Thanks,

GWFAMI

]]>The hypothesis are written exactly as my stat teacher (PhD) directed. Using the equation

t = ((u₁ - u₂ ) - σ) / (√(s² (1/n₁ + 1/n₂))

σ = 45 in my case, 0 in what you are suggesting.

u₁ =308

u₂ = 254

s² = 3672.5

n₁ = 20

n₂ = 20

If you use σ = 0, then the resulting t = 2.8178, but if you use σ = 45, then the resulting t = 0.4696

Does that help?

Thanks,

GWFAMI

Sorry for the way the equation is written, but it's the best I could get in as it will not allow me to post any pictures.

]]>Secondly, the two hypotheses can be rewritten to have both means on the same side:

u_{1} - u_{2} = 0

u_{1} - u_{2} > 0

I'm not totally sure, but I believe writing the hypotheses this way and replacing the zero with 45 (or whatever difference you're testing for) may allow you to do the test.

]]>I have an instance where I'm doing a 2 sample t test, using the following:

H_{0}: u_{1} <= u_{2} + 45

H_{1}: u_{1} > u_{2} + 45

I have no issue doing a standard 2 sample t test where

H_{0}: u_{1} <= u_{2}

H_{1}: u_{1} > u_{2}

But how or where can I put in the "+ 45" in the built in stat test?

Thanks,

GWFAMI

]]>