AP Statistics – Chapter 17-20 Test Review Name: ________________________________ Period: ____

On a separate sheet of paper, answer all of the questions below. Be thorough in your responses, as this is what
will be required for the Free Response and Short Answer portions of the Test.

1. Several post-operative treatments have been used to help patients recover from bypass surgery. The most popular
treatment has been used for the past 3 years. In all of the cases in the past 3 years when the treatment was used, 72%
of the patients showed significant health improvement. A new treatment is being studied in which the sample of 120
patients who had suffered heart attacks was given a new treatment and 95 showed significant health improvement.

a. Does the study give evidence that the treatment is better than the old one at a significance level of 5%?
[ NOTE: Be sure to run a complete inference test, including defining all variables and checking conditions, with
proper interpretations and conclusions. Follow your detailed Steps for Inference Tests. And finally, be sure to
answer the question above.) ]

VOI = % health improvement p = 0.72, phat = 95/120 = 0.7917 Ho: p = 0.72 Ha: p > 0.72

Conditions: We will assume the sample was randomly selected and independently done.
10x check: We will assume there are at least 1200 patients (safe assumption!)
np checks: np = 95, n(1-p) = 120 – 95 = 25, both > 10… All Conditions are Met! 😊😊

phat − p 0.7917 − 0.72

Z = = 1.749
p (1 − p ) 0.72(0.28)
n 120
P( Z >1.749) = 0.0401

Since the p-value (0.0401) < alpha (5%), we Reject the Ho claim.
Based on our sample, we have reason to believe the proportion of patients whose health improved is greater than 72%
(i.e., believe Ha).

b. If the significance level is changed to 1%, what impacts, if any, are made to the decision in the previous part?
If your decision changed, what does that tell you about your concluding statements in both scenarios?

If alpha (significance level) changed to 1%, our decision would change…

Since the p-value (0.0401) > alpha (1%), we Fail to Reject the Ho claim. And based on this new information, we have
reason to believe the proportion is actually 72% (and not greater).

c. For the same scenario, create the corresponding confidence interval and describe if this reconfirms your decision
in the previous part. Explain why or why not.
[ Hint: 90% CI means 10% is outside the interval, 5% on each tail… 2 x 5% significance level ]
[ Hint: What value would we look for in the Confidence Interval range, to confirm our conclusion in part (a)? ]

A 5% significance level corresponds to a 90% confidence interval (i.e., 5% at both the lower/upper tails).
p (1 − phat ) 0.7917(0.2083)
90%CI = phat ± Z * hat 0.7917 ± 1.645
=
For a 90% CI, the Z* value is 1.645. n 120
90%CI = (0.731, 0.852)

We are checking to see if 0.72 is found within the CI range… It is NOT, and is just slightly below the lower bound, so
this is consistent with our decision in part (a)… Reject Ho claim (because 0.72 is NOT within the range).
 d. For the same scenario, create a 98% confidence interval. How does this range compare to the decision found in
part (b) and the range found in part (c)?

0.7917(0.2083)
For a 98% CI, the Z* value is 2.326. 98%CI =
0.7917 ± 2.326 (0.705, 0.878)
=
120
You will notice this range of (0.705, 0.878) is wider than the range found in part (c)… Checking the range, 0.72 is found
WITHIN the range. This is also consistent with the Fail to Reject Ho claim decision in part (b).

e. Describe a Type I and Type II error associated with this scenario, and describe the impacts of those errors? Be
specific and respond in context to the scenario.

Type I Error: We believe Ha, but Ho is actually true. We believe the % of patients who showed health improvement is
more than 72%, but it actually only 72%. Impact: We think the new treatment is helping, but it may not actually be. We
might put more money into treatment program, but possibly shouldn’t have.
Type II Error: We believe Ho, but Ha is actually true. We believe the % of patients who showed health improvement is
72%, but it actually greater than 72%. Impact: Thinking the new treatment is not working, we do not put money into the
program, but possibly we actually should have.

2. A building inspector believes that the percentage of new construction with serious code violations may be even
greater than the previously claimed 7%.

a. Define your Variable of Interest. State the Null and Alternative Hypothesis statements.

VOI: % of homes with code violations. Ho: p = 0.07 Ha: p > 0.07

b. Describe a Type I and Type II error associated with this scenario, and describe the impacts of those errors.

Type I error: we believe % of homes with code violations is greater than 7%, but it actually 7%. Impact: We put in
unnecessary procedures and programs to improve quality.
Type II error: we believe % of homes with code violations is 7%, but it actually greater than 7%. Impact: We did not
establish any procedures or programs to improve quality, but should have. Homes are being built with poor quality.

c. She conducts a hypothesis test on 200 new homes and finds 23 with serious code violations. Is this strong
evidence against the 0.07 claim? Explain your reasoning.
(Note: You do not need to check conditions, but you should identify all key variables; i.e., p, phat, x, n)

phat − p 0.115 − 0.07

=Z = = 2.49
p (1 − p ) 0.07(0.93)
n = 200, x = 23, phat = 23/200 = 0.115
n 200
P( Z > 2.49) = 0.006
Since the p-value < 5%, we Reject the Ho claim. Based on the sample, we have strong evidence that the proportion of
homes with code violations is greater than 7% (i.e., believe Ha).
 3. You want to design a study to estimate the proportion of students on your campus who agree with the statement, "The
student government is an effective organization for expressing the needs of students to the administration." You will
use a 95% confidence interval and would like the margin of error to be 0.05 or less. Find the minimum sample size
required for this margin of error and confidence interval level.

We are given the Margin of Error (5%), and the 95% CI yields a Z* value of 1.96… Use these values to solve for n.
The most conservative sample size (i.e., the largest value) is found when we use p = 0.50… Rearrange the equation to
solve for n. Then, round the value up to the next integer for the desired sample size.

* p (1 − p )
ME Z=
= SE Z *
n
0.50(0.50)
0.05 = 1.96 ...
n
=n 384.15
= 385

4. In looking at the Confidence Interval formula, what are ways we could ensure a larger Confidence Interval range
value? (i.e., a larger or smaller … )

phat (1 − phat )
CI phat ± Z *
%= Ways to have a Larger CI:
n
Larger Z* (i.e., larger confidence interval % value), a Smaller n (sample size), phat values closer to 0.50 (50%)

5. A recent election poll indicated that in a sample of 250 voters, 130 indicated that they intended to vote for a particular
candidate. The pollsters were quoted as saying, “it’s too close to call.” Are they correct?
[ Hint: Consider creating a confidence interval, or using multiple confidence interval levels, to justify your answer. ]

phat = 130/250 = 0.52 For 90% CI, Z* = 1.645

p (1 − phat ) 0.52(0.48)
90%CI = phat ± Z * hat 0.52 ± 1.645
= (0.468, 0.572)
=
n 250
95%CI = (0.458, 0.582)
80%CI = (0.478, 0.560)

ALL 3 CI ranges include 0.50… so it is safe to assume that most confidence intervals include 50%. Since 0.50 is IN each
of the CI ranges, it is too close to call for the election.

6. For the scenario in question #5, define the following variables:

a. Variable of Interest % vote for a candidate

b. x, n, Phat x = 130, n = 250, phat = 130/250 = 0.52
phat (1 − phat ) 0.52(0.48)
c. Standard Error =
(SE) SE = = 0.0316
n 250
d. Z* value, if a 90% confidence interval is chosen Z* (Area 0.95) = 1.645
e. Margin of Error (ME), for the same 90% CI ME = Z*SE = 1.645(0.0316) = 0.0520
7. A sample of 50 is drawn from a factory that makes Thingamabobs. Prior experience has shown that the defective rate
for Thingamabobs is 10%. What is the likelihood the defective rate of your sample is less than 12%?
[ Hint: your Standard Error is derived entirely from the sample proportion. ]

phat − p 0.12 − 0.10

Z = = 0.471
p (1 − p ) 0.10(0.90)
n 50
P( Z < 0.471) = 0.681
For the sample of 50 Thingamabobs, there is a 68.1% chance we would see a defective rate less than 12%.
(Note: This is a significantly higher % because we are dividing by the sample standard error.)

8. A sample of 30 items is drawn from a factory that manufactures widgets. Prior experience has shown that the weight
for these widgets are normally distributed, with an average of 25 oz and standard deviation of 4 oz.

a. On average, what would we anticipate the mean weight of the sample to be? Why?
Sample mean takes on the population mean, especially for sample sizes at or above 30… so Sample mean ~ 25 oz
b. What is the standard deviation for the sample?
Sample standard deviation takes into account the sample size (by dividing by sqrt(n) ).
σ 4
σ X= = = 0.730
bar
n 30

c. What is the likelihood the weight of one widget (sample) is less than 24 ounces?

X −µ
24 − 25
Z= = = −1.369
σ 4
n 30
P( Z < −0.012) =0.085

9. You found the Null & Alternative Hypothesis statements in each of these scenarios previously. State what a Type I
and Type II error would be and list one possible impact of each.

a. Mrs. Smith’s Apples claims that the proportion of apples with a blemish is 12%. The local grocers is
concerned with an unusually high number of bruised apples. The local grocer took a sample of 49 apples and
found 14% had a blemish.
VOI = % of apple blemishes Ho: p = 0.12, Ha: p > 0.12
Type I error: We believe % of blemishes is > 12%, but it is actually 12%. Impact: Pull good apples off of the shelf.
Type II error: We believe % of blemishes is 12%, but it is actually > 12%. Impact: Sell bad apples to customers.

b. A large city’s Department of Motor Vehicles claimed that 80% of candidates pass driving tests, but a local
newspaper reporter is highly doubtful that the rate is this high. The reporter took a survey of 90 randomly
selected local teens and found 68 who passed.
VOI = % pass driving test Ho: p = 0.80, Ha: p < 0.80
Type I error: We believe % passing test is < 80%, but it is actually 80% (or less). Impact: Recommend possibly
unnecessary DMV programs to improve passing rates.
Type II error: We believe % passing test is 80%, but it is actually < 80%. Impact: Should have possibly implemented
DMV program improvements, but didn’t.
 c. A certain part in a factory historically had a failure rate of 10%. The process used to make this part was
altered so the factory managers are curious is the failure rate has improved. The factory managers took a
random sample of 50 parts and found 1 that were defective.
VOI = % of failure rate Ho: p = 0.10, Ha: p < 0.10
Type I error: We believe % failure rate < 10%, but it is actually 10%. Impact: Left factory processes alone, thinking
quality had improved… but maybe should have implemented processes.
Type II error: We believe % failure rate = 10%, but it is actually < 10%. Impact: Implemented some factory processes, but
possibly shouldn’t have.

10. In your own words, define the following terms:

a. phat b. p c. p-value d. statistically significant

phat – sample proportion… statistic found when a particular sample is taken

p – population proportion… generally accepted proportion value
p-value – the likelihood of the observed sample occurring (or a sample as extreme as this, purely by chance)
Statistically significant – a sample that is significant enough to occur not likely by chance… i.e., a very low p-value

11. For the following scenarios, identify the following: Population, Parameter, Sample, Statistic
a. Each month, the Current Population Survey interviews a random sample of individuals in about 55,000 U.S.
households. One of the goals is to estimate the national unemployment rate. In December 2009, 10.0% of
those interviewed were unemployed.

Population – US households Parameter – unemployment rate % (not listed)

Sample – 55,000 households Statistic – 10%

b. A factory produces plate glass with a mean thickness of 4 mm and a standard deviation of 1.1 mm. What is
the mean and the standard deviation for the sampling distribution in this situation? Find the probability that
in a SRS of 100 sheets of glass, the mean thickness of the 100 sheets is more than 3.85.

Population – Factory plate glass Parameter – 4mm (mean thickness)

Sample – 100 glass sheets Statistic – 3.85mm

12. A botanist has produced a new variety of hybrid wheat that is better able to withstand drought than other varieties.
The botanist knows that for the parent plants, the seed germination was 80%. To test this claim, 400 seeds from the
hybrid plant are tested, and it is found that 334 germinated. Interpret a p-value of 0.040059 and make a conclusion
based on α = .05. [ NOTE: You do not to conduct an Inference Test, only interpret the p-value referenced. ]

The p-value of 0.040059 is less than the significance level (alpha, 5%), so our decision would be to Reject the Ho claim.
Ho: p = 0.80 (% seed germination) Ha: p > 0.80 (because our phat = 334/400 = 0.835)
BONUS Questions…

Some conceptual questions for you…

13. Circle the correct word or phrase to complete the sentence. You might consider the reason for your
Increase/Decrease response.

a. As 𝑝𝑝̂ gets closer to 0.5, minimum sample size to achieve a set margin of error { increases or decreases }
phat = 0.5 will yield the highest value on the numerator… so a higher n is needed to maintain the same ME
b. As sample size (n) increases, standard deviation of sampling distribution { increases or decreases }
Standard Error (SE) is inversely related to sample size (n)
c. As confidence interval % increases, width of a confidence interval { increases or decreases }
Larger CI % values yield larger confidence interval ranges.
d. As margin of error decreases, width of a confidence interval { increases or decreases }
same as ( c )
e. As confidence interval % increases, the Z* value { increases or decreases }
CI % values are directly related to the Z* values
f. As alpha increases, the likelihood of deciding to ‘Reject the Ho claim’ { increases or decreases }
As alpha (significance level) increases (e.g., 5% to 10%), there is a greater chance of Rejecting the Ho claim.
g. All other things being equal, the sample size (n) increases, the margin of error { increases or decreases }
Sample size (n) is inversely related to the ME
h. All other things being equal, the sample size (n) decreases, the standard error { increases or decreases }
Sample size (n) and SE are directly related