Unit 3 – Sampling Gone Bad

Sampling Gone Bad
Unit 3 - Sampling Methods
Where does the
Data
come from?
Ann Landers (an advice columnist) once asked her
readers to respond the following question:
"If you had it to do over again, would you have children?"
With about 10,000 responses....
70 percent of the respondents said no!
Later, many other newspapers and
researchers performed a better sample of
America and found 90-95% of parents would
still have kids if they could do it all over again.
Bias: Most letters to newspapers are written by
disgruntled people - voluntary response bias
http://www.stats.uwo.ca/faculty/bellhouse/stat353annlanders.pdf
What about a poll on the Internet?
http://espn.go.com/sportsnation/polls
Voluntary Response Bias occurs when we use a voluntary response
sample.
A voluntary response sample consists of people who choose
themselves by responding to an open invitation to participate in
the survey.
Voluntary response samples show bias because people with
strong opinions (often in the same direction) are most likely to
respond.
To prepare for her book Women and Love, Shere
Hite sent questionnaires to 100,000 women asking
about love and relationships.
– 4.5% responded
– Hite used those responses to write her book
What are some problems with this?
A few things that occurred with the survey:
– respondents “were fed up with men and eager to fight
them…”
– “the anger became the theme of the book…”
– “but angry women are more likely” to respond
Nonresponse occurs when part of the sample invited
to respond does not respond (either intentionally or
unintentionally).
This is different from voluntary response because the
survey was not open to EVERYONE in the population,
just a sample.
http://www.youtube.com/watch?v=Rg5bBJQOL74
Answer the question given to you.
Speeds guessed:
contacted:
smashed:
Calculate the 5 number summary for your data set
(contacted or smashed). Be sure to check for outliers!
Parallel Boxplots:
contacted:
smashed:
Loftus and Palmer (1974) looked into eye witnesses for
car crashes.
Loftus and Palmer determined that how you ask
questions will affect the answer.
Real speed about 20 mph
in the video shown in class.
http://www.youtube.com/watch?v=SP8kJ5A5xU8
What is wrong with the following questions?
a) Don't you agree that campus parking is a problem?
How can we fix this question?
Revised question: Is parking on campus a problem?
b) Don't you agree that the new rule is a problem?
Revised Question: Do you agree or disagree that the new rule is a
problem?
c)
Do you want to eat a hamburger or the usual
vegetable sandwich?
unfair because it favors hamburger over vegetable sandwich.
Usually companies use biased questions in their
advertisements or marketing surveys to make people
favor their products over others.
d) More people in the city are using our beauty products
than any other brand. Do you use our beauty products?
Clearly the question indicates that the respondent should be using XYZ
beauty products
e) A company manufactures product A. The company conducts a
survey about the product. The following is one of the questions
in the questionnaire.
How would you rate our product?
A. Excellent
B. Good
C. Satisfactory
The question is biased, because, NO negative option is provided.
f) Is green your favorite color?
asked on an assumption
Leading Questions:
http://www.youtube.com/watch?v=G0ZZJXw4MTA
Response Bias is when the wording of the question or the
person asking the question can in any way
influence the response. Leading questions are an
example of response bias.