Page 36 - Statistics for Dummies

20
Part I: Vital Statistics about Statistics
Correlation, regression,
and two-way tables
One of the most common goals of research is to find links between variables.
For example,
✓ Which lifestyle behaviors increase or decrease the risk of cancer?
✓ What side effects are associated with this new drug?
✓ Can I lower my cholesterol by taking this new herbal supplement?
✓ Does spending a large amount of time on the Internet cause a person to
gain weight?
Finding links between variables is what helps the medical world design better
drugs and treatments, provides marketers with info on who is more likely to
buy their products, and gives politicians information on which to build argu-
ments for and against certain policies.
In the mega-business of looking for relationships between variables, you find
an incredible number of statistical results — but can you tell what’s correct
and what’s not? Many important decisions are made based on these studies,
and it’s important to know what standards need to be met in order to deem
the results credible, especially when a cause-and-effect relationship is being
reported.
Chapter 18 breaks down all the details and nuances of plotting data from two
numerical variables (such as dosage level and blood pressure), finding and
interpreting correlation (the strength and direction of the linear relationship
between x and y), finding the equation of a line that best fits the data (and
when doing so is appropriate), and how to use these results to make predic-
tions for one variable based on another (called regression). You also gain
tools for investigating when a line fits the data well and when it doesn’t, and
what conclusions you can make (and shouldn’t make) in the situations where
a line does fit.
I cover methods used to look for and describe links between two categorical
variables (such as the number of doses taken per day and the presence or
absence of nausea) in detail in Chapter 19. I also provide info on collecting
and organizing data into two-way tables (where the possible values of one
variable make up the rows and the possible values for the other variable
make up the columns), interpreting the results, analyzing the data from two-
way tables to look for relationships, and checking for independence. And, as I
do throughout this book, I give you strategies for critically examining results
of these kinds of analyses for credibility.

3/25/11 8:18 PM
05_9780470911082-ch01.indd 20 3/25/11 8:18 PM
05_9780470911082-ch01.indd 20