Chapter 10 – Relationships Between Measurement Variables
Correlation is the strength of a certain type of relationship between 2 measurement variables.
Regression is the numerical method for trying to predict the value of one variable by knowing the value
of another one. A statistical relationship is different from a deterministic relationship because in the
second one we can directly determine the value of one variable if we have the other. Example: cm m,
kg lb. in a statistical relationship there is natural variability within both variables. They are good for
describing what happens to a population, or aggregate. The stronger the relationship the more useful it
will be for predicting for an individual. Statistical relationships do not hold for everyone.
To find out if a statistical relationship exists between 2 variables researchers measure a sample of
individuals. A relationship may exist or it could just be luck. Some numbers are too small to be
convincing. Researchers determine if results are statistically significant by asking if the chances of the
relationship would be the same if there was nothing going on in the population. If the chances are small
the results are statistically significant. If it is less than 5% it is significant. If the relationship is stronger
than 95% then it is significant. It is easier to rule out chance if the sample is really large. Results with
statistical significance may not even mean the relationship is strong.
Correlation between 2 measurement variables is an indicator of how close their values fall to a straight
line. This is called the Pearson product-moment correlation or the c