Problem 1

Acme Corporation manufactures light bulbs. The CEO claims that an average Acme light bulb
lasts 300 days. A researcher randomly selects 15 bulbs for testing. The sampled bulbs last an
average of 290 days, with a standard deviation of 50 days. If the CEO's claim were true, what is
the probability that 15 randomly selected bulbs would have an average life of no more than 290
days?

Note: There are two ways to solve this problem, using the T Distribution Calculator. Both
approaches are presented below. Solution A is the traditional approach. It requires you to
compute the t score, based on data presented in the problem description. Then, you use the T
Distribution Calculator to find the probability. Solution B is easier. You simply enter the
problem data into the T Distribution Calculator. The calculator computes a t score "behind the
scenes", and displays the probability. Both approaches come up with exactly the same answer.

Solution A

The first thing we need to do is compute the t score, based on the following equation:

t = [ x - μ ] / [ s / sqrt( n ) ] 
t = ( 290 - 300 ) / [ 50 / sqrt( 15) ] = -10 / 12.909945 = - 0.7745966

where x is the sample mean, μ is the population mean, s is the standard deviation of the
sample, and n is the sample size.

Now, we are ready to use the T Distribution Calculator. Since we know the t score, we select "T
score" from the Random Variable dropdown box. Then, we enter the following data:

 The degrees of freedom are equal to 15 - 1 = 14.

 The t score is equal to - 0.7745966.

The calculator displays the cumulative probability: 0.226. Hence, if the true bulb life were 300
days, there is a 22.6% chance that the average bulb life for 15 randomly selected bulbs would
be less than or equal to 290 days.

Solution B:
This time, we will work directly with the raw data from the problem. We will not compute the
t score; the T Distribution Calculator will do that work for us. Since we will work with the raw
data, we select "Sample mean" from the Random Variable dropdown box. Then, we enter the
following data:

 The degrees of freedom are equal to 15 - 1 = 14.

 Assuming the CEO's claim is true, the population mean equals 300.
 The sample mean equals 290.
 The standard deviation of the sample is 50.

The calculator displays the cumulative probability: 0.226. Hence, there is a 22.6% chance that
the average sampled light bulb will burn out within 290 days.

Problem 2

Suppose scores on an IQ test are normally distributed, with a mean of 100. Suppose 20 people
are randomly selected and tested. The standard deviation in the sample group is 15. What is
the probability that the average test score in the sample group will be at most 110?

Solution:

To solve this problem, we will work directly with the raw data from the problem. We will not
compute the t score; the T Distribution Calculator will do that work for us. Since we will work
with the raw data, we select "Sample mean" from the Random Variable dropdown box. Then,
we enter the following data:

 The degrees of freedom are equal to 20 - 1 = 19.

 The population mean equals 100.
 The sample mean equals 110.
 The standard deviation of the sample is 15.

We enter these values into the T Distribution Calculator. The calculator displays the cumulative
probability: 0.996. Hence, there is a 99.6% chance that the sample average will be no greater
than 110.
Exercise Problem A:

A university administrator is interested in whether students returning to college after a break

have a higher GPA than students who go straight through college after high school. She knows
that the mean GPA for the population of student who go straight through college after high
school is m = 2.9, and that the standard deviation of the population is s = 0.394. To determine
whether students returning to college after a break have a higher GPA than students who go
straight through college after high school, the administrator selects a sample of n = 10 students
who returned to college after taking a break. The mean GPA for these 10 students is |X = 3.11.
Conduct a single sample t-test to determine whether students returning to college after a
break have a higher GPA than students who go straight through college after high school. [Use a
= .05.] Be sure to report the value of t that you get for this sample mean, and what the critical
value of t is for this problem.

 The research hypothesis in this case is a one-tailed because we want to know whether
the population mean for GPA for returning students is GREATER THAN the mean GPA for
the population of student who go straight through college from high school. So the
research hypothesis is H1: mstraight < mbreak. We start by assuming that the null
hypothesis is true. The null hypothesis is: H0: mstraight =mbreak.
 From this, we can assume that the population that the sample of returning students
was drawn from has a population mean of 2.9 (since we assumed H0 is true).
 From the population of GPAs for student who did not take a break in school, we
construct a distribution of sample means using a sample size of n = 10. We use a
sample size of n = 10 because that's the sample size of your sample. So the sample
mean of your sample of size n = 10 is one of the scores in this distribution of sample
means (n = 10).
 Now we need to determine the mean and the standard deviation of this distribution of
sample means so we can locate the sample mean you have with respect to the
distribution of sample means. The mean of the distribution of sample means is easy --
it's the same as the mean of the distribution of raw scores that the distribution of
sample means was constructed from, i.e., 2.9. The standard deviation of the
distribution of sample means -- called the standard error of the mean -- is just the
standard deviation of the distribution of raw scores that the distribution of sample
means was constructed from divided by the square root of n, i.e., 0.394/sqrt(10) =
0.394/3.162 = 0.125
 Now we can compute the t-score for the sample mean we have. Remember t = (score-
mean)/std.dev., where here the score is your sample mean (i.e., 3.11), the mean is
the mean of the distribution of sample means (i.e., 2.9), and the std.dev. is the
standard error of the mean (i.e., 0.125):

t = (3.11-2.9)/0.125 = 0.21/0.125 =1.68

(So the sample mean of 3.11 is 1.68 standard deviations greater than the mean of the
distribution of sample means.)

 (Almost there....) We need to determine what the critical value of t is that bounds our
alpha area in the (right) tail of the t-distribution. We choose alpha to be .05 (which
means that the probability of our making a Type I error if we reject the null hypothesis
is 1 in 20). Anyway, to determine the critical value of t for our one-tailed test, we first
have to determine the degrees of freedom, df. df for a single-sample t-test is df = n-1 =
10-1 = 9. If you look up the critical value of t in Table T for alpha = .05 for a one-tailed
test, you'll find that tcrit = 1.833 for df = 9.
 We fail to reject the null hypothesis because our value of t = 1.68 is NOT further into
the tail of the t-distribution than the critical value of t, which is 1.833.
 
INFERENCE EXAMPLE USING t DISTRIBUTION
(ONE-SAMPLE PROCEDURE)
 In this example, there is no population standard deviation (s) provided. This parameter
is unknown. The statistic s (computed from a sample) is used in place of s.
 =========================
 A machine is calibrated to produce washers with a thickness of 0.05 inches. A random
sample of ten measurements yields the following thicknesses (unit = inches).

.053 .054 .049 .050 .054 .053 .048 .055 .054 .053

 At the 5% level of significance, test the hypothesis that the machinery is functioning
properly.
 Ho: m = 0.05 (sample came from population with m = 0.05).
Ha: m is not equal to 0.05 (sample did not come from population with m =
0.05)
Type of test: 2-tail.
Degrees of freedom: 10 - 1 = 9.
Mean of sample: x(bar) = 0.0523.
Calculated s from sample: s = 0.0024.
 Standard error for sample means: s/SQRT(N) = 0.0024/SQRT(10) = 0.00076.
 Calculation of sample t statistic: t = (0.0523-0.05)/0.00076 = 3.03.
 Critical values of t with 9 degrees of freedom: t < -2.262 and t > 2.262.
 Conclusion: Calculated value of t is in the critical region. At the 5% level of
significance, we reject Ho. At the 5% level of significance, there is evidence to
suggest that the machinery is not functioning properly.
 =========================
 Alternate approach using confidence intervals: Using the data above, the 95%
confidence interval for the population mean is
 0.0523 plus/minus (2.262)(0.00076) = 0.0523 plus minus 0.00172.
 This interval is [0.0506,0.0540]. Since this interval does not contain 0.05, we would
reject Ho at the 5% level of significance.
 ==========================
 Alternate approach using probability: With 9 degrees of freedom, the probability of
obtaining a value t = 3.03 if Ho is true is 0.007, or about 0.7%. [This is
tcdf(3.03,1E99,9) on the TI-83 calculator]. This would cause rejection of Ho at the 5%
level of significance.

Lesson 10 Practice Exercises answers

Confidence Intervals and t-tests for paired data and data from two samples

1. The text works through the steps for a confidence interval and a paired t-test of the mean difference in 7α-
HCO(ng/mL) from ‘before’ to 1 month after cholecystectomy for 51 patients (Sauter study). Data from this study for
7α-HCO(ng/mL) at 3 months after cholecystectomy are also available.

The 3-month minus ‘before’ differences were calculated for each patient. Summarized below are the sample mean and
SD of the difference, the SE(diff) and the critical value for a two tailed test with α = 0.05.

Use the data below to answer the questions

a. The correct t-test to use to test the null hypothesis of no mean change in 7α-HCO from ‘before’ to 3 months after
cholecystectomy is the paired t-test for the mean difference – this is usually just called a ‘paired t-test’
b. Write out the null hypothesis and the alternative hypothesis for this test

Ho δ = 0 (mean difference = 0)

HA

δ ≠0 (mean difference is not equal to 0)

c. The SE(diff) = SD / 51 = 27.91 .7/ 14 = .3 91

d. The critical value is from the t- distribution with n-1 df = 50 df.

e. Calculate the test statistic for this test: 27.27/3.91 = 6.97

f. Is the null hypothesis rejected? Yes Why or why not?

The critical values of this two-tailed test are 2.01 and -2.01. The test statistic = 6.97 which is greater than 2.01. The test
statistic is in the right tail rejection region of the distribution of the test statistic so the null hypothesis of ‘no mean
change’ in 7α-HCO from before to 3 months after cholecystectomy is rejected.

f. If the null hypothesis is rejected, indicate whether there was a mean increase or decrease in 7α-HCO from ‘before’ to
3 months after cholecystectomy.

There was a significant mean increase in 7α-HCO.

We know that there was a mean increase because the mean ‘3-month’ minus ‘before’ difference is positive indicating
that, on average, patients had an increase in 7α-HCO from ‘before’ to 3 months after cholecystectomy. If the difference
had been calculated as ‘before’ minus ‘3-month’, a negative mean difference would indicate an increase.

g. Construct a 95% confidence interval for the population mean difference in 7α-HCO from before to 3 months after
cholecystectomy.

95% CI = mean difference ± coefficient*SE(diff) = 27.27 ± 2.01*3.91 = (19.41, 35.13)

Sample Mean Difference 27.27

SD of differences 27.91

SE(diff) 3.91

critical value for α = 0.05 2.01Lesson 10 Practice Exercises answers

Confidence Intervals and t-tests for paired data and data from two samples

The confidence coefficient for a 95% CI is the same as the critical value for a two-tailed t-test with α = 0.05. The same
SE(diff) is used to calculate the t-statistic and in the confidence interval formula. The 95% confidence interval limits
are consistent with the paired t-test result of rejecting the null hypothesis of ‘no mean change’. The value 0 (indicating
‘no mean change’) is not contained between the interval limits so you would conclude that there was a significant mean
increase from before to 3 months after cholecysectomy

2. The text works through the steps for a 95% confidence interval of the difference in

mean pulse oximetry between Emergency Department patients with suspected pulmonary
embolism (PE) and those without suspected pulmonary embolism (no PE).

Data used for this confidence interval are summarized below

Pulse oximetry PE yes

n=181

PE no

n=740

Mean 93.4 95.8

SD 5.9 4

The pooled SD was calculated for the difference of the two means = 4.44

The confidence coefficient for a 95% CI of the difference of means = 1.96

a. Write out the null hypothesis and the alternative hypothesis for this test

Ho 1− 2 = 0 (means are equal in the two groups)

HA

1− 2 ≠0 (means are not equal in the two groups)

b. Calculate the SE(diff of means) = .0 368

11

12

=+

nn

SDp

c. Calculate the t-statistic for a two sample t-test of the difference of means

assuming equal variance = .6 52

0.368

93 4. 95 8.

−=

d. What are the critical values of a two-tailed test of the null hypothesis that the
difference of means = 0? Use alpha significance level = 0.05.

± 1.96 for a test with α = 0.05

The critical values for a two-tailed test with α = 0.05 are the same as the confidence coefficients for a 95% confidence
interval. Lesson 10 Practice Exercises answers

Confidence Intervals and t-tests for paired data and data from two samples

e. Based on the test statistic and critical values is the null hypothesis rejected? Why or why not?

The null hypothesis of ‘no difference between the means is rejected because the test statistic, -6.52, is in the left tail
rejection region: -6.52 is < -1.96.

f. Are the test results consistent with the conclusion from the 95% confidence interval of the difference in means? The
95% confidence interval constructed in the text is (1.67, 3.13)

Yes – you would conclude from this confidence interval that there is a significant difference between the means
because 0, the value indicating ‘no difference in means’ is not contained between the 95% confidence interval limits.

g. The test statistic has a t-distribution with 181 + 740 – 2 df = 919 df Notice that the critical values on the t-distribution
with large df are the same as the critical values on the standard normal distribution. As the sample size increases, the
tdistribution approaches a standard normal distribution. h. Write out a conclusion and summary statement for the t-test
of the difference in mean pulse oximetry between patients with and without suspected PE. Results of a two-sample t-
test (with alpha = 0.05) of the difference in mean pulse oximetry between Emergency Dept. patients with and without
suspected pulmonary embolism (PE) indicate that mean pulse oximetry is significantly lower for patients with
suspected PE.

You could use the TDIST function in Excel to find the p-value of the test. Remember that you need to enter the positive
value of the t-statistic =TDIST(6.52,919) = 1.16E-10 which = 0.000000000116. This p-value would typically be
reported as < 0.0001 instead of as the actual value.