribute level are:
1. Price calculation (P): fix, temporal variable, consumption variable (coding 1 to 3)
2. Communication: (A): Mail, e-mail, online portal, messaging service via app (coding 1 to 4)
3. Service: (S): E-Mail, chat (bot), app (coding 1 to 3)
4. Additional Device (D): no device, mechanical, local connected, app controlled (coding 1 to 4)
5. Price attribute (C): monthly fee additional to (P)

Example: One respondent perceives the service channel "chat (bot)" as having the highest maturiry (MMS2==3) with the range of available service channels. Therefore in my third model, I obtain the preference/WTP for "chat (bot)" that are perceivd of having a high maturity level.

Base Model: V = ... beta_S1 * S==1 + beta_S2 * S==2 + beta_S3 * S==3 + ...
Model 1: V = ... (beta_S2 * beta_S2_MM * S2MM) * S2==2 + ...
Model 2: V = ... beta_S2MM * S2MM==1 + beta_S2MM * S2MM==2 + beta_S2MM * S2MM==3 + ...

Does this answer your question? Or did I take a wrong turn?

With regards to your concern of having error free explanatory variables, is that not covered by ϵ_i ? Or should I introduce additional error terms?

Thank you!!

Have a nice Sunday!
Sutsch
stephanehess
Site Admin
Posts: 974
Joined: 24 Apr 2020, 16:29

Re: Person attributes in unlabelled choices

Post by stephanehess »

Hi

sorry, but this is still not making sense to us, really.

I assume with

(beta_S2 * beta_S2_MM * S2MM) * S2==2

you are trying to capture an interaction between the sensivity to S2 (i.e. beta_S2) and whether S2MM applies? But this is not how the utility would be coded in that way. It would be more something like:

(beta_S2 + beta_S2_MM * S2MM) * S2==2

In relation to the perception variables being treated as error free, the extreme value term is not helping you with this. You should have a look at the literature on hybrid choice models instead

Best wishes

Stephane
--------------------------------
Stephane Hess
www.stephanehess.me.uk
Sutsch
Posts: 8
Joined: 13 Sep 2021, 12:51

Re: Person attributes in unlabelled choices

Post by Sutsch »

Hi,

you are right, I got mixed up in the post. Currently i have "(beta_S2 + beta_S2_MM * S2MM) * S2==2" in my code. I am very sorry for the confusion! And yes, i would like to quantify any interaction effects between S2 and the corresponding individual maturity perception for that attribute level.

I will look at hybrid choice models, thank you very much for that direction!

Have a very nice Day!

Thank you!

Best Wishes
Sutsch
Post Reply