Bias and Misrepresented Data

Bias and
Misrepresented Data
I've randomly
chosen the 2 students
for standardized
testing...
Bias
• Bias is a when a question or sampling method
unfairly favours a particular answer.
• If you ask the employees
at a particular restaurant,
"who has the best
chicken wings in town",
they'll likely pick their
own. This is why K-W has
200 places that have the
"best wings in town".
• Random sampling means that any member of
the population has an equal chance of being
selected to be part of the sample.
• When a sample is selected at random from a
population, it is said to be an unbiased
sample.
• If the sample is selected in a non-random way,
it may result in a biased sample which means
that the conclusions based on the sample may
not reflect the population that they're from.
Examples:
• If you're trying to find out what
percentage of the student
population at school was born
in a different country, choosing
an ESL class as your sample will
give a biased result.
• If you want to know what
percentage are smokers, and
you take your sample from
students who hang out in the
school's smoking area, you'll get
a biased result.
Types of Bias
• There are a few different types of bias that
may be introduced if the sample is not
selected correctly.
• A researcher must avoid bias when sampling
otherwise the results will be invalid.
Sampling Bias
• Sampling bias occurs when the sample does
not accurately represent the population.
• Example: You want to find the average age of
the students at your school, and choose 2
grade 9 classes and a grade 11 class as your
sample.
Non-Response Bias
• Non-response bias can occur when the
method of data collection (such as
questionnaires) are not returned and so
results are influenced.
• Example: If students are given a questionnaire
about how many would like to participate in
math contests, only those who actually want
to participate are likely to complete and
return them.
Household Bias
• Household bias occurs when strata (divisions)
from the sample group are not equally
represented.
• Example: A survey to determine the average
speed on a highway would be biased if taken
only during rush hour. Other strata (in this
case, time intervals) need to be included (such
as late evenings, etc.).
Response Bias
• Response bias occurs when people completing
the surveys may answer questions in a way
that they think the person conducting the
survey wants them to answer.
• This can be based on the conditions of the
survey or the nature of the questions.
• Example: If a taste
test between Pepsi
and Coke is being
conducted, the results
will likely differ
depending on
whether the person
conducting the survey
is dressed neutrally or
wearing a jacket and
hat from one brand.
• In the words of Steven
Colbert...
• "George W. Bush...
GREAT President, or the
GREATEST President?"
• Questions worded this
way will likely have
response bias.
Example 1:
• In order to collect your sample, you ask
students who are in the cafeteria during
period 1.
• What bias may be introduced with this
sampling?
Example 2:
• You realize that the method described in
Example 1 is biased.
• To fix this, you decide to provide a
questionnaire to one person from each
homeroom. You ask the students to take the
completed questionnaire to the activity office
when they are done.
• What type of bias might be
introduced with this sample?
Example 3:
• You realize that the method described in
Example 2 is still biased!
• To correct the problem of non-response bias
you decide to wait in each class until the
questionnaire is returned to you.
• What bias may be introduced with this
sampling?
"Ignore it.
It's unscientific."
Misrepresented Data
• Many groups try hard to misrepresent data by
displaying it in improper ways. It's always
important to look closely at what the data
says, rather than the way it looks at first.
What's wrong with
these graphs?