1
$\begingroup$

I ran a survey and one question was to indicate which accommodation (hotels, apartment, b&b, ...) respondents prefer when travelling. Thery were allowed to select more than one option. The survey was completed by two different groups: university students and workers aged 25-35.

I would like to show that workers prefer to stay in hotel. Which statistical test should I use? I thought about using chi-square test but my total percentages are higher than 100%.

$\endgroup$
1
  • 2
    $\begingroup$ Since you want to associate this multiple choice question with another question, then using an indicator variable for whether or not the hotel choice was made can be used in an ordinary statistical test of association. But it would be advisable to first guage the overall evidence for there being some category with a difference. This could be done by getting a likelihood ratio $\chi^2$ test in a binary logistic model where age group is the dependent variable and all the indicators are independent variables. $\endgroup$ May 10, 2019 at 16:43

1 Answer 1

2
$\begingroup$

To clarify the first part of Frank's comment, make a new variable where 0 = does not prefer hotel and 1 = prefers hotel. Those percentages will sum to 100. Then run a chi square test between that variable and the student vs. workers variable.

$\endgroup$
5
  • $\begingroup$ I have a doubt. If I want to say that workers prefer hotels, what should I do? Can I use the chi-square? $\endgroup$
    – Denisa
    May 12, 2019 at 19:11
  • $\begingroup$ Can you clarify what your doubt is? A chi square test between the new variable (does vs. does not prefer hotel) and the student vs. workers variable will test whether or not workers prefer hotels. $\endgroup$
    – Emily
    May 12, 2019 at 19:16
  • $\begingroup$ My doubt is related to the fact that I use chi-square to assess the association between two variable. So how can I write the null hypothesis and the alternative one? I know, maybe it is a stupid question but at the moment I am very confused about all these things $\endgroup$
    – Denisa
    May 12, 2019 at 20:28
  • $\begingroup$ I'm not following what your doubt is but the null hypothesis is that there is no difference, and the alternative hypothesis is that there is a difference. $\endgroup$
    – Emily
    May 12, 2019 at 22:33
  • $\begingroup$ @EmilyIf there is a multiple choice, construction of new variable (the way you have proposed) shal leave out/exclude the other simultaneous responses e.g. BB. Probaly, the approach for analysis of multiple responses is bereft of any logic. $\endgroup$
    – user10619
    Jul 18, 2019 at 4:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.