thank you for the valuable advice you provide on this forum to all STATA newbies like me.
As part of a research that I am carrying out with my group, we have developed a questionnaire based on the best to worst scaling approach. The survey plans to ask an interviewee through an online survey to choose between a set of alternatives the best and the worst (a choice for the best, a choice for the worst).
BEST WORSTOPT1a [_] [_]
OPT1b [X] [_]
OPT1c [_] [_]
OPT1d [_] [X]
OPT1e [_] [_]
In this example, OPT1b is the best, OPT1d is the worst.
This exercise is developed 5 times by the same interviewee. To develop the survey, we used LimeSurvey.
The software reports the results for each person interviewed horizontally. So the results for each interviewee take the form:
ID | OPTB1 | OPTW1 | OPTB2 | OPTW2 | OPTB3 | OPTW3 | OPTB4 | OPTW4 | OPTB5 | OPTW5 | AGE | INCOME
where ID is the indicator of the interviewee, OPTB1 is the "best" option chosen by the interview in the first group and OPTW1 is the "worst" option chosen by the interview in the first group, OPTB2 and OPTW2 in the second, and so via up to 5, AGE and INCOME the socio-economic variables for each interviewee.
In order to develop the analyzes we need the data to have a vertical format where there are 5 lines for each interviewee:
ID | OPTB1 | OPTW1 | AGE | INCOME
ID | OPTB2 | OPTW2 | AGE | INCOME
ID | OPTB3 | OPTW3 | AGE | INCOME
ID | OPTB4 | OPTW4 | AGE | INCOME
ID | OPTB5 | OPTW5 | AGE | INCOME
The groups of questions are preset. In fact there were 20 macro groups (randoms) that include the 5 different groups that are automatically selected by the LimeSurvey software for each interviewee in order to have a balanced dataset. The random number selected is available for each interviewee. I don't know if this last information can be useful.
I thank you very much for your support.
I hope you can help me.
Federico
0 Response to Reshaping to long
Post a Comment