statistician around?

[bartholomew]

Hi thanks for your previous answers.

I have the following problem: i was asked this question

are there differences in the number of incidents of violence recorded for patients of different ethnic background?

I know that in the patient population there were 543 attacks on staff. the problem is that in 80% of cases ethnicity was not recorded.

Do u think that the analysis makes sense

i did a chi test in the 20% observable spectrum but I do not think the inferences are valid for the whole poulation.it doesnt What do u think?

Is the question unswerable?

[c00ky83]

Well 20% WERE recorded, which is a sample of 109 - a good amount in my book. It's down to your assumptions - if you think the 20% is a fair representation (unbiased) then you're OK. But you just don't know - that's the thing about stats, you can use them to prove or disprove anything you like!

[Peter Pedant]

I hope you find an answer you can use B, you fist asked this a few months ago.

I think you need to go and SEE a statistician.