STT 100 Notes

Prepare
1. Context
. What does the data represent?
. What is the goal of the study?
2. Source of the data
. Are the data from a source with a special interest so that there is pressure to obtain
results that are favorable to the source?
3. Sampling Method
. Were the data collected in a way that is unbiased, or were the data collected in a way
that is biased (such as a procedure in which respondents volunteer to participate)?

Analyze

1. Graph the Data

2. Explore the Data
. Are there any outliers (numbers very far away from almost all the other data)?
. What important statistics summarize the data (such as the mean and standard
deviation)?
. How are the data distributed?
. Are there missing data?
. Did many selected subjects refuse to respond?
3. Apply Statistical Methods
. Use technology to obtain results.

Conclude
1. Significance
. Do the results have statistical significance?
. Do the results have practical significance?

A voluntary response sample (or self-selected sample) is one in which the

respondents themselves decide whether to be included.
Voluntary samples are seriously flawed because we should not make conclusions
about a population on the basis of samples with a strong possibility of bias.
Examples:
. Internet polls
. Mail-in polls
. Telephone call in polls
Statistical significance: achieved in a study when we get a result that is very
unlikely to occur by chance. There is a statistical significance if the likelihood of
an event occurring by chance is 5% or less.
. Getting 98 girls in 100 random births is statistically significant.
. Getting 52 girls in 100 random births is not statistically significant.

Sample Data Reported Instead of Measured: When collecting data from people,
it is better to take measurements yourself instead of asking subjects to report
results. Ask people what they weight, and you are likely to get their desired
weights, not actual weights. People tend to round down, sometimes way down.
When asked, someone with a weight of 187lb might respond that he or she
weighs 160Ib. Accurate weight are collected by using a scale to measure weights,
not by asking people what they weigh.

1. Computer Virus In an AOL survey of Internet users, this question was posted
online: “Have you ever been hit by a computer virus?” Among the 170,063
responses, 63% answered “yes. “A) What term is used to describe this type of
survey in which the people surveyed consist of those who chose to respond?
B) What is wrong with this type of sampling method?
A) Voluntary sampling
B) It is a flawed sampling method because the respondents only consist of
people that wanted to respond or were motivated to respond. In this case,
the respondents felt motivated to respond because they themselves were hit
with a computer virus, therefore they wanted to volunteer and respond to
the survey. Additionally, it is a flawed sampling method because the sampling
method is not random, meaning not everyone had an equal chance of
participating in the survey. Lastly, this type of sampling method leads to
strong bias, causing misleading conclusions. For example, the conclusion that
all computer users have been hit by a computer virus based on this survey is
not reliable and misleading.

2. Reported Versus Measured: In a survey of 1046 adults conducted by Bradley

Corporation,
subjects were asked how often they wash their hands when using a public
restroom, and 70% of
the respondents said “always.”

a. Identify the sample and the population

The sample is the 1046 adults that participated in the survey
The population is all adults that use public restrooms
b. Why would better results be obtained by observing the hand washing
instead of asking about it?
Because it would prevent inaccuracies and the desire to provide
misleading answers due to pressure or desire to be liked or accepted by
 society. More accurate results would be collected by observing the adults
washing their hands instead of asking them, as there is a possibility that
the adults are lying and providing misleading answers which will result
into an unreliable survey. Observing instead of asking would result in
more accurate data collection and a more reliable survey, which would
result into a conclusion backed by stronger evidence.

3. Statistical Significance Versus Practical Significance: When testing a new

treatment, what is the difference between statistical significance and practical
significance? Can a treatment have statistical significance, but not practical
significance?
When testing a treatment, statistical significance would mean that the
treatment was found to have benefits after an experiment or study was
conducted, in which the results were unlikely to occur by chance. In practical
significance, the treatment would be given to patients, and if found that there
is a benefit, that would result into practical significance.

4. Correlation: One study showed that for a recent period of 10 years, there was
a strong correlation (or association) between the per capita consumption of
margarine and the divorce rate
in Maine (based on data from National Vital Statistics reports and the U.S.
Department of Agriculture). Does this imply that increasing margarine
consumption is the cause of an increase in
the divorce rate in Maine? Why or why not?

No, it does not imply that increasing margarine consumption is the cause of
an increase in the divorce rate in Maine? Just because two variables are
changing together, it does not mean that change in one variable directly
causes change in the other variable. Correlation does not mean causation.

In Exercises 5–8, determine whether the given source has the

potential to create a bias in a statistical study

The American Automobile Association (AAA) is a not-for-profit federation of

motor clubs that provide automotive and travel services. AAA conducts a
survey of its members
about their use of public transportation versus private automobiles. The
given source does indeed have the potential to create bias due to voluntary
sampling and asking instead of observing or measuring.

Data Set 1 “Body Data” in Appendix B includes pulse rates of subjects, and
those pulse rates were recorded by examiners as part of a study conducted by
the National
Center for Health Statistics.
There could be a potential bias if the sampling method is not random, and
there is an examiner bias since different examiners may have varying
consistencies, accuracies, and precisions when measuring pulse rates

A data set in Appendix B includes brain volumes from 10 pairs of monozygotic

(identical) twins. The data were collected by researchers at Harvard
University, Massachusetts
General Hospital, Dartmouth College, and the University of California at
Davis.

There is a sampling bias because the sample of 10 pairs of identical twins is

not representative of the general population. Additionally, there is potential
for researcher bias based on the objective and expectations of the study
and they could unintentionally influence or manipulate the study design,
data collection and results that could lead to a misleading conclusion.

An article in Journal of Nutrition (Vol. 130, No. 8) noted that chocolate is rich
in flavonoids. The article notes “regular consumption of foods rich in
flavonoids may reduce
the risk of coronary heart disease.” The study received funding from Mars,
Inc., the candy company, and the Chocolate Manufacturers Association.

There could be a bias as journals are not a reliable source. Additionally, the
article has received funding from Mars and the Chocolate Manufacturers
Association, which could influence the study and it’s objective and
expectations, resulting in bias. There is a financial incentive to come up
with a conclusion that is beneficial to mars.

Researcher bias as the researcher may intentionally or unintentionally be

influenced by the funding which could influence the study methodology,
design, analysis, and interpretation of results.

In Exercises 9–12, determine whether the sampling method appears

to be sound or is flawed.

9. Nuclear Power Plants In a survey of 1368 subjects, the following question

was posted on the USA Today website: “In your view, are nuclear plants
safe?” The survey subjects were
Internet users who chose to respond to the question posted on the electronic
edition of USA
Today.
Flawed
 10. Clinical Trials Researchers at Yale University conduct a wide variety of
clinical trials by using subjects who volunteer after reading advertisements
soliciting paid volunteers.
Flawed

In a Password Boss survey of 2030 randomly selected adults, 39%

said that they never share passwords with anyone.
Flawed

12. Social Media Usage In a survey of social media usage, the Pew Research
Center randomly selected 2002 adults in the United States.
Sound

In Exercises 13–20, determine

whether the results appear to have statistical significance, and also determine whether the
results appear to have practical significance.

13. Diet and Exercise Program In a study of the Ornish weight loss program, 40 subjects
lost a mean of 3.3 lb after 12 months (based on data from “Comparison of the Atkins, Ornish,
Weight Watchers, and Zone Diets for Weight Loss and Heart Disease Risk Reduction,” by Dan-
singer et al., Journal of the American Medical Association, Vol. 293, No. 1). Methods of statis-
tics can be used to show that if this diet had no effect, the likelihood of getting these results is
roughly 3 chances in 1000.

The results appear to have statistical significance as 40 subjects lost 3.3lb after 12 months
which was then proven to be unlikely to have been occurred by chance. However, the study
lacks practical significance as many overweight people would not see the effort and
commitment required worth it, as 3.3lbs isn’t that big of an amount to lose over the course of
12 months.

A study compared surgery and splinting for subjects suffering

from carpal tunnel syndrome. It was found that among 73 patients treated with surgery, there
was a 92% success rate. Among 83 patients treated with splints, there was a 72% success rate.
Calculations using those results showed that if there really is no difference in success rates be-
tween surgery and splints, then there is about one chance in a thousand of getting success rates
like the ones obtained in this study

The results appear to have statistical significance due to the high success rates and the
unlikelihood of it to have occurred by chance. There appears to be no practical significance as
more information is required.
A parameter is a numerical measurement describing some characteristic of a
Population.

A statistic is a numerical measurement describing some characteristic of a

Sample.

Hint The alliteration in “population parameter” and “sample statistic” helps us

remember the meanings of these terms.

Page 14

Discrete data: number of values are finite, values are quantitative, number of values can be
counted.
Continuous data: infinite possibilities of quantitative values, collection of values cannot be
counted.