# Which method to use for predicting variable a (score based on a likert scale) with an variable b and c

I have three variables. The energy behavior is a score based on a Likert scale. The other two variables age and education are nominal.

Now I want to show that the two nominal variables, age and education influence the energy-saving behavior.

What is a good suitable method in this situation? I am using SPSS.

Thank you.

## 1 Answer

Answering directly to the way you placed your question, if you think it's acceptable to assume that the Likert scale variable follows a Normal distribution and that both age and education are nominal, you can do a 2-way ANOVA.

However, assuming Normality on a Likert scale variable is far from being consensual. I'll assume that it is not a continuous variable, hence it may be treated as discrete or as ordinal.

Furthermore, it's common to have age as a continuous variable, when its range is large enough, or if you have the date of birth and the date of evaluation.
As for education it is common to use it as ordinal or a discrete scalar if you have the number of years that people went to school.

Hence, an ordinal regression may be the way to go if you assume that the Likert scale variable is ordinal.

Otherwise, if you assume that the Likert scale, age and education can be reasonably treated as scalars, a linear regression may be applicable.

• Age: 15-25, 26-35, 36-45, Over 45 // Education: Less than high school, High school, Graduate, College diploma, Bachelor's degree, Master's degree, Doctorate degree, Others // The energy saving behavior score is the mean out of 5 questions (Likert Scale) - Thank you Commented Mar 31, 2019 at 11:57
• I suggest that you add the above information into the question. Note that age and education are ordinal. In this case, the ordinal regression might be the more adequate approach. Commented Apr 1, 2019 at 2:52