I am trying to perform an ordinal regression. My dependent variable is ordinal as well as my independent variables. However, I've created concepts for the independent variables by taking the mean of the questions that were 'assigned' to it (these questions were measured on the same Likert-scale). Now that I see the output of the ordinal regression, I have a large amount of categories. For example, 1 variable (sorry for the weird layout):
[Quality_of_Personalization=1,17] [Quality_of_Personalization=1,33] [Quality_of_Personalization=1,50] [Quality_of_Personalization=1,67] [Quality_of_Personalization=1,83] [Quality_of_Personalization=2,00] [Quality_of_Personalization=2,17] [Quality_of_Personalization=2,33] [Quality_of_Personalization=2,50] [Quality_of_Personalization=2,67] [Quality_of_Personalization=2,83] [Quality_of_Personalization=3,00] [Quality_of_Personalization=3,17] [Quality_of_Personalization=3,33] [Quality_of_Personalization=3,50] [Quality_of_Personalization=3,67] [Quality_of_Personalization=3,83] [Quality_of_Personalization=4,00] [Quality_of_Personalization=4,17] [Quality_of_Personalization=4,33] [Quality_of_Personalization=4,50] [Quality_of_Personalization=4,67] [Quality_of_Personalization=4,83] [Quality_of_Personalization=5,00] [Quality_of_Personalization=5,17] [Quality_of_Personalization=5,33] [Quality_of_Personalization=5,50] [Quality_of_Personalization=5,67] [Quality_of_Personalization=5,83] [Quality_of_Personalization=6,00] [Quality_of_Personalization=6,17] [Quality_of_Personalization=6,50]
My question : can I treat the independent variables as continuous because of all the different possibilities as number when calculating the mean? I know techniqually you should have a 'higher' level of measurement in order to treat it as another measurement level, but treating it as continuous gives a much cleaner output in SPSS, which I can actually interpret... so I really hope this has been done before or can be done.
* If you have any resources if this would be possible that would be even greater.
If not, how should I interpret all these different categories of all the variables? It is a very large table..
Yes, means and other summary statistics are treated as continuous because they can hypothetically take on a continuum of values. For example the Bernoulli probability is a continuous estimate (bounded by 0 and 1, but can take any value in between) of a binary system.
Thank you for your answer, James! Do you maybe have a source for this?
- Badges
- Science topic
- Similar topics
- Psychometrics
- Measurement