Re: t-test question
- To: mathgroup at smc.vnet.net
- Subject: [mg64186] Re: [mg64150] t-test question
- From: Darren Glosemeyer <darreng at wolfram.com>
- Date: Sat, 4 Feb 2006 04:13:41 -0500 (EST)
- Sender: owner-wri-mathgroup at wolfram.com
As an aside to the responses that have already been posted, if you have the
data at hand, you can use MeanDifferenceTest to perform the test directly
from the data.
In[1]:= << Statistics`HypothesisTests`
In[2]:= data1 = Table[Random[], {20}];
In[3]:= data2 = Table[Random[Real, {.3, 1.3}], {35}];
This will give the signed t-statistic for a difference of 0, which is the
difference in the case you described, and the p-value for the two-sided test.
In[4]:= MeanDifferenceTest[data1, data2, 0, TwoSided -> True, FullReport ->
True]
Out[4]= {FullReport -> MeanDiff TestStat Distribution ,
-0.265388 -3.22715 StudentTDistribution[41.354]
> TwoSidedPValue -> 0.00244793}
The two-sided p-value is the p-value for the absolute value of the
t-statistic, which is the case you described.
Darren Glosemeyer
Wolfram Research
At 07:09 PM 2/2/2006 -0500, Csukas Attila wrote:
>Dear all,
>
>I have found an equation for comparing two means.
>
>Ue = (na - 1)*Ua + (nb - 1)Ub/na + nb - 2 This is for the pooled
>value of distribution and
>
>t0 = |xa-xb|/ Root Ue*(1/na+1/nb) This is for the t value.
>
>Sample values
>na=342
>nb=170
>Ua=6.39
>Ub=6.07
>xa=177.19
>xb=170.37
>
>Does anybody have idea how this works in Mathematica?
>I am not familiar with Mathematica syntax but interested how would this
>work...
>
>Any help and idea appreciated.
>
>Attila