paired samples t test in imsl via Fortran 77 or 90+

Can anyone tell me how imsl's atwob subroutine can be used to conduct
a 2 paired samples t test? Suppose you have x1 = 3, 4, 6, 9, 11, and
x2 = 4, 4, 9, 8, and 12. Thanks!

On May 4, 1:42*pm, Shlomo Sawilowsky <professorshl...@gmail.com>
wrote:
> Can anyone tell me how imsl's atwob subroutine can be used to conduct
> a 2 paired samples t test? Suppose you have x1 = 3, 4, 6, 9, 11, and
> x2 = 4, 4, 9, 8, and 12. Thanks!

Oops! I figured it out: it is EFSS(5) and EFSS(7). IMSL remains as
cryptic as ever!

Shlomo Sawilowsky <professorshlomo@gmail.com> wrote in news:3468c71a-4877-
4c1d-92bf-907d6eb02950@36g2000yqi.googlegroups.com:

> On May 4, 1:42*pm, Shlomo Sawilowsky <professorshl...@gmail.com>
> wrote:
>> Can anyone tell me how imsl's atwob subroutine can be used to conduct
>> a 2 paired samples t test? Suppose you have x1 = 3, 4, 6, 9, 11, and
>> x2 = 4, 4, 9, 8, and 12. Thanks!
>
> Oops! I figured it out: it is EFSS(5) and EFSS(7). IMSL remains as
> cryptic as ever!
>

Not as cryptic as both your mails.

The 't' test is better known as the ' Student's t test'.
"A student" was the publication psuedonym of R.A. Fisher.

t= |(error in Mean Value)|/(Standard Error of mean).

The divisor is the standard deviation divided by the square root of the the
number of measures. The dividend is simply the positive difference of the
population mean and the sample mean. So you measure the standard error of
mean and to calcualate 't', use the knowledge of the actual mean for the
universe of possible events.

The 't' value so calculated is then used to find a locus between two
adjacent curves of probability significance of a logarithmic plot of the
caculated t value against the logarithmic plot of the number of degrees of
freedom.
These curves of significance level is known as the 'Student's t
distribution' for levels of confidence (from near-zero to near certainty).
See a statistics text for the equations of the distribution, which are quite
complex. I can offer a look-up table of factors and a simple equation to
generate a very close approximation. Conf=f((t,n))

So for two samples you can now compare their t-values and their confidence
levels.
From this you can now calculate the probability that the two samples belong
to the same population.

Terence <tbwright@bigpond.net.au> wrote:
> The 't' test is better known as the ' Student's t test'.
> "A student" was the publication psuedonym of R.A. Fisher.

> t= |(error in Mean Value)|/(Standard Error of mean).

And the reason why Guiness is the statistician's favorite beer.

-- glen

On 2012-05-05 03:50:40 -0300, Terence said:

> The 't' test is better known as the ' Student's t test'.
> "A student" was the publication psuedonym of R.A. Fisher.

From Google one gets the more usual history. Nothing to do with Fisher
but much more to do with Guiness and their fine beers.

>> History
>> The t-statistic was introduced in 1908 by William Sealy Gosset, a
>> chemist working for the Guinness brewery in Dublin, Ireland ("Student"
>> was his pen name). Gosset had been hired due to Claude Guinness's
>> policy of recruiting the best graduates from Oxford and Cambridge to
>> apply biochemistry and statistics to Guinness's industrial processes.
>> Gosset devised the t-test as a cheap way to monitor the quality of
>> stout. He published the test in Biometrika in 1908, but was forced to
>> use a pen name by his employer, who regarded the fact that they were
>> using statistics as a trade secret. In fact, Gosset's identity was
>> known to fellow statisticians.


>
> t= |(error in Mean Value)|/(Standard Error of mean).
>
> The divisor is the standard deviation divided by the square root of the the
> number of measures. The dividend is simply the positive difference of the
> population mean and the sample mean. So you measure the standard error of
> mean and to calcualate 't', use the knowledge of the actual mean for the
> universe of possible events.
>
> The 't' value so calculated is then used to find a locus between two
> adjacent curves of probability significance of a logarithmic plot of the
> caculated t value against the logarithmic plot of the number of degrees of
> freedom.
> These curves of significance level is known as the 'Student's t
> distribution' for levels of confidence (from near-zero to near certainty).
> See a statistics text for the equations of the distribution, which are quite
> complex. I can offer a look-up table of factors and a simple equation to
> generate a very close approximation. Conf=f((t,n))
>
> So for two samples you can now compare their t-values and their confidence
> levels.
> From this you can now calculate the probability that the two samples belong
> to the same population.

You are quite correct.
'Student' was the psuedonym of W.S. Gosset, and who published the idea of
the measure; R.A. Fisher is the man most associated with the with the
research and theory development of the validity of the test as a confidence
measure, and the development of the maths of the confidence level
calculations of the distribution.

Before using the T-test, one should check the validity of the assumption
that the variation in the sample measures are suffficeintly close to be
independent measures of the same popultion. This test, the variance ratio
measure, is known as Snedecor's F test, which was so named by him in honour
of R.A. Fisher.

This was why I trotted out the wrong name (my maths was done in 1953 when
the origins and associated names of the t and F test symbols were all fresh
to me; but time passes..).
I still have my 1951 "Facts from Figures" by M.J. Maroney.