Even if it were continuous-interval, you can transform complex data to be simpler. For example, on Becks Depression Inventory, you receive a score for depression at the end. You can leave it as the score, but you can also transform it to determine level of depression (i.e., mild, moderate, severe), based on point ranges (‘if you earned between these two amounts, you have earned this label). It really depends on what you do with the variable. How are you treating the variable? How are you operationalizong the variable? ANOVA is blatantly only for categorical, but if you treat your data as categorical, it would still work.

Thank you for your answer, Deborah.

I have a dependent variable that measures the intentionality to pay for a service. This variable is measured with a 5 point likert scale (1=little intention to 5=much intention). In turn, I have created 3 different stimuli, and I want to analyze if there are differences in the intention to pay for the service between the groups exposed to these stimuli.

Im not sure if I can use the likert scale measure because it is not strictly continuous, but ordinal.

I see what you mean. If you use the likert scale for group placement, it would be appropriate but it sounds like you may need to consider chi square. Have you given any thought to that analysis?

To clarify, you could still do ANOVA because you have a score from your variable. The Likert response options are categorical but we almost always transform it into interval. It really depends on what you want to do. I would recommend you make the decision based on what analysis would provide an answer to the research question and let that guide you. But ANOVA would still work

Thanks Deborah Pauli

Your response has been very helpful. I will think about the chi-square option also.

Thanks again.