I have a dataset with one dependent variable and several independent categorical variables.

I want to explore the amount of variability each of the independent variables explain in the dependent variable.

This can be done by entering all the variables in a one-way ANOVA and calculating the proportions of explained variability of each factor (SSeffect / SStotal).

However, I am also interested in partitioning the proportion of explained variability of a specific independent variable between the different levels of that variable to see if some levels of a factor have more effect on the dependent variable than others.

How can I approach this task?


First of all, you stated something like

entering all the variables in a one-way ANOVA

Probably, you would like to use multiway ANOVA (maybe you mistaken multiway with Multivariate MANOVA?).

For example, you have a such situation:

  • DV: FlatPrice
  • IVs: City, District, Street, SquareMeters, Storey.

The District is nested in City and Street is nested in District as well as in City. In R environment it's expressed as FlatPrice ~ City/District + ...

Those IVs are properly nested and not crossed, so you could nested ANOVA or hierarchical ANOVA. Look, at this http://www.biostathandbook.com/nestedanova.html

By the way, you said something about assessing the severity of each IV. Maybe you could look at this question and answer, because it may help you Tests I sequential and III marginal

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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