I am trying to evaluate consumer preferences in relation to new transport alternatives by using choice modelling and hoping to use above models in the analysis. I am wondering how to form questions in the survey to use these models in analysis?
Hi,
The difference between the multinomial and ordered logit models concerns the nature of the scale that underlies the item. Let's say that you want to use an item as a dependent variable in one of the logit models you mention. If the item is unordered (e.g., "what factor most determines your choice of transportation: a) cost b) convenience c) cleaniness d) safety") then a multinomial logit model is appropriate. The item is multinomial in the sense that it made up of multiple categories. An ordered item (e.g., "how satisfied are you with your transport choices: a) dissatisfied b) indifferent c) satisfied") could be analyzed using an ordered logit model because the results allow you to interpret the effects of predictors on the underlying ordered (less to more) scale of satisfaction.
For the unordered item, contrasts of a vs b, b vs. c, c vs. d, etc. can be arbitrarily constructed--any particular set of contrasts the analyst prefers could be used. For the ordered logit model the contrasts a vs. b and b vs. c make the most sense because b reflects "more satisfaction" than a, and c reflects "more satisfaction" than b.
One more thought I had was that it might be easier for you to write your items first and then decide how best to analyze them later. There are item-writing tutorials available in texts an online that might be helpful:
http://www.socialpsychology.org/teach/itemtips.htm
Hope this helps,
Scot
Thank you very much for explaining the models. The confusion is very much cleared now.
Hi Divyakant,
Thank you very much for explaining the models and providing the link in order to further explore discrete choice choice modelling. I will go through it.
Thank you.
Hi Scot and Divyakant,
I have attached a word file including a choice experiment can you please take a look at that? My question is actually related to this one specific choice experiment. How can I model the choices in relation to this one choice experiment?
If you have time please respond. Thanks
Hi Nitin,
If you do not mind investigating an advanced analysis of choice experiment. I suggested to attempt yourself rank-ordered logit model which you can refer from this source: https://www.stata.com/manuals13/rrologit.pdf.
Good luck!
You can read my thesis : Book Réformes des marchés agricoles: coûts de transaction, choix ...
I applied Nested Logit Model and justifiy why it must be used.
I would go with the idea that was suggested by @Divyakant Tahlyan
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix