In This Chapter
Another look at covariance
Why and how we analyze covariance
Analysis of Covariance (ANCOVA) in Excel
One of the major points of this book is that Excel comes with a surprising number of analytical tools and formulas. The toolset isn't as extensive as you'd find in a dedicated statistics package, but it's still impressive.
Some analyses, unfortunately, aren't part of Excel. And some of those might turn out to be important for you. In many cases, with a little ingenuity you can use the existing parts of Excel to perform those analyses anyway. In this Appendix, I focus on one of those analyses.
In Chapter 15, I mention covariance in connection with correlation. I spoke about it briefly as the numerator of the correlation coefficient. I also mention that covariance represents two variables changing together.
What does that mean, exactly?
Imagine a group of people on whom we measure mathematical ability and sociability. (Let's just assume we have valid, reliable ways of measuring both.) If we find that the people with high mathematical ability are the most sociable, and the people with low mathematical ability are the least sociable, this thing called covariance is numerically high and positive. This type of positive relationship is called a direct relationship.
A different result is possible: The people with high mathematical ability might turn out to be the least sociable, and the people with low mathematical ability ...