Correlation Part 2
For this exercise, you'll be working with some questionnaire data I collected. The questionnaire included five attitude measures:
You can download the dataset here:
Creating a Scatterplot
Let's compute the correlation between AWS and Agency that we saw on the last page. First, plot the data: Select Graphs -> Legacy Dialogs -> Scatter/Dot -> Simple Scatter, then put aws in the Y-axis box and agency in the X-axis box:
Then click OK. You should get a scatterplot like this:
Notice how high scores on AWS tend to be found with low scores on Agency. This would lead you to expect a negative correlation.
Computing a Correlation
To compute the correlation between Agency and AWS, select Analyze -> Correlate -> Bivariate (bivariate means "dealing with two variables"). Move AWS and Agency into the variables box and press OK. You should get output looking like this:
These tables can be hard to read at first, so go slowly. The first thing to notice about correlation tables is that they are symmetric - the upper right corner is a mirror image of the lower left corner. You can ignore the information in the lower left corner because it is all contained in the upper right corner.
- is the actual correlation coefficient r. The correlation between any variable and itself (AWS with AWS) is a perfect 1.0. The correlation between AWS and Agency is r = -.843. There are two stars after this number, which, if you look at the table above, indicates that the correlation is significant at p < .01. The exact p-value is given in the second row:
Here you can see that the correlation is significant at p < .001 (remember that when SPSS says p = .000, it really means p < .001). In the third row:
are the number of observations that went into the correlation. This means that the correlation was based on responses from 201 people.
To report the correlation above in APA style, you would write...
Important: always provide an interpretation after you report the correlation.