There are so many discussions in our circle about validity test for nominal data in SPSS. Is it possible to do a validity test for nominal data in SPSS?
Hello Novel,
Without specifying what aspect(s) of validity you're referring to, it's hard to give a definitive answer. However, many types of validity evidence hinge on the strength of correlations between a measure (no matter its scale strength) and some suitable, external variable.
With nominal data, measures of association could include: (a) phi; (b) Cramer's V; or (c) Cohen's kappa (which gives association beyond that expected by chance). If there are just two levels to the nominal variable, and the external variable is continuous, you could use a Pearson correlation. If there are more than two levels, and the external variable is continuous, you could dummy code the nominal variable, and use it as a set of IVs in a multiple regression model, and use the model R (or R-squared) as the indicator of association.
There are other options as well, but perhaps this will give your inquiry a start.
Good luck with your work.
Hello Novel,
Without specifying what aspect(s) of validity you're referring to, it's hard to give a definitive answer. However, many types of validity evidence hinge on the strength of correlations between a measure (no matter its scale strength) and some suitable, external variable.
With nominal data, measures of association could include: (a) phi; (b) Cramer's V; or (c) Cohen's kappa (which gives association beyond that expected by chance). If there are just two levels to the nominal variable, and the external variable is continuous, you could use a Pearson correlation. If there are more than two levels, and the external variable is continuous, you could dummy code the nominal variable, and use it as a set of IVs in a multiple regression model, and use the model R (or R-squared) as the indicator of association.
There are other options as well, but perhaps this will give your inquiry a start.
Good luck with your work.
Dear Novel Idris Abas,
There is a Chi-squared Test to measure its significance to confirm its validity in the study.
I give below some links to understand how to perform the Chi-squared test for the variable validity.
http://learntech.uwe.ac.uk/da/Default.aspx?pageid=1435
http://learntech.uwe.ac.uk/da/Default.aspx?pageid=1440
Hope the above will help you in your studies. Good luck.
Regards.
If you follow the logic of construct validity, then you need to create hypotheses about how your new measure will be related to other existing measures, through processes such as convergent validity, divergent validity, predictive validity, etc.
David Morse has given you good advice about to assess those relationships, once you have specified them.
Please help me validate ny questionnaire to measure practice of self medication. It's nominal questionnaire with 50 responses as a oart of pilot study. I have SPSS ver.20 do don't have cohen kappa for that much number of participants.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS