Hi,
I am analysing data from a choice experiment with two alternatives and just read in the article from Trine Kjaer - A review of the discrete choice experiment - with emphasis on its application in health care (see page 137 in attached document or PDF page 141, respectively) that analysing data from discrete choice experimets with binary tasks require to set up data as differences.
I thought to set up my data in the usual way (as described on the next page in the paper), each row showing one alternative including a column for the respondent ID, the choice set ID, the alternative, the attribute levels as well as the choice (0 or 1).
My question: Do I need to set up the data as differences and if yes which alternative do I include? Referring to the paper, it is not so obvious for me. I would include the chosen alternative for each choice set but in the paper, they seem to take each possible choice - car, bus and opt out - once for all three choice sets per respondent.
I would use the following Stata commands depending on the assumptions made:
mixlogit answer duration walk price, group(question) id(id)
clogit answer duration walk price, group(question_id)
Thank you and best regards
Felix
I am analysing data from a choice experiment with two alternatives and just read in the article from Trine Kjaer - A review of the discrete choice experiment - with emphasis on its application in health care (see page 137 in attached document or PDF page 141, respectively) that analysing data from discrete choice experimets with binary tasks require to set up data as differences.
I thought to set up my data in the usual way (as described on the next page in the paper), each row showing one alternative including a column for the respondent ID, the choice set ID, the alternative, the attribute levels as well as the choice (0 or 1).
My question: Do I need to set up the data as differences and if yes which alternative do I include? Referring to the paper, it is not so obvious for me. I would include the chosen alternative for each choice set but in the paper, they seem to take each possible choice - car, bus and opt out - once for all three choice sets per respondent.
I would use the following Stata commands depending on the assumptions made:
mixlogit answer duration walk price, group(question) id(id)
clogit answer duration walk price, group(question_id)
Thank you and best regards
Felix
