Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We have measured reading speed in patients with diabetic eye disease. What we have found is that reading speed is faster in younger patients and in those who has better vision as you would expected. However, we have developed an novel measurement which is much faster to do than measuring reading speed. Both reading speed and the novel measurement correlate well with each other but we would like to see whether this correlate is independent of age and vision level. How can I do that in SPSS. Many thanks.

share|improve this question
Are you asking how to design an experiment to determine this, or do you already have the data? Do you only want to know how to get SPSS to do this (in which case, the question is off-topic, per our FAQ), or are you wondering about the statistical methods involved in assessing this issue? – gung Dec 4 '12 at 13:24
We have the data but what statistical methods should I use to determine that. V – user17952 Dec 18 '12 at 22:17

I see an easier but fragmented way and a more difficult but comprehensive way.

The easier way is to divide up your subjects into, say, 3 age categories and 3 vision-level categories, and to compute correlations for each group. (In SPSS, use the "split file" command). Then you could assess the degree to which these 9 correlation coefficients differ by group. You could do it graphically in any number of ways and/or using formal statistical tests of the difference between correlations; the latter would involve first transforming each correlation using Fisher's Zr method (available through online calculators but not in SPSS). In order to test every difference, you would need to run 9!/(2!(9-2)!) tests, or 36 of them.

The more difficult, but also more elegant, way would be to preserve age and vision level just as they are--not truncating into groups--and use them to build interaction terms into a regression. Your question would now be, "To what extent does the connection between the new measurement (Y) and the old measurement (X1) vary depending on the level of X2 (age) and X3 (vision level)?" You would need to get familiar with linear regression, its assumptions, and the use of continuous variables to test for interactions.

share|improve this answer
+1 but I think the "more difficult" way is actually in the end simpler and far better. One of the main uses of regression is to answer this sort of question. – Peter Ellis Feb 17 '13 at 1:42

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.