Multinomial Logit for travel destinations

Chhatoq — Wed, 14 Jun 2017 00:29:22 GMT

Dear Friends,

I am trying to estimate how the characteristics (attributes) of a destination influence the decision of an individual to choose that destination for travel.

The choice set for each individual won't be the same (non-universal choice set) and each choice set is composed of 10 different alternatives from which the individual chooses only one alternative. The total number of travel alternatives in the database is 200.

The independent variables represent a range of attributes that are common across all alternatives ( e.g., average price, number of attractions, rankings, among others). Please see below for a sample of the data arrangement.

Would the "logistic" or "bchoice" procedures be good for this?

Any help would be very appreciated.

Thank you.

Datoq

Choice	Individual	Attr1	Attr2	Attr3	Nobs
1	1	5	82	0.6	10
0	1	2	33	0.27	10
0	1	3	14	0.6	10
0	1	7	54	1.3	10
0	1	6	63	0.5	10
0	1	2	85	0.3	10
0	1	12	74	2	10
0	1	5	71	3	10
0	1	21	22	4.2	10
0	1	24	36	1	10
0	2	5	93	3.9	10
0	2	6	4	2	10
0	2	8	43	5.7	10
…	…	…	…	…	…
1	2	6	7	5.9	10
0	2	2	34	2.9	10

Re: Multinomial Logit for travel destinations

WarrenKuhfeld — Wed, 14 Jun 2017 12:57:48 GMT

BChoice, PHReg, and MDC fit multinomial logit models.

PHREG is not documented as a discrete choice procedure, but you can look here http://support.sas.com/techsup/technote/mr2010c.pdf and elsewhere for more information about it.

topic Multinomial Logit for travel destinations in Statistical Procedures

Multinomial Logit for travel destinations

Re: Multinomial Logit for travel destinations