0
$\begingroup$

I have a dependent continuos variable and two independent variables, both categorical. The first independent variable is the "gender" of respondents (two levels), the second one is their "first language" (5 levels: english, french, ... and other 3).

I would like to test whether the dependent continuos variables is significantly different between male and females, controlling also for their "first language". On some tutorials I found that ANCOVA could be a solution as an extended version of ANOVA, controlling for a covariate. However, I also read that ANCOVA requires covariate to be CONTINUOS variable and not Categorical, such as is my case with the Categorical variable "first language".

Is there any other solution available? Is SPSS (i'm using this software) working the same if I use a CATEGORICAL covariate instead of a continuous one? What do you suggest could be the best option to pursue? Maybe a Two-Way Anova could work? But in the case of a Two-Way Anova I already know the design would be unbalanced (= without having the same number of cases for each group). What would be then the implications?

$\endgroup$
0
0
$\begingroup$

You might like to have a look at this post asking the exact same question about categorical covariates.

"Is SPSS (i'm using this software) working the same if I use a CATEGORICAL covariate instead of a continuous one?"

The answer to this I have from Baguley (2012). He says that different softwares implement it differently. Probably different commands also results in different results. He reccommends to create one's own syntax unless one is sure what the software does. With his guide I created a repeated ANOVA and multilevel models in R. I'm not a statistician, mathematician or econometrician, hence I like to draw on people with more knowledge.

Regarding your unbalanced design question, I did a quick search on this forum and discovered a wealth of answers. I guess this is why your question hasn't been answered until now.

$\endgroup$

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