Page 343 - Statistics for Dummies

Chapter 20: Ten Tips for the Statistically Savvy Sleuth
✓ The statistical term correlation is only used in the context of two
numerical variables (such as height and weight). It does not apply to
two categorical variables (such as political party and gender).
For example, voting pattern and gender may be related, but using the

word correlated to describe their relationship isn’t “sc” (statistically cor-
rect, get it?). You can say two categorical variables are associated.
✓ If a strong correlation and scatterplot exist between two numerical
variables, you should be able to draw a straight line through the points,
and the points should lie close to the line. If a line doesn’t fit the data well,
the variables likely won’t have a strong correlation (r), and vice versa. (See
Chapter 18 for information on line-fitting, also known as linear regression.)
A weak correlation implies that a linear relationship doesn’t exist between

the two variables, but this doesn’t necessarily mean the variables aren’t
related at all. They may have some other type of relationship besides a
linear relationship. For example, bacteria multiply at an exponential rate
over time (their numbers explode, doubling faster and faster).
✓ Correlation doesn’t automatically mean cause and effect. For example, 327
suppose Susan reports based on her observations that people who
drink diet soda have more acne than people who don’t. If you’re a diet
soda drinker, don’t break out just yet! This correlation may be a freak
coincidence that only happened to the people she observed. At most, it
means more research needs to be done (beyond observation) in order
to draw any connections between diet soda and acne. (Susan can read
Chapter 17 to find out how to design a good experiment.)
Reveal Confounding Variables
A confounding variable is a variable that isn’t included in a study but whose
influence can affect the results and create confusing (confounding) conclu-
sions. For example, suppose a researcher reports that eating seaweed helps
you live longer, but when you examine the study, you find out that it was
based on a sample of people who regularly eat seaweed in their diets and are
over the age of 100. When you read the interviews of these people, you dis-
cover some of their other secrets to long life (besides eating seaweed): They
slept an average of 8 hours a day, drank a lot of water, and exercised every
day. So did the seaweed cause them to live longer? You can’t tell, because
several confounding variables (exercise, water consumption, and sleeping
patterns) may also have contributed.
The best way to control for confounding variables is to conduct a well-
designed experiment (see Chapter 17), which involves setting up two groups
that are alike in as many ways as possible, except that one group receives a
specified treatment and the other group receives a control (a fake treatment,
no treatment, or a standard, non-experimental treatment).You then compare

3/25/11 8:12 PM
29_9780470911082-ch20.indd 327 3/25/11 8:12 PM
29_9780470911082-ch20.indd 327