|
V17
Sampling Distribution Theory Welcome to Part 12 of our video
series on sampling distribution theory. In this video, we're going to cover
the F distribution. We're going to cover the F probability table or,
otherwise known as, the table of critical values of the F distribution. We
are going to cover how to use this table and do some practice examples with
it as well. I'm Renee Clark from the Swanson School of Engineering at the
University of Pittsburgh. Okay, so, this is what the f
distribution looks like, or a picture of it. It looks very similar to the Chi
Square distribution which we covered in the previous video. It is a continuous distribution as well. Okay, it is skewed to
the right also because it has a long right tail. Okay, therefore, it is not symmetric…
is not a symmetric distribution like the T distribution is or the Z
distribution. Okay, it ranges from 0 to positive Infinity along the x-axis
just like the Chi Square distribution does. But, where
it differs from the Chi Square distribution is that it has two parameters: nu
one and nu two. So, it has… the F distribution has two different degrees of freedom.
Okay, each one is the sample size minus one. Okay, however, the order in which
you specify the two different degrees of freedom matters, and we will see
that in a… in an example problem. Okay, the convention is to write the
critical value of the F distribution as F sub alpha, just like you see right
there. So, F sub alpha sits along the x-axis. Alpha, again, is the area or
probability to the right of f sub alpha. Okay, so, alpha
here represents all the area under the curve to the right of the critical
value, F sub alpha. That is the convention. Okay, so, this is one of the F
probability tables in the back of your book, and it differs somewhat from the
table of critical values for the Chi Square distribution. And,
the reason for that is because we now have two different values for nu, or
two parameters, nu one and nu two. Nu one values are
specified across the top of the table. Nu two values
are specified down the left hand side. Okay, and,
because these column and row headers are now taken up by values for nu one and nu two, that limits the alpha values that can
be specified in these tables. Okay, so, the… for this particular
table, show the alpha value is .05. And that you know because the
subscript there is shown as 0.05. Okay, let me draw your attention
to the visual legend. Okay, hard to see but F sub alpha is your critical
value that sits along the x-axis. The area to the right of it under the curve
is alpha. Okay, so, alpha refers again to the area to the right of the
critical value. Okay, now, in the back of your book, there are actually two pages associated with alpha of .05. Okay, the
first page has nu values- nu one values- that range from… from 1 up to 9, and
then, that's page one, and then the second page picks up from there with a nu one value starting at 10 all the way up to Infinity. Okay,
the values for nu two remain the same across both pages. Okay, values for f,
of course, sit in the middle of the table and are read from the middle of the
table. Okay, so, let's try an example. What is the F value for nu 1 equal 4
and New 2 equal 6 degrees of freedom with an area of 0.05 to the right? Okay,
so, that means alpha equal .005. Okay, the convention is alpha is area to the
right. So, we just simply need to read this off of
the table. Okay, we need to make sure we're in the alpha equal .05 table, and
we know that because this header up here has a subscript of 0.05. Okay, so,
we're in the right table. We just have
to find nu 1 of four, right there, nu 2 of
six, right there, and read off the value that sits at the intersection of
those two points. And that value is 4.53. Alright, in the back of your book,
there are also two pages of critical values for alpha equal .01. Okay, so,
you can see right there- F sub .01. Okay, so, there are critical values for alpha
of .01 as well, and there's two separate pages. Okay, first
page goes from nu 1 equal 1 up to nu 1 equal 9. The second page picks up from
there- goes from nu 1 equal 10 out to infinity. Okay, okay, so, in the back of your book, you know, as far as your book is
concerned, critical values are only given for two different alpha values: 0.05
and .01. And that's just a matter of space in the back of the book. Other alpha
values could certainly be specified, but there's limited space in the back of
the book. Okay, and…and that's because, you know, our column and row headings
are now taken up by two different values of nu, or two parameters, and so
that's why you know the alpha values are limited in this case. Okay, so, in reality, generally we allow software to calculate
probabilities associated with the F distribution. However, it still is good
to be able to use the F table and have an understanding of
it. Okay, so, in terms of being able
to directly read critical values of f out of these
tables, you can only directly read critical values for f sub .05 and F sub .01-
the two different alpha values that are available. However, there's a bit of
a silver lining. You can also calculate critical values for two additional
values… critical values of f. Those having an alpha of 0.95 and 0.99. And what you'll notice
here is that these two additional alpha values are complimentary, if you will,
to the ones that are in the table, right? .95 and
0.05 add to one. .99 and 0.01 also add to one. Okay, so, um… um… theorem… by
theorem 87 in your book, that is how we can… we can use that to calculate
these two additional critical values for f. And eight… 87 reads as follows:
the f of 1 - alpha nu one nu 2, okay, is equal to the reciprocal of f of alpha. Let's take a note
that this subscript, 1 minus alpha, and alpha are complimentary, and that
they add to one, right? And that's why I was saying this- and this add to one,
right? Just as .99 and 0.01 add to one. Okay, so, it's equal to the
reciprocal, but the order of the nu’s is now changed. Instead of nu one, nu two, it's in the denominator: nu 2, nu 1. Okay,
so, let's try an example that will help to clarify this. Okay, say we want to
determine the value of f having 11 and 7 degrees of
freedom and an area of .99 to the right. Okay, so, what we're trying to find
out is f of N- .99, 11, 7. That's what we're trying to determine. Okay? Okay,
so, that's going to be equal to the reciprocal of the f value having a
complementary alpha subscript, right? Go back to our theorem here: if 1 - alpha
is .99, then alpha in the denominator is going to be .01, and those values we
can read directly from the table. But, you do have
to watch in that, here, nu
one is 11, nu two is 7. Those get reversed in the denominator, so you say 7
comma 11. And, remember, on the first slide, or… or
an early slide, we said the order of specification of the nu…
the nu values matters. Okay? Okay, so, when we then
finish this out, we need to find the f critical
value alpha .01, 7, 11. Okay, easy enough, we can read that directly out of
the table. This is our alpha equal .01 table. We know that because it's
specified right there. Okay, nu
one is seven. Nu one is across the top. Nu 2 is 11, down the left hand
side. We're going to read off the f value… val
that's at the intersection of those two. Okay, 4.89. So, in the denominator is 4.89. Okay, and when we do the math on that
we get .204. Okay, so, that's the critical value from the F distribution
having an area of 0.99 to the right and 11 comma 7 degrees of freedom. Just
fill this in 1 over .01, 7, 11 over 4.89. Okay, we wish to thank the
National Science Foundation for supporting our work under Grant 233582. Thank
you for watching. |