Statistics - Correlation Matrix

If you ask for cross tabulation between two number fields or weighted score questions, Raosoft SURVEYWin computes Pearson correlation between the two fields. Here is a quick example to describe what correlation means:

If you have two questions on your form such as 'How much money do you make?' and 'What type of car do you drive?' and you want to see if people who make more money drive expensive cars, you would do a correlation matrix on these two questions.

If you are using weighted score questions then you need to make sure you set up the options accordingly. For example if your income options were:

  1. $10k
  2. $20k
  3. $30k
  4. $40k
Then in order to correlate properly you would want your vehicle choices in ascending order of cost:
  1. $5k
  2. $10k
  3. $15k
  4. $40k

With our questions set up like this, the following can be inferred:

  • If you get a positive correlation, then people who make more money drive more expensive cars.
  • Conversely, if you get a negative correlation, then people who make more money drive less expensive cars.

  • If the correlation value is marked with a + symbol, then your correlation is significant. A non-significant correlation means that your two variables are not strongly related (ie: income has no effect on car ownership).

    To create a Correlation Matrix:

    1. Select Statistics Correlation Matrix or use the keyboard combination Ctrl+Shift+M. The Correlation Analysis dialog box will appear.
    2. In the listbox labeled Select Questions, select the question to use in the plot and move it into the Questions Selected window by using the right arrow button. You must move in two weighted score or write in number questions to create a correlation matrix.
    3. Click Finish. A cross-hair cursor will appear. Locate the cross-hair where you would like the plot to appear and press the left-mouse button.