$36WDWV&KDSWHUReview 4.

The following are percents of fat found in 5 samples of each of two brands of ice cream:
Part 1: Multiple Choice. Circle the letter corresponding to the best answer. A 5.7 4.5 6.2 6.3 7.3
B 6.3 5.7 5.9 6.4 5.1
Use the following for questions 1 – 3: Which of the following procedures is appropriate to test the hypothesis of equal average fat
content in the two types of ice cream?
Sixty-eight people from a random sample of 128 residents of Uppsala, Sweden, had blue eyes. (a) Paired t test with 5 df.
45 people from a random sample of 110 people from Preston, England, had blue eyes. Let p1 (b) Two-sample t test with 4 df.
(c) Paired t test with 4 df.
represent the proportion of people in Uppsala with blue eyes, and let p2 represent the proportion
(d) Two-sample t test with 9 df.
of people in Preston with blue eyes. (e) Two-proportion z test

1. If researchers suspected that the distribution of eye color is different in these two countries 5. An ecologist studying differences in populations of a certain species of lizards on two
before collecting the data, which of the following pairs of hypotheses would be appropriate different islands collects lizards in live traps, weighs them, and then releases them again. (He
to test? marks them so he won’t weigh the same lizard twice). During one study period, he collects
(a) H 0 : p1 0.53, p2 0.41; H a : p1 z 0.53, p2 z 0.41 the following data. All weights are in grams.
(b) H 0 : p1 p2 0.47; H a : p1 z p2 z 0.47 n Mean (gm) Std. Dev. (gm)
(c) H 0 : p1 p2 ; H a : p1 ! p2 Sheep Island 24 46.5 5.97
(d) H 0 : p1 p2 ; H a : p1 z p2 Pig Island 30 44.2 4.24
(e) H 0 : p1 p2 ; H a : p1  p2 Which of the following is the correct expression for the test statistic to test the hypothesis
that the mean weights on the two islands are equal?
2. Which of the following conditions are necessary in order to perform the test in Question 1? 46.5  44.2
(a) t 2
I. There must be at least 1280 people in Uppsala, Sweden and at least 1100 people § 5.97 4.24 ·
in Preston, England. ¨  ¸
© 24 30 ¹
II. np and n(1 – p) must be large enough for Normal calculations to be reasonably
accurate. 46.5  44.2
(b) t
III. Two independent random samples must be taken. 5.97 4.24

(a) I only 24 30
(b) II only 46.5  44.2
(c) III only (c) t
5.97 2 4.242
(d) I and III are necessary 
(e) I, II, and III are all necessary 24 30
46.5  44.2
(d) t
3. The P-value for this test is 0.06. If the researchers chose a significance level of 0.05, which 5.97 2 4.242
of the following represents the correct conclusion to draw from this result? 
24 30
(a) Reject H 0 : there is evidence to suggest a difference in the proportion of blue-eyed people
46.5  44.2
in these two countries. (e) t
5.97 2  4.242
(b) Fail to reject H 0 : there is convincing evidence to suggest a difference in the proportion of
54
blue-eyed people in these two countries.
(c) Fail to reject H 0 : there is insufficient evidence to suggest a difference in the proportion of
blue-eyed people in these two countries.
(d) Accept Ha: there is convincing evidence to suggest a difference in the proportion of blue-
eyed people in these two countries.
(e) Accept H0: there is insufficient evidence to suggest a difference in the proportion of blue-
eyed people in these two countries.
Use the following for questions 6 – 8: Part 2: Free Response
Show all your work. Indicate clearly the methods you use, because you will be graded on the
An experiment was conducted to assess the efficacy of spraying oats with malathion (at 0.25 correctness of your methods as well as on the accuracy and completeness of your results and
lb/acre) to control the cereal leaf beetle. A sample of 10 farms was selected at random from explanations.
southwest Manitoba. Each farm was assigned at random to either the control group (no spray) or
the treatment group (spray). At the conclusion of the experiment, a plot on each farm was 9. Daphnia pulicaria is a water flea—a small crustacean that lives in lakes and is a major food
selected, the number of larvae per stem was measured, and a one-tailed test of significance was supply for many species of fish. When fish are present in the lake water, they release
performed to determine if malathion reduced the number of beetles. Here are two possible chemicals called kairomones that induce water fleas to grow long tail spines that make them
outputs from MINITAB, only one of which is correct (some output hidden): more difficult for the fish to eat. One study of this phenomenon compared the relative length
Two-sample T for not spray vs spray of tail spines in Daphnia pulicaria when kairomones were present to when they were not.
Below are data on the relative tail spine lengths, measured as a percentage of the entire
length of the water flea.
N Mean StDev SE Mean
Not spray 5 4.09 **** ****
Spray 5 3.05 **** **** Relative tail spine length
n x s
Difference = mu (not spray) - mu (spray) Fish kairomone present 214 37.26 4.68
Estimate for difference: 1.04
T-Test of difference = 0 (vs > 0): T-Value = 1.896 P-Value = ***** DF = *****
Fish kairomone absent 152 30.67 4.19

Paired T for not spray - spray (a) Do the data provide convincing evidence that the mean relative tail spine length of
Daphnia is longer in the presence of fish kairomones? Assume the conditions for
N Mean StDev SE Mean
Not spray 5 4.09 **** **** inference have been met.
Spray 5 3.05 **** ****
Difference 5 1.04 **** ****

T-Test of mean difference = 0 (vs > 0): T-Value = 1.887 P-Value = *****

6. Which of the following is the appropriate test statistic and a possible P-value?
(a) 1.896, 0.013
(b) 1.896, 0.065
(c) 1.896, 0.131
(d) 1.887, 0.059
(e) 1.887, 0.118

7. In which one of the following cases would a Type II error occur?

(a) We do not conclude malathion is effective when in fact it was effective.
(b) We conclude malathion is effective when in fact it is ineffective.
(c) We conclude malathion is effective when in fact it is effective.
(d) We do not conclude malathion is effective when in fact it is ineffective.
(e) We conclude malathion is neither ineffective nor effective.

8. What does power refer to in this situation?

(a) The ability to detect an effect of malathion when in fact there is no effect. (b) What additional information would you need to confirm that the conditions for this test
(b) The ability to not detect an effect of malathion when in fact there is no effect. have been met?
(c) The ability to detect an effect of malathion when in fact there is an effect.
(d) The ability to not detect an effect of malathion when in fact there is an effect.
(e) The ability to generalize the results of this controlled experiment to other insect pests
besides the cereal leaf beetle.
10. An association of Christmas tree growers in Indiana sponsored a sample survey of 500
randomly selected Indiana households to help improve the marketing of Christmas trees.
One question the researchers asked was, “Did you have a Christmas tree this year?”
Respondents who had a tree during the holiday season were asked whether the tree was
natural or artificial. Respondents were also asked if they lived in an urban area or in a rural
area. The tree growers want to know if there is a difference in preference for natural trees
versus artificial trees between urban and rural households. Among the 160 who lived in rural
areas, 64 had a natural tree. Among the 261 who lived in an urban area, 89 had a natural tree.
(a) Construct and interpret a 95% confidence interval for the difference in the proportion of
rural and urban Indiana residents who had a natural Christmas tree this year.
$QVZHUV6ROXWLRQV
Part 1
1. d The null (“no effect”) hypothesis is that the proportion of blue eyes in each population is
the same. Prior to the study the researchers did not express an expectation that one
group’s proportion would be higher than the other’s, so the alternative hypothesis is
p1 z p2 .
2. e Statements I and III both refer to the independence condition (statement I for finite
populations, in particular). Statement II is the Normality condition for two-proportion z-
procedures.
3. c Since the P-value (0.06) is larger than D (0.05), we have no evidence against the null, so
we cannot reject it. A test of significance only provides evidence against the null
hypothesis—it does not provide evidence for the alternative if the P-value is small or for
the null if the P-value is large.
4. b These are samples from two independent populations with unknown D, so a two-sample
t-test is appropriate. The conservation choice for degrees of freedom is 1 less than the
smaller sample size, or 4.
5. c Formula for two-sample t-statistic for the difference of two means is t
 x1  x2   0 .
s12 s22

n1 n2
6. b The appropriate test is the two-sample test, not the paired test, so t = 1.896. According to
Table B, this test statistic with df = 4 yields a P-value between 0.05 and 0.10, and the
only choice in that range is 0.065.
7. a Type II error is failing to reject H0 when it’s false, so in this context, it’s concluding that
malathion has no effect on mean beetle density when it does.
8. c Power is 1  P  Type II error  , so it’s the probability of rejecting H0 when there is a
difference in beetle density between the two groups.
Part 2
9. (a) State: We are testing the hypotheses H 0 : P1  P 2 0; H a : P1  P 2 ! 0 , where P1 is the mean
length of Daphnia tail spines when fish kairomones are present and P 2 is the mean length of
Daphnia tail spines when fish kairomones are absent. We will use a significance level of D =
0.05. Plan: The procedure is a two-sample t-test for the difference of means. The problem
states that the conditions for this procedure have been met. Do:
(b) Suppose the tree growers asked 10 different questions about Christmas tree preferences in an
effort to find differences between urban and rural residents. Explain why it would be unwise
to perform 10 different 2-sample significance tests (D = 0.05) on their data and conclude that
any significant result establishes a difference between the two groups.
 37.26  30.67   0 6.59
t 14.111 . This very large t-statistic yields an extremely small P-value
4.68
2
4.19
2
0.467

214 152
(by calculator: P-value | 2 u 10 36 ). Conclude: Since the P-value is much less than any
conventional value for D, we reject H0: we have strong evidence that the mean length of Daphnia
pulicaria tail spines in the presence of fish kairomones is longer than when the kairomones are
absent. (b) We need to know more about the design of the experiment. If this was a controlled
experiment, were individual water fleas randomly assigned to the two treatments? If water fleas
were merely sampled from environments that did or did not have fish kairomones, can we view
the samples as independent SRSs from the two populations? The sample sizes are very large, so
we are not concerned about the shapes of the population distributions. 10. State: We wish to
estimate, with 95% confidence, the difference pR  pU , where pR and pU are, respectively, the
proportion of rural and urban residents who had a natural Christmas tree this year. Plan: We
should use a 2-sample z-interval for pR  pU . Conditions: Random: The study used 500
randomly-selected Indiana households. We can view this as SRSs from rural and urban
populations. Normal: Number of successes and failures in the two groups are 64, 96, 89, and
172, all of which are at least 10. Independent: It seem reasonable to assume that the samples are
independent, and there are certainly more than 5000 households in Indiana. Do: The critical z
64 89
for 95% confidence is 1.96 and pˆ R 0.4; pˆ U 0.34 . The interval is given by
160 261
§  0.4  0.6   0.34  0.66  ·
 0.4  0.34  r 1.96 ¨  ¸ 0.06 r 0.095 , or  0.035, 0.155  . Conclude: We are
© 160 261 ¹
95% confident that the interval from –0.035 to 0.155 captures the true difference in the
proportion of rural versus urban residents of Indiana who had a natural Christmas tree this year.
(b) At the D = 0.05 level, we expect to make a Type I error and reject a true H0 about once in
every 20 tests we perform. If we perform 10 such tests, the chances are quite good that we will
make at least one Type I error. (In fact, the probability is 1  0.9510 | 0.4 ).