# When to use t distribution?

• Mar 11th 2010, 06:51 AM
Allan89a
When to use t distribution?
s(x) = 1.7 s(y) = 1.4
x-bar = 6 y-bar = 7.2
n(x) = 40 n(y) = 50

Find a 99% CI for the difference between mean x and mean y.

Now I solved this with a t distribution (which according to the key in the book is correct), but I don't understand why the central limit theorem can't be used so that I can use z-value from the cumulative normal table instead? Could anyone help explain why?
• Mar 11th 2010, 03:31 PM
mr fantastic
Quote:

Originally Posted by Allan89a
s(x) = 1.7 s(y) = 1.4
x-bar = 6 y-bar = 7.2
n(x) = 40 n(y) = 50

Find a 99% CI for the difference between mean x and mean y.

Now I solved this with a t distribution (which according to the key in the book is correct), but I don't understand why the central limit theorem can't be used so that I can use z-value from the cumulative normal table instead? Could anyone help explain why?

Do you know the population standard deviation .....? Or are you estomating it using the sample standard deviation ....?