Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.

I have a set of cancer counts (incidence) for difference racial groups for 4 different types of cancer. I'd like to know if there is a difference in cancer incidence among the racial groups. So the data look a bit like this:

Cancer Sites         All Races    White     Black   Asian/Pacific Islander 

Digestive System    144,010       119,330   17,071  5,242

Colon and Rectum    73,183        61,450    8,177   2,243

Colon excl. Rectum  50,492        42,366    5,911   1,380

Rectum              22,691        19,084    2,266     863

Any ideas about how to approach this?

share|improve this question
Related to Michael's response - what is your target population? Do you know the number of individuals from each racial group in your target population? – Macro Jun 29 '12 at 15:34
If the answer to Macro's question is no then I would be very cautious about trying to analyze the count data. Large counts for one race relative to the others might simply mean that the population for that race is much larger than for the others – Michael Chernick Jun 29 '12 at 15:42
You should still be able to say something if the proportions of the cancers are very different. – Douglas Zare Jun 29 '12 at 16:13
You say "$4$ different types of cancer," but it looks like the second category is the disjoint union of the third and fourth. Are these all contained in the first? – Douglas Zare Jun 30 '12 at 6:55

1 Answer

I think the actual count comparison could be misleading because in your population the races are probably not equally distributed. Change the counts to proportions by dividing total number of individuals in your population from that particular racial group. You can then compare the proportions using contingency table analysis (chi square approximation or the Fisher exact test).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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