MathGroup Archive 2001

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: t distribution

  • To: mathgroup at smc.vnet.net
  • Subject: [mg28016] Re: [mg27984] t distribution
  • From: Tomas Garza <tgarza01 at prodigy.net.mx>
  • Date: Wed, 28 Mar 2001 02:40:45 -0500 (EST)
  • References: <200103270626.BAA21777@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

There is, indeed. Load the Statistics package and you'll find it is a real
treasure chest for the statistician. Since the t distribution is usually
associated with a specific hypothesis test, you may go to HypothesisTests:

In[1]:=
<< Statistics`HypothesisTests`;

and there you search for StudentTPValue which gives the p-value for
"teststat" in terms of the Student "t" distribution with dof  degrees of
freedom.

Another illustrative approach is to load the ContinuousDistributions
package, plot the Student-t cumulative distribution function, and search for
the t values corresponding to a given probability directly inside the plot
values. This is a useful resource. For example, suppose you have 20 degrees
of freedom,

In[2]:=
<< Statistics`ContinuousDistributions`;
In[3]:=
tt = StudentTDistribution[20];
In[4]:=
fd = Plot[CDF[tt, x], {x, -4, 4}];

Now use FullForm of the plot, which will give a huge list of things, among
which you'll see the actual plot points, i.e., the t values and their
corresponding cumulative probability. Then, you can choose whatever value
you desire, not being restricted to pre-specified levels of significance.

In[51]:=
FullForm[fd]

Try this, just for fun. You may refine the plot values by using the option
PlotPoints in Plot.

Tomas Garza
Mexico City

----- Original Message -----
From: "Jose Lasso" <jml at accessinter.net>
To: mathgroup at smc.vnet.net
Subject: [mg28016] [mg27984] t distribution


> Hi MathGroup,
>
> I want to know if Mathematica have some function that return the t
> value for a known confidence level, with known degrees of freedom?
> Thx in advance. Regards
>
> Jose M Lasso
>
> PS:Sorry for my english!
>



  • Prev by Date: Re: binomial distribution
  • Next by Date: Re: Exponential fit question.
  • Previous by thread: t distribution
  • Next by thread: Re: t distribution