I’m sure this is no big deal, but the subtleties are beyond me:
What are the differences between using chi squared and a t test when examining data? I know both can be used to calculate the probability of a null hypothesis, but are there times when one is more appropriate than the other?
Thanx in advance,
mycoman
The test that you should use for the null hypothesis depends on the distribution of your test statistic. If the test statistic is distributed according to a t distribution, you use a t test. If it’s distributed according to a chi squared distribution, you use a chi squared test. If it’s distributed according to an F distribution, you use an F test. If it has some other distribution, you use the appropriate test.
Thanx, Ultrafilter, I think you’ve helped me realize where my deficiency in understanding lays.
How does one recognize the aforementioned distributions?
(apologies for the unplumbed depths of my ignorance, but the help really is appreciated.)
mycoman
Unless you either have a graduate degree in statistics or are working on one, you generally need to have that told to you.
If you have a specific problem in mind, post that, and you’ll get more specific (i.e., useful) answers.
Believe it or not, Ultrafilter, that pretty much answered my question.
I had learned how to do a chi-squared in the past, and was being asked to do a t test now, and was uncertain why. Although I’m sure a more thorough knowledge of the background would be interesting, in this case it would only distract me from what I need to accomplish.
Thanx again
mycoman
Well, if you have some handy-dandy software (such as Minitab,) then you can actually put in a set of data and test it to almost any type of distribution. I’m currently taking a class all about this kind of stuff, and let me tell you, having a computer do it for me makes it a lot less tedious.