1
$\begingroup$

I have a small survey dataset with 10 variables (Gender, International (Yes or No), and some Likert scale data) and all of them have categorical data. Having performed descriptive analysis, I was curious about what can I do to make good inferences. I have worked with a regression with numerical variables but never with so many categorical variables.

I was looking into dummy coding but I found articles that said too many dummy variables may give bad results. What are your options here?

My hypotheses are:

H0: There will be no significant difference in English menu preference between the international and national groups

H1: The international group will have a preference for an English menu compared to the national group.

The variables I have are as follows:

  • Language (English, German, Other)
  • International (Yes and No)
  • Difficulty with the current menu (Likert scale data)
  • Preference for English Menu (Likert scale data)
  • Gender
  • Age
  • Number of years lived in Germany (If the number is more, he or she might be okay with the current menu)
  • and three other variables.

I want to see how all these variables can affect Preference for English Menu.

$\endgroup$
5
  • 4
    $\begingroup$ What’s the objective of these analyses? $\endgroup$
    – utobi
    Mar 13, 2023 at 23:05
  • $\begingroup$ Before you run your statistical analysis, what hypothesis are you testing and how do you believe your data can answer that question? You can run a number of different types of stats tests with categorical data, but the goals of this analysis should be clear to narrow down what would work for you. $\endgroup$ Mar 14, 2023 at 1:18
  • $\begingroup$ This amazing book can be a good resource mregresion.files.wordpress.com/2012/08/… $\endgroup$
    – dherrera
    Mar 14, 2023 at 2:39
  • $\begingroup$ @utobi I live in a non english speaking country so what I am trying to see with this study is, how international and national people feel/prefer to the idea of having an English menu in a restaurant. My hypothesis are as follows H0: There will be no significant difference in English menu preference between international and national group and H1: International group will have preference for an English menu compared to national group. $\endgroup$ Mar 14, 2023 at 8:43
  • $\begingroup$ @Inzemam you should mention that in your question (you can edit it to add these information), because many people overlook comments. You should also detail how you want to model your hypothesis, because in your comment you mention two variables (nationality and menu preference), while in your question you mention 10 variables. So why were you mentioning these 10 variables? Was it just to give some context? Or are there additional things you want to test or control for? Add all these details in your question, it will help people to answer your question. $\endgroup$
    – J-J-J
    Mar 14, 2023 at 8:50

1 Answer 1

0
$\begingroup$

For your Likert scale questions if you have psychometricr questions, you can try drivers analysis. But again, before running this analysis you need to establish some objective/hypothesis.

key driver analysis, Or relative importance analysis, quantifies the importance of a series of predictor variables in predicting an outcome variable. It's very common in market research.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.