I have conducted an online survey and I want to run a statistical test to determine of there is any correlation between, for example:
1. The number of years of experience a person has (selected choice question) and
2. What images are taken during an ultrasound examination (select all that apply question)
I can run frequency tables and crosstabs in SPSS Statistics via Analyze>Multiple Response>Define Variable Sets but I don't see any options to perform statistical tests here.
I have tried Analyze>Descriptive Statistics>Crosstabs but I believe I can't perform a Chi-Square test on multiple response sets when there is a "select all that apply" response?
What statistical test should I be using in this instance?
Thanks in advance :)
Hello Samantha,
Much of the answer hinges on how you characterize the multiple options variable (in your example, is image A chosen? is image B chosen? etc.).
1. If there are "right/preferred" and "wrong/not preferred" choices (e.g., image B should be chosen, image D should not be), you could create a score of correct responses (and correct omissions) over the set of choices. This then would be correlated with the age variable (via Spearman correlation if age categories/ranges are used as response options; via Pearson otherwise).
2. If the pattern of choices is classifiable qualitatively (e.g., someone concerned about abnormal liver function would choose images A, B, C and F), then you could assign each case to its closest matching category, and run your chi-square (with the associated phi or Cramer's V to reflect the correlation).
3. Otherwise, you might have to be content to evaluate phi values for each image individually (Yes/No).
Good luck with your work.
Thank you so much David Morse
I used your second suggestion and can now run chi-square and associated Phi and Cramer's V!!!!
Thanks for taking the time to explain this to me :)
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS