# Hypothesis Testing: One proportion or Two?

• Dec 7th 2009, 05:29 PM
Empirical-Design
Hypothesis Testing: One proportion or Two?
In my statistics class we've been learning about hypothesis testing as of late, and while I feel I get it, my professor has sent me home with a problem that I can't for the life of me solve.
A report on health care in the US said that 28% of Americans have experienced times when they have been able to afford medical care. A news organization randomly sampled 801 American Minorities, of whom 38% reported that there had been times in the last year when they had not been able to afford medical care. Does this indicate that this problem is more severe among American Minorities?
I'm hung up on whether or not I'm supposed to test for one proportion by mostly ignoring the original 28% and use the null hypothesis p = .38 and n value 801, or if I'm supposed to compare two proportions, in which case the n value for 28% of Americans is mostly unknown (no definitive number given to explain the sample of the test.)

On another hand, I could be off entirely on what I am supposed to do here, in which case the need for help is even more dire. ><

Got anything for me, MHF? All help is greatly appreciated. (Happy)
• Dec 9th 2009, 12:43 AM
ANDS!
Read the question again. They are asking you if the rate of medical coverage is worse in minorities than the general population (meaning are minorities bad off, or are they pretty much average with the entire population). Thus the hypothesis you are testing is whether or not those 801 minorities sampled actually come from a population whose true proportion is 28% or if they indeed come from a different population. So yes, you need the original proportion since that is what you are testing.

Do a thought experiment; assume the report on health care is true. If you sampled 801 people (all minorities), and found that the proportion of those who didn't have medical coverage was 38%, how far from the population proportion is that? See if you can take the problem from here.