I have survey data to find when people continue to use a device or stop using it. The preferences are ordinal values, and the response is binary value of using the device or not. Which statistical model should I use
The obvious answer is: logistic regression. For binary logistic regression, the DV is binary, and the IVs can be numerical or categorical (ordinal or nominal) or a combination of these. If your sample size is large, CHAID analysis could add surprising insights into your data.
It might be interessting to think about linear discriminant analysis (LDA) as well, especially when it comes to classification questions as indicated by the binary nature of the dependent variable. LDA is a special case of regression analysis. The ordinal IVs however could pose probems.
Thanks Francois and everybody else for your answers. I was also thinking of binary logistic regression. My follow up question is: Do I need to create any dummy variables for my ordinal IVs?
It depends on the software. In SPSS, for example, you have to distinguish between "factors" (categorical variables) and "covariates" (scale variables). In the case of factors, the program automatically creates dummy variables with one of the categories as reference category. If you do not specify that an IV is a factor, it will auomatically treat that variable as a continuous variable.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling