Hypothesis Testing - Z-test (Excel)

In this video, we'll see how to do the full hypothesis testing procedure using Excel and using the
In this video, we'll see how to do the full hypothesis testing procedure using Excel and using the
z-test. I'm gonna create a spreadsheet that will calculate and display and organize all I think we
need, labeling it for testing rain here, and this is where the population deviation is looking at
example 9.1 from the book. Here's where you're not given sample data; rather, you're given
summary statistics, so you can easily adapt this or use sample data.

We'll start with the hypotheses, so we need a null and alternative hypothesis. We're testing the
mean, so we're going to use the lowercase Greek letter mu. Now to determine the inequalities,
we need to find the claim and the problem and then translate that to hypotheses. You had some
experience with that in the previous unit, and here we can actually see that there's a claim by
Frank that the goggles help Jeffrey swim faster than 16.43 seconds. So the claim is that mu,
which represents the population mean of Jeffrey swimming with the goggles, is less than his
swimming faster time, less than his previous average time of 16.43. So that creates the
alternative hypothesis, and then we use the same number in the null hypothesis.

So we can ascertain the type of test in terms of the direction; this is a left-tailed test that should
be apparent from the inequality in the alternative hypothesis pointing to the left the type of
distribution we use is an X and we'll be using the normal distribution since the population
standard deviation it is represented by the letter K squeek letter Sigma isn't. Next, we're gonna
state our level of significance that's represented by a lowercase Greek letter alpha and we're
told that that is 0.05. You now record our sample statistics just summarize the sample that were
not actually given it's it's a test for the mean we should have a sample mean and sample size
you always have that and then a standard deviation is usually given in the mean tests in this
case it's the population standard deviation sample mean is 16.

Yeah, so simple mean a 16 it says Jefferies mean time was 16 seconds the sample size is 15 it
says for the 15 swims and the population standard deviation is 0.8 0.8 we're now going to adjust
the standard deviation to get the standard deviation of the sampling distribution and remember
from experience 8 that is found by taking the population standard deviation and dividing by the
square root of the sample size. Next, I'll do that for us and we're now ready to calculate a
p-value that's a left-tailed test so we will have the left tailed p-value but so we can do fewer
examples and generalize what's going on I'm going to go ahead and put in formulas for all the
p-values and so we're using the normal distribution and we want the area to the left for a left tail
so do I missed the mark off for the tail is gonna be the sample mean the mean of the sampling
distribution is the hypothesized population mean the number that shows up in the hypotheses
the standard deviation is the adjusted standard deviation the one for the sampling distribution
because that's where the p-value is represented and then cumulative we're gonna use one so
that gives us the left tailed one now if this were a right tailed test we'd want to get area to the
right and so then you'd need to do 1 minus the same norm just default finds area to the left and
the total area is one area they're gonna be 1 minus just so I'm just gonna put this stuff in so you
can change these values later if you want.
Yeah, right-tailed. Now, if it's a two-tailed test, you would have done the left tail to the right tail
based on, you know, which side that sample value was from the mean. And then you would just
double it, and since the p-value is usually the smaller of these two, I'm gonna go ahead and say
this will work most of the time, but of course, double-check that kind of thing.

Now, of course, the one we're using for this problem is the first one just so you can reuse this
spreadsheet later. Now we want to get a picture of this p-value so we know what's going on.
Let's go to that online tool which is linked from the experience 12 folder. We're gonna get area
from the value P values in the area. The mean would be the number that shows up in the
hypotheses. The standard deviation is that adjusted one. There's the same numbers we used
when we put them in the norm dist, and we'll just brown that to 207. And then in a left-tailed test,
we would want well below the left 16. So we can now get a nice picture of this. All right, that
looks pretty good. So there's a picture of the p-value, which is something that you're asked to do
in the application, and I think this is the ideal way to do it. You, of course, can draw this. You
should have something similar drawn and then have this labeled where you're working off the
tail, okay, which that tail is marked off at 16 because 16 was the sample mean, and the area of
that tail is the p-value. So we did a hand-drawn graph, mark off where that tail is with 16, and
then I marked off the mean in the middle. Hi. Okay, you know, let's to make this right. Let's just
grab all this, that way I don't know if you, so go ahead and just take a screenshot or a snip of the
whole thing. That way somebody can easily check your book. So now we know it's the area
below 16, and we know what that area is, and we can see that the correct mean is standard
deviation being used. All right, decision time. So remember the decision. We compare the
p-value with alpha, and the p-value was about 0.02, and alpha is 0.05. So we'd say, since P is
less than alpha, we reject the null hypothesis. All right, now we got to have a conclusion which
relates this decision to reject the null back to the original claim or that the original claim was
made by Frank that the goggles would help Jeffrey swim faster. And that corresponded to the
alternative hypothesis since we rejected the null. We actually do have evidence to support the
alternative, and we could say the following: Sample data supports the claim that the goggles
helped Jeffrey swim faster at the 0.05 level of significance. There's something along those lines.
All right, and that's a complete hypothesis test for this problem. No, you could also do this same
Z test for the mean where you were given sample data, and how would this differ? Let's take a
look. So in the situation of working with sample data, you would still diagnose your hypotheses
in the past and type of distribution method significance. But when you get to sample statistics,
you would have to calculate these. So we couldn't create a column here where you'd actually
put in the sample data, and you would have to paste in the numbers here. So I'm just going to
have to put in a bunch of numbers. Have this? So 16, and let's get 15 of those. Right? So here's
$15, and the sample mean is 16. Just so that the numbers actually match up. In general, you'll
have a list of data there, and then the sample mean will be the average, and we'll just go ahead
and select that. I can go way down here and not really cause any problems in that way if I have
a longer data set later on. You know, as long as it doesn't go past row 119, it'll just include those
numbers. And when you have those blank cells, it doesn't include the blank cells. So that'll work
really well there. The sample size can be a formula to the count function. We'll do the same
thing. It'll just count, and it just counts the ones that have numbers put in there. So you can now
overwrite this. Remember, if you put in a shorter data set, you'd want to delete the additional
numbers that you are overriding. Everything else is automated. You were given population
standard deviation for the Z-test, and then this formula is automatically updated, and the
p-values are automatically updated. So you would just be changing that. And this is the Z-test.
So maybe you followed along, and you've created one of these. Maybe not. I provided this as a
model spreadsheet that you can edit, and I've shown which cells should be edited, which should
be left alone. At some point, we'll be looking at doing a t-test.