2
$\begingroup$

The goal of the analysis is to find a correlation between birth context and stature, while controlling for age.

Each person has an ID number, age, stature, and birth context. That last part is what is giving me trouble. There are three type of birth context: foreign, which means they were born in another country, immigrant, which means they were born within 9 years of their mother moving to the US, and established, which means that they were born after 10 years of the mother being in the US.

I'm not sure how to control for age in this situation. Would I have to make separate dummy variables for each age, such as age 10, in the established group?

Anything that points me in the right direction is appreciated.

$\endgroup$
  • $\begingroup$ This may be useful to you. $\endgroup$ – Metrics Apr 20 '13 at 13:21
1
$\begingroup$

How to treat age depends on 1) How old the population is 2) How you've recorded age.

If you have recorded age only as a grouped variable - well, that's a shame, but you'd have to call it categorical. If you have age recorded in years, you can treat it as continuous. If you have the whole range from birth to adulthood, you will want to use some method (such as splines) that allows for changes in the relationship (people grow faster at different ages). Even if you have it from birth through adolescence, this would be true. If your age range is quite narrow, you might get a good model treating age linearly.

$\endgroup$

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.