Lab 6.

5 & ESC

Statistics 200: Lab Activity for 6.5 and Essential Synthesis C

Inference for means and proportions using theory-based distributions

Learning objectives:
 Distinguish between separate (independent) samples and paired data when comparing means
 Identify the advantages of using paired t procedure with data sampled as matched pairs
 Use Minitab to perform hypothesis tests and construct confidence intervals
 Demonstrate an understanding of when to use two theoretical distributions: normal or t with
inference for means and proportions when answering questions about real data

Confidence Interval Standardized test statistic

Sample statistic ± (multiplier)×SE sample statistic−null value
SE

Note: This lab uses Minitab. Have your guide ready!

Activity 1: Matched pairs or independent/separate samples

For each of the prompts below, decide whether the parameter of interest is a paired difference in means
(matched pairs, μd ), or a difference in means (independent samples, μ1−μ2 ). Also determine if the
research question is asking for a confidence interval or hypothesis test.

1. Students want to know if it matters where they have their cell phone screen repaired. A researcher
used available data from a sample of eight cell phones with broken screens. For each phone, an
estimate for the screen repair in U.S. dollars ($) was obtained from both a local store and an on-line
merchant where you will have the additional cost of shipping. Your goal is to determine if, on
average, the estimate in dollars for the total repair from the on-line merchant is larger?

Identify the case: parameter of interest:

Test or Interval:

2. A study wants to determine how free Wi-Fi affects data usage on a long-distance bus ride. A group
of eighteen buses traveling from State College to New York City were randomly assigned where
nine buses gave free Wi-Fi to passengers and the other nine buses, while traveling the same route,
offered Wi-Fi for a one-time charge for its passengers. The study measured the amount of data
used, in gigabytes, on each bus for the ride. The goal is to determine, on the average, how different
is the amount of data used when comparing free Wi-Fi to a paid service in gigabytes.

Identify the case: parameter of interest:

Test or Interval:

11/6/22 © Pennsylvania State University

 Lab 6.5 & ESC

3. An advertising firm seeks to compare two advertisements for the same product. Selected
participants are matched as a pair when based on age, sex, and income. A coin is flipped to assign
the two advertisements within each matched pair. The advertising firm compares average users'
willingness to purchase the product, when reported on a licker score, for the two advertisements.
Does the data suggest that there is a difference in the two mean rating scores?

Identify the case: parameter of interest:

Test or Interval:

Activity 2: Advantage of using paired samples

Research Question: Is there a difference in the average noise level (in decibels) when considering two
distances “30 centimeters” and “45 centimeters” away from a hand dryer in public restrooms?

The biggest cause of hearing loss is not aging, as many people believe, but noise. Exposure to sounds with
dangerously high decibel levels – 85 decibels or higher – can lead to permanent hearing impairment over
time. People of all ages are susceptible to noise-induced hearing loss – even kids. Moreover, Health
Canada prohibits the sale of toys with peak loudness greater than 100 decibels.

A 13-year-old Nora Keegan from Canada questioned whether hand dryers might be a source of noise-
induced hearing loss, so she decided to find out as part of a science fair project. Her interest was spurred
after she observed kids in bathrooms covering their ears and avoiding using the hand dryers. She herself
had noticed they sometimes caused her ear pain. This ambitious teen’s study was published 1.

In one part of this study, Keegan measured the noise levels of twenty different hand dryers found in
public restrooms in her hometown of Calgary for over a year. A decibel-meter was used to measure the
peak loudness of hand dryers at two distances from the wall: 45 centimeters (the distance commonly
used by manufacturers for testing) and at 30 centimeters (the distance many children’s ears are away
from the wall) when using a hand dryer at the authors height.

1
Keegan, N.L. (2019). “Children who say hand dryers ‘hurt my ears’ are correct: A real-world study
examining the loudness of automated hand dryers in public places.” Paediatrics & Child Health, Volume
25(4).

Exploratory Data Analysis

Consider the two variables ‘Dist_30’ and ‘Dist_45’. Create a graph that displays a dotplot for each
variable.

Graph > Dotplots > Multiple Y Variables > Simple

1. Did any of the hand dryers have a loudness that was greater than 100 decibels?

11/6/22 © Pennsylvania State University

 Lab 6.5 & ESC

2. Are the centers similar?

3. The study found that, in general, older-appearing hand dryers tested quieter than newer hand
dryers. Could this help to explain the unusual distribution found with each dotplot?

4. This is an example of paired data. Why is this a true statement? What is the case?

Analysis
What happens when we build a confidence interval for the difference in means without taking the pairing
into account? We are doing this to make a numerical comparison.

5. Use Minitab to build the 95% confidence interval with the two-sample t procedure.

Stat > Basic Statistics > 2-sample t > Select ‘Each sample is in its own column’ from the drop-
down menu. Sample 1: ‘Dist_30’ and Sample 2: ‘Dist_45’

x 1−x 2 =______ decibels 95% confidence interval for µ1 – µ2 is ( ) decibels

6. What conclusion can be made with the 95% confidence interval?

7. With the corresponding two-sided test, what is the…

p-value:
Generic conclusion:

Note: Remember that an advantage of using a paired means approach is the ability to remove extra
variation and focus on the variation of interest. We accomplish this task by using the differences rather
than the original two samples. With this activity, the calculated difference = (Dist_30 – Dist_45).

8. State in words what extra variation are we trying to remove.

9. Create a dotplot of the Sample: ‘Diff (30-45)’ What do you notice? Does this sample reasonably
meet the normality condition?

10. With all 20 brands, is the decibel level higher at a distance of 30 centimeters from the wall?

11. Use Minitab to find the summary statistics for all three variables found in Table 1.

Stat > Basic Statistics > Display Descriptive Statistics [select all 3 variables]

Fill in Table 1 for the missing standard deviations.

11/6/22 © Pennsylvania State University

 Lab 6.5 & ESC

Table 1: Summary statistics for all three variables

Variables sample mean (decibels) sample standard deviation (decibels)
Dist_30 92.54
Dist_40 91.90
Diff (30 – 45) 0.64

12. What do you notice when you compare the standard deviations for the three samples?

13. Use Minitab to build a 95% confidence interval for µd:

Stat > Basic Statistics > Paired t > Sample 1: ‘Dist_30’ and Sample 2: ‘Dist_45

x d =¿ ___________ decibels 95% confidence interval for µd: ( ) decibels

14. What conclusion can be made with the 95% confidence interval?

15. With the corresponding two-sided test, what is the…

p-value:
Generic conclusion:

16. Compare the 95% confidence interval for µd with the 95% confidence interval for µ1 – µ2. What is
the primary difference between the two intervals when comparing the conclusions?

17. With the confidence interval for µd, do you believe that we have found practical significance
(importance)? Note: the difference between (between 30 centimeters and 45 centimeters) is
about 6 inches.

18. Compare the p-values from the two tests. What do you notice?

19. With this example, you have learned about the advantage of using a paired design when the
samples are paired. Why is this true?

Activity 3: Testing hypotheses with your class data

This activity uses your class dataset. Making the dubious assumption that this is a representative sample
of University Park students, we are going to test the following hypotheses.

11/6/22 © Pennsylvania State University

 Lab 6.5 & ESC

A. A recent survey (eMarketer, 2022) found that TikTok users in the US spend an average of 45
minutes (or 0.75 hour) per day on the platform. Does our sample provide convincing evidence that
the true average time Spent on TikTok is different from 0.75 hours?

B. According to recent research, the average cost of textbooks and supplies for the 2020 to 2021
school year at 4-year institutions was $1,226 per year (or $613 per semester). Does our sample
provide convincing evidence that Stat 200 students spend more per semester on books and
supplies than the reported average of $613 per semester?

C. Is there a difference in the proportion of students who identify as liberals based on their housing
status (on-campus vs off-campus)?

1. Complete the table below for each of the hypotheses above.

Table 2:

A B C

Parameter of interest

Standardized test statistic 18.66 -3.84 -0.79

(Given)
Which distribution should
we use to calculate the p-
value? t or Z
Direction of Ha

P-value (Given) < 0.001 1.000 0.475

Generic conclusion
(reject, fail to reject)

Which type of error is

possible?

2. Overall, are the conclusions what you expected, or did you find any surprises?

11/6/22 © Pennsylvania State University