(female narrator)
This is just a couple of tips
on how to use the
significance test flow chart.
As you go through each problem, this
flow chart can help you to determine
what type of significance test you
should perform on that problem.
The first thing you want to do is you
want to read the problem.
When you're done reading the
problem, you need to ask yourself,
"What was the data collected from
each member of the sample?"
For instance, did you collect
a weight, or a height,
or did you take their temperature,
or did you count the number
of dogs that they had?
That would be numerical data.
When you read the problem,
did you see some key words like mean,
or average, or standard deviation.
All of those things indicate that you
have a quantitative data problem.
On the flip side of that, were there some
keywords like proportion, percent, rates,
or was each member of the
sample asked a yes or no question,
such as, "Have you had
a heart attack? Yes or no?"
Maybe they were asked, "Are you
overweight? Yes or no?"
"Do you support the
president? Yes or no?"
So that type of a thing. That
would be qualitative data.
Once you determine the type of data,
then you follow the decision
tree down that side.
For instance, if you have quantitative
data, then you would ask yourself,
"Do I have two populations or
do I have one population?"
Now remember, an easy way to tell
whether you have one or two populations
is to look for the samples
that are given in the problem.
If you're only given one sample,
then you only have one population.
If you're given two samples,
then you have two populations.
You have to be given the complete
information about both of those samples,
like the sample mean and sample standard
deviation, for two separate samples
in order to determine
you have two populations.
Once you determine how
many populations that you have,
if you have one population, you're going
to do the t-stat one sample.
If you have two independent populations,
then you're going to do the
t-stat two samples.
Keep in mind that you need to determine
whether you're going to pull the variances
or not pull the variances.
Down here, we have the little
tip on how to decide that.
It's just the ratio of the larger
sample standard deviation
divided by the smaller sample
standard deviation,
and looking to see if it's greater
than two, then you remove
the pulled variances check mark,
and if it's less than two,
then you keep the pulled
variances check mark.
Now remember if you have
two dependent samples,
the two samples are related to each other,
such as a before and an after,
a pre and a post test. The two samples are
related because it's the same person
taking the pre-test and the
same person taking the post-test,
or maybe you're testing
two different kinds of tires,
so you take a car and you
drive it with the first tire,
and then you take the same car and you
drive it with the second tire.
Those two samples are related because
it's the same car driving both times.
If they're dependent samples,
then you have a choice
of either finding the differences
between the two samples
and doing a one sample t-test, or under
t-stats you can do the paired t-test.
If you have more than two
populations, we only have one test
where we had more than two populations
and that was from chapter 14,
and that was where we had three, four,
five, six, as many populations
as were needed, and that was to perform
the one-way ANOVA test.
Now, if you have two quantitative
variables, like you were collecting
the height of a person and
the weight of a person,
and you were asked to decide if the
variables were dependent,
or associated, or independent, if
you saw those words in there,
but you have two quantitative
variables, a height and the weight,
then that would be a
regression type problem,
and you would perform that significance
test using the test on beta, the slope,
using the regression simple linear
command in StatCrunch,
so again that's for two
quantitative variables,
and you're going to look for phrases
like "Determine if the variables
are dependent, or
associated, or independent"
Back over here on the qualitative
side, it's a little bit shorter
on the qualitative side, but starting out
the same way we did on the quantitative
side, we need to determine if we
have one or two populations.
Again, if you have two populations then
you're going to have information
for two samples. You're going to be
given the number of successes
and the total sample size
for two different samples.
If you do determine that
you have one population,
then you would do a
proportion stat one sample.
If you do determine that you
have two populations,
then again you're going
to do a proportion stat,
but you're going to do two samples.
Our last test for qualitative data,
is if we have two variables.
For examples, we might be looking at
gender and happiness.
That would be two qualitative variables.
For gender, you would be male or female.
For happiness, they might indicate very
happy, or happy, and not happy,
so those are word answers to those.
"How happy are you?"
Very happy, happy, not happy.
Those are word answers so those are
definitely qualitative variables,
and it was two variables.
We were collecting gender
and we were collecting level of happiness
so we have two qualitative variables.
In the problem you can also look to see
if it says the variables are dependent,
associated, or independent.
Also note, a little hint here, that when
you do have two qualitative variables,
it's typically the data will be
shown in a contingency table,
so all of these clues here
help you determine
that you're going to perform a
chi-square independence test.
So this is a little bit about how
to use this decision tree,
or significance test flow chart.
Sometimes I call it a decision tree
because it branches off,
or sometimes I just call it a
significance test flow chart.
But anyway, this is a little bit about
how to use this document.