lOMoARcPSD|9327020

Lab Activity #12: More Practice with Hypothesis Testing (KEY)

You will definitely see questions like this on the final exam, and today is the day to get some help and to
ask questions! We hope this assignment will also help you tackle the next homework assignment.

1. In a 2015 survey, Gallup reported that Americans planned to spend an average of $830 on gifts during
the holiday season. You believe that Americans plan to spend less than $830. You survey a random
sample of 45 individuals and find they report planning to spend an average of $815 on holiday gifts,
with a sample standard deviation of $50.

Use the information above to conduct the appropriate hypothesis test by going through the steps we
discussed in lecture. Assume the alpha level is .01 (or 1%). Fill in the following table with your
answers, and be sure to show all your work.

Hypotheses Ho: µ = $830

Ha: µ < $830

Test Statistic 𝑥̅ − 𝜇 815 − 830 −15

𝑧= 𝑠 = = = −2
50 7.5
√𝑛 √45

p-value From Table B, we see the percentile corresponding to a z-value of -2 is 2.27%. Our p-
value is therefore 2.27%/100% = .0227.

What is your Since .0227 is greater than (or larger than) .01, our decision is to fail to reject the null
decision, and hypothesis. With this sample of data, we do not have evidence to suggest the average
why? amount spent on gifts is not $830.

2. If the sample size in Question 1 had been 75 rather than 45, how do you think this would affect the p-
value? Please explain.

With a larger sample size, we would obtain a larger test statistic. This in turn would lead to a
smaller p-value.

3. If the alpha level had been .05 (or 5%) rather than .01 (or 1%), would your decision have changed?

Yes. Since .0227 is less than (or smaller than) .05, our decision would have been to reject the
null hypothesis.

4. Suppose that one of your classmates wrote down the following hypotheses for this example from
Question 1. What is wrong with these hypotheses?

Ho: 𝑥̅ = $815
Ha: 𝑥̅ > $815

We write our hypotheses in terms of population parameters, not sample statistics. The
hypotheses should be:
 lOMoARcPSD|9327020

Ho: µ = $830
Ha: µ < $830

5. If a researcher obtained a p-value of 0.08 for a hypothesis test that has a significance level (or alpha
level) of 0.05, what conclusion should the researcher make?

a. The null hypothesis is true.

b. The null hypothesis should not be rejected.
c. The null hypothesis should be rejected.
d. None of the above.
6. If a researcher obtained a p-value of 0.08 for a one-sided (or one-tailed) alternative hypothesis, what
would the p-value be if the alternative hypothesis was two-sided (or two-tailed)? Assume here that
all that would change would be the direction of the alternative hypothesis; the sample data would be
the same.
a. 0.08
b. 0.04
c. 0.16
d. 0.28
e. It’s impossible to answer this question with the given information

7. A hypothesis test is conducted in which the alternative hypothesis is that more than 10% of a
population is left-handed. The p-value for the test is calculated to be 0.25. Which statement is correct?

a. We can conclude that more than 10% of the population is left-handed.

b. We can conclude that more than 25% of the population is left-handed.
c. We can conclude that exactly 25% of the population is left-handed.
d. We cannot conclude that more than 10% of the population is left-handed.
8. A test for ESP involves having someone repeatedly try to guess which one of four photographs is the
chosen “target,” with a different set of photographs used for each trial. Define p to be the probability
of a correct guess. Suppose someone guesses 100 times, and gets 30 correct, so 𝑝̂ = 0.30. What is
the appropriate null hypothesis in this situation?

a. H0: p = .25
b. H0: p = .30
c. H0: 𝑝̂ = .25
d. H0: 𝑝̂ = .30
9. In a newspaper article about whether the regular use of Vitamin C reduces the risk of getting a cold, a
researcher is quoted as saying that Vitamin C performed better than placebo in an experiment, but the
difference was not larger than what could be explained by chance. In statistical terms, the researcher
is saying the results are

a. due to non-sampling errors.

b. definitely due to chance.
c. statistically significant.
d. not statistically significant.

2
 lOMoARcPSD|9327020

Each of the following three statements is FALSE. Please explain why each statement is false.

10. The p-value of a hypothesis test is the probability of obtaining a result as extreme (or more extreme)
as the one obtained assuming the null hypothesis is false.

The p-value of a hypothesis test is the probability of obtaining a result as extreme (or more
extreme) as the one obtained assuming the null hypothesis is TRUE.

11. The larger the p-value, the more evidence there is against the null hypothesis.

The larger the p-value, the more evidence there is IN FAVOR of the null hypothesis. A smaller
p-value gives us evidence against the null hypothesis.

12. If the p-value for a test is 0.15, the probability that the null hypothesis is true is .15.

A p-value tells us nothing about the probability of the null hypothesis being true (or false). It tells
us the probability of obtaining our sample results IF the null is true.

Read carefully through the following (for Questions 13-15), and the problems that accompany each
question. Each problem has a mistake in it, and we would like you to (a) identify what the mistake is, and
(b) correct the mistake.

13. In the past, about 10% of people were left-handed. You suspect the percentage is increasing because
people are less likely to dissuade a child from being left-handed these days. You observe a random
sample of 100 first-graders at recess to see whether they throw with their left or right hand, and you
find that 15 use their left hand. What is the probability that you would observe 15 or more first-
graders using their left hand if the population proportion was still 10%?

Problem: A student attempts to work through this problem by calculating the following
test statistic. Since the value of z is not in Table B, the student concludes the p-value
must be very small (and clearly less than .05).

15−.10
𝑧= = 497
.10(1−.10)
√
100

The calculation here is incorrect. The numerator should be “.15 - .10” rather than “15 -
.10.” if the correct value was plugged in for the sample proportion, the test statistic
would end up being 0.05/0.03 = 1.7.

14. Researchers at Stanford studied whether reducing children’s television viewing might help to prevent
obesity. At the beginning of the study, children were asked to report how many hours of television
they watch in a typical week. The 198 responses had a mean of 15.41 hours and a standard deviation
of 14.16 hours. Do these data provide evidence at the .05 significance level for concluding that third
and fourth graders watch more than 14 hours of television per week on average?

Problem: A student approaches this problem by writing the following hypotheses.

Ho = 14
Ha > 14

3
 lOMoARcPSD|9327020

The null and alternative hypotheses are missing the symbol for the correct
population parameter. They should be:

Ho: = 14
Ha:  > 14

15. Suppose that a golfer keeps track of how far he hits the ball when he practices his long shots, and he
has found over time that his distances have an average of 180 yards and a standard deviation of 20
yards. Suppose that he starts using a new brand of golf ball and wants to see whether his average
distance seems to improve. He hits 64 shots with the new ball and calculates his average distance to
be 190 yards. Does this provide strong evidence that the mean of his driving distances is greater than
180 with the new ball?

Problem: A student answers this question by saying “Yes, because 190 is larger than 180.”

This response ignores sampling variability completely. It’s not good enough to compare 190 to 180,
because we need to consider the fact that the sample mean distance will vary from sample to
sample. The key issue is how much variability there is in the sample mean distances (how far they
tend to fall from the population mean from sampling variability alone), and whether a sample mean
distance of 190 yards is surprising. To figure this out, we must standardize that sample mean distance
and use Table B to determine the probability of obtaining that sample mean distance or a more
extreme sample mean distance if the true average is 180 yards.

16. (If time permits: More practice with hypothesis testing for proportions) The makers of M&Ms claim
that a bag of candies should contain 24% blue (or 0.24 blue). You believe that a larger proportion of
blue M&Ms exist in bags of candies. You obtain a random sample of 100 candies and find that 28%
(or 0.28) of the sample is made up of blue M&Ms. Conduct a hypothesis test with α= .05 (or 5%).
Hypotheses Ho: p = 0.24

Ha: p > 0.24

Test Statistic The standard deviation of our sampling distribution is

𝑝̂ − 𝑝
𝑧=
√𝑝(1 − 𝑝)
𝑛

The test statistic is z = (0.28-0.24)/0.04 = 1

p-value From Table B, we find the percentile below 1 to be 84.13%. The percentile above would be 100% -
84.13% = 15.87%, so our p-value would be .1587.

What is your Here, we fail to reject the null hypothesis since the p-value of .1587 is larger than our alpha level of
decision, and .05. We do not have evidence with this sample that the proportion of blue M&Ms is significantly
why? larger than 0.24.