Correlational Research

A. Basics

1. Typically, you have ONE group of subjects that is representative of your target population. Each subject is measured on TWO variables (X and Y).

2. The goal is to determine if a RELATIONSHIP exists between the two variables.

3. You do NOT manipulate (influence) the variables. (i.e., no IV’s or DV’s)

4. You do NOT assign participants to different groups or conditions.

5. Simply provide operational definitions and MEASURE the two variables for each subject.

B. Graph the Relationship

        1.      Plot scores on a scatterplot (X-axis and Y-axis).
        2.    If the pattern tends to run from the lower left to upper right, you have a positive correlation.
               As one variable rises (or falls) in the subjects, so does the other variable (e.g., as hours of TV
              watching per day increases, so typically, does weight).
        3.   If the pattern tends to run from the upper left to the lower right, you have a negative
             correlation. As one variable rises (or falls), the opposite occurs in the other variable. (e.g., as IQ
              scores rise, errors on tests tend to fall).
         4.   If there seems to be no pattern, you have no correlation (or a very weak one).

C. Calculate the Correlation Coefficient (r)

          1.   The correlation coefficient (r) is always a value between -1.00 and +1.00.
          2.   The closer the value is to positive or negative 1.00, the stronger the relationship
                 between the two variables.
          3.    The closer the value is to 0, the weaker the relationship.
          4.    The sign (+ or -) simply tells you if the relationship is positive or negative.
          5.    If the corresponding p-value is .05 or less, it means that your correlation is
                  statistically significant (i.e., there is less that a 5% probability that your results occurred
                  simply by chance.

D. Correlation is NOT Causation

           1.   Even with a strong correlation coefficient, you often can’t determine if X is affecting Y,
                 or if Y is affecting X (e.g., does watching a lot of TV cause you to be overweight, or does
                 being overweight cause you to watch more TV?)
           2.   There could always be a “3^rd variable” that is affecting both X and Y (e.g., lack of
                  parental supervision could lead to more TV watching AND poor eating habits).