testing claims about the proportion
TRANSCRIPT
TESTING CLAIMS ABOUT THE PROPORTIONLecture 17
TESTING CLAIMS ABOUT THE PROPORTION Last lecture was the hard part
Now you know everything you need to know about testing claims and can apply your knowledge to testing claims about other parameters besides the mean
This lecture we’ll test claims about the population proportion… but only a very small part of this process will be at all new to you
You remember the basics from when we covered estimating population proportions in Lecture #14 If a variable is binomial, we call p the proportion in the population for which the variable was
in one of the two categories
In a sample from that population, the sample proportion is called 𝑝 and equals 𝑥
𝑛, where x is
the number of times the variable was in the category of interest and n, of course, is the sample size
TESTING CLAIMS ABOUT THE PROPORTION STEPS IN TESTING CLAIMS
A. State the hypotheses and identify the claim.
B. State the values of x and n.
Perform the test using the sample data or statistics, and make a diagram showing the distribution, the parameter and the statistic.
Shade and label the p-value, which should be rounded to the nearest thousandth.
C. Decide whether or not to reject the null hypothesis.
D. Summarize the results.
You’ll note that these steps from before are perfectly applicable to testing claims about the proportion, with the addition of stating the values of x and n.
The only differences are:
That the parameter is p instead of 𝜇
The value of the parameter mentioned in the claim is 𝑝0 instead of 𝜇0
The statistic is 𝑝 instead of 𝑥
We will never use the t-distribution in testing claims about the proportion, only z
TESTING THE FIRST CLAIM
Here’s the claim: At most one-third of people stopped by the CHP for suspected DUI test above the legal limit
To test this claim, a random sample of 200 people stopped for suspected DUI is selected, and of these it turned out that 36% tested above the legal limit
Let’s choose the 5% significance level
A. Remembering to use p for the parameter, and that ‘at most’ encompasses the ‘is less than’ and the ‘equals’ options, the hypotheses become
𝐻0: 𝑝 ≤1
3(Claim)
𝐻0: 𝑝 >1
3
This is a right-tailed test
TESTING THE FIRST CLAIM
Let’s skip the build-up and go directly to the Stat Test menu on the calculator
We’ll be using 1-PropZTest
Putting 1
3in for 𝑝0 seems obvious
when we enter that the calculator uses this repeating decimal in all its glory: .3333333333…
But the next line asks for x, which is the number of people out of the 200 who tested above the
legal limit, and that’s information we weren’t given directly
We’ll have to use a little arithmetic, or a little algebra
TESTING THE FIRST CLAIM
Perhaps you realize that since 36% of the 200 people in the sample tested above the legal limit,
to find out how many people that actually is you’ll take 36% of 200 and come up with x = 72
If you don’t readily see that, you’ll have to work harder
Take the equation 𝑝 =𝑥
𝑛and substitute 0.36 for 𝑝 and 200 for n:
0.36 =𝑥
200
Multiply both sides by 200 and again you’ll get x = 72
TESTING THE FIRST CLAIM
Either way, the 1-PropZTest screen looks like this:
Upon calculation, the screen becomes:
Watch out for all the p’s
The one with the hat is the sample proportion (which we didn’t actually input – the calculator
computed it from the x and the n), and the one without the hat is our p-value
TESTING THE FIRST CLAIM
B. x = 72; n = 200
C. Here again is the way to decide whether or not to reject the null hypothesis: If 𝑝 < 𝛼, reject 𝐻0 If 𝑝 > 𝛼, don’t reject 𝐻0
In this case, 0.212 > 0.05, or 𝑝 > 𝛼, so do not reject 𝐻0
TESTING THE FIRST CLAIM
D) Here again are the rectangles for summarizing the results:
Since we didn’t reject 𝑯𝟎 and since the claim was 𝑯𝟎, we’re in the upper-right triangle
There isn’t sufficient evidence to reject the claim that at most one-third of people stopped by the CHP for suspected DUI test above the legal limit
Reject 𝑯𝟎 Do not reject 𝑯𝟎
The claim is 𝑯𝟎
There is sufficient
evidence to reject the
claim that…
There isn’t sufficient
evidence to reject the
claim that…
The claim is 𝑯𝟏
There is sufficient
evidence to support
the claim that…
There isn’t sufficient
evidence to support
the claim that…
TESTING THE SECOND CLAIM
Here’s the second claim: The majority of Mendocino College students are female. Let 𝜶 =𝟎. 𝟏𝟎. Our sample is the Class Data Base, in which 60 of the 102 people in the sample are female
What is a majority? It’s more than half
If a group is split 50-50, there isn’t a majority
Any time you see the phrase “A majority…” you don’t have to read any further to translate it to a mathematical sentence It’s always p > 0.5.
Likewise, “A minority…” always translates as p < 0.5
And of course it’s always the alternative hypothesis. The null hypothesis has to include the ‘is less than’ and ‘equals’ options
TESTING THE SECOND CLAIM
Here’s the second claim: The majority of Mendocino College students are female. Let 𝜶 =𝟎. 𝟏𝟎. Our sample is the Class Data Base, in which 60 of the 102 people in the sample are female
A. 𝐻0: 𝑝 ≤1
3
𝐻0: 𝑝 >1
3(Claim)
B. x = 60; n = 102
TESTING THE SECOND CLAIM
Here’s the second claim: The majority of Mendocino College students are female. Let 𝜶 =𝟎. 𝟏𝟎. Our sample is the Class Data Base, in which 60 of the 102 people in the sample are female
Notice that the calculator found 𝑝 for us. Here’s the diagram:
C. 0.037 < 0.10, or 𝑝 < 𝛼, so reject𝐻0
D. Since we did reject 𝐻0, and since the claim was 𝐻1, we’re in the lower-left rectangle
There is sufficient evidence to support the claim that the majority of Mendocino College students are female
ACTIVITY #17: TESTING CLAIMS ABOUT THE PROPORTION
STEPS IN TESTING CLAIMSA. State the hypotheses and identify the claim.B. State the values of x and n.
Perform the test using the sample data or statistics, and make a diagram showing the distribution, the parameter and the statistic.
Shade and label the p-value, which should be rounded to the nearest thousandth. C. Decide whether or not to reject the null hypothesis.D. Summarize the results.
Use the 5% significance level in testing these claims.
1. It has been claimed that a minority of Americans are in favor of national health insurance. In a survey in which 400 people were interviewed, it was found that 47% were in favor of national health insurance.
2. A researcher studying demographics thinks that the majority of Mendocino College students are natives of Lake or Mendocino County. Test this claim.
3. Two-thirds of Mendocino College students own at least one pet.