Regression with dummy variable as dependent variable and likert scale as independent variable

[Thea Olsen]

Hello,

We have conducted a survey and aim to do regressions on the data.

We have a dependent variable, dummy/binary, 0 "Non-marginal customers" and 1 "Marginal customers". The independent variables (10) are on the same 5 Likert Scale ("Very important", "Quite important", "Neither", "Quite unimportant" and "Very unimportant"). In addition, we have data on gender, education, age (in groups) +++. We have 312 respondents.

Do you have any suggestions on which regression method we should use and how to conduct it? How do we treat the independent variables (should we generate dummy variables here?)

In advance, we appreciate your help.

[Carlo Lazzaro]

Thea:
welcome to this forum.
Exception made for linear probability models, two-level categorical regressands call for -logit- or -logistic-.
If you are dealing with survey data, -see -svy- prefix.

[Thea Olsen]

Thank you for the response, Carlo! We aim to use logit/logistic.

We have defined "Diversion" as dummy variables. Will it be correct to use i. in front of the independent variables to be able to conduct the regression? 5 is "very important" in this case. Or is there another way to treat the independent variables in a logistic regression?

Example:
logistic Diversion b5i.Interestrate b5i.Customerservice b5i.Bankslocation

[Carlo Lazzaro]

Thea:
your code is legal.
However, be informed that it consider level 5 of your predictors as the reference category.
See -fvvarlist- helpfile/entry for more information.