Statistics for the Behavioral Sciences by Frederick J. Gravetter, Larry B. Wallnau ISBN 10: 1305504917 ISBN 13: 9781305504912

522 CHAPTER 15 | Correlation
To compute a partial correlation, click Analyze on the tool bar, select Correlate, and click
on Partial. Move the column labels for the two variables to be correlated into the Variables
box and move the column label for the variable to be held constant into the Controlling for
box and click OK.
To compute the Spearman correlation, enter either the X and Y ranks or the X and Y scores
into the first two columns. Then follow the same Data Analysis instructions that were presented
for the Pearson correlation. At step 3 in the instructions, click on the Spearman box before the
final OK. (Note: If you enter X and Y scores into the data editor, SPSS converts the scores to
ranks before computing the Spearman correlation.)
To compute the point-biserial correlation, enter the scores (X values) in the first column
and enter the numerical values (usually 0 and 1) for the dichotomous variable in the second
column. Then, follow the same Data Analysis instructions that were presented for the Pearson
correlation.
The phi-coefficient can also be computed by entering the complete string of 0s and 1s into
two columns of the SPSS data editor, then following the same Data Analysis instructions that
were presented for the Pearson correlation. However, this can be tedious, especially with a large
set of scores. The following is an alternative procedure for computing the phi-coefficient with
large data sets.
Data Entry
1. Enter the values, 0, 0, 1, 1 (in order) into the first column of the SPSS data editor.
2. Enter the values 0, 1, 0, 1 (in order) into the second column.
3. Count the number of individuals in the sample who are classified with X = 0 and Y = 0.
Enter this frequency in the top box in the third column of the data editor. Then, count how
many have X = 0 and Y = 1 and enter the frequency in the second box of the third column.
Continue with the number who have X = 1 and Y = 0, and finally the number who
have X = 1 and Y = 1. You should end up with 4 values in column three.
4. Click Data on the Tool Bar at the top of the SPSS Data Editor page and select Weight
Cases at the bottom of the list.
5. Click the circle labeled weight cases by, and then highlight the label for the column
containing your frequencies (VAR00003) on the left and move it into the Frequency
Variable box by clicking on the arrow.
6. Click OK.
7. Click Analyze on the tool bar, select Correlate, and click on Bivariate.
8. One by one move the labels for the two data columns containing the 0s and 1s (probably
VAR00001 and VAR00002) into the Variables box. (Highlight each label and click the
arrow to move it into the box.)
9. Verify that the Pearson box is checked.
10. Click OK.
SPSS Output
The program produces the same correlation matrix that was described for the Pearson
correlation. Again, you want the correlation between X and Y which is in the upper right
corner (or lower left). Remember, with the phi-coefficient, the sign of the correlation is
meaningless.
FOCUS ON PROBLEM SOLVING
1. A correlation always has a value from +1.00 to –1.00. If you obtain a correlation outside
this range, then you have made a computational error.
2. When interpreting a correlation, do not confuse the sign (+ or –) with its numerical value.
The sign and the numerical value must be considered separately. Remember that the sign