Statistics hypothesis test

[shaikht]

Hello,

If this isn’t the right place, please let me know or point me to where I should ask.

I’d like to do a hypothesis test, specifically, a 2-tailed test of the difference in means. The 2var app works fine for where the. Hypothesized mean is 0. In my case, the hypothesized mean is not 0 but I can’t seem to find a test for this.

Any suggestions, short of writing my own function?

[Wes Loewer]

(12-01-2022 03:48 PM)shaikht Wrote:  I’d like to do a hypothesis test, specifically, a 2-tailed test of the difference in means. The 2var app works fine for where the. Hypothesized mean is 0. In my case, the hypothesized mean is not 0 but I can’t seem to find a test for this.

Any suggestions, short of writing my own function?

I don't see a built-in way to do this on the Prime.

(It looks like it is available on NumWorks but not on the TI Nspire.)

[phobos]

In the Inference App under Hypothesis test, select the Alt Hypothesis μ1≠μ2.

[Wes Loewer]

(12-02-2022 12:37 AM)phobos Wrote:  In the Inference App under Hypothesis test, select the Alt Hypothesis μ1≠μ2.

That's the equivalent of μ1-μ2≠0. Please correct me if I'm mistaken, but I think the question was whether it was possible to compare the difference in the two samples' means to a value other than 0, such as μ1-μ2≠5.

[phobos]

If you subtract the value from x̄1 it should work.

[rawi]

Shaikht wrote:

Quote:I’d like to do a hypothesis test, specifically, a 2-tailed test of the difference in means. The 2var app works fine for where the. Hypothesized mean is 0. In my case, the hypothesized mean is not 0 but I can’t seem to find a test for this.

You can use the inference app:
Method: Hypothesis test
Type: Z-Test: Mue1-Mue2
Alt. Hypoth: Mue1 different Mue2
Goto Num and put in the needed values.
Example: Mean1 = 0.5; n1=50; Sigma1=0.2
Mean2 = 1; n2 = 80; Sigma2 = 0.3
Alpha = 0.05 (so you test with a significance level of 95%)
We want to test the hypothesis that the difference of the means of the universe is -0.333333.
Press Calc
You get two boundaries, lower = -0.585993, upper= -0.414007
-0.333333 is outside of the interval given by the boundaries.
Therefore the hypothesis that the difference can be 0.333333 has to be rejected.
Hope that helps.

[shaikht]

(12-01-2022 07:04 PM)Wes Loewer Wrote:  I don't see a built-in way to do this on the Prime.

(It looks like it is available on NumWorks but not on the TI Nspire.)

Hello Wes,

Yes, you are correct: the hypothesized mean is not zero in my case. My workaround is to split the test statistic calculation into 2 fractions, the first one for u1-u2 and the second for the hypothesized mean. Then, subtract the latter from the former. Thanks for taking the time to understand my original question!

[Wes Loewer]

(12-02-2022 09:42 PM)phobos Wrote:  If you subtract the value from x̄1 it should work.

Well, that was easy!

It makes sense, but I have never seen the formulas for when the difference is not 0 so I wasn't sure if a simple shift would be equivalent. I wasn't sure if the test statics and the p-values would be exactly the same, but a little experimenting on NumWorks shows that indeed they are.

Sometimes the simplest answer really is the best. :-)