Linear regression is a way of analysing the trendline that we talked about in the context of correlation. Regression provides an straight-line equation of the form:

y = a + bx

where y is the variable being predicted, x is the variable you're predicting it from, and a and b are two numbers calculated by your analysis.

a is the intercept of the regression line (the value of y when x is zero) and b is the slope. (NB you may see some books that call them 'm' and 'c'. Very confusing, but don't worry about it too much - they're just different names for the same thing. We could call them 'Bert' and 'Ernie' if we like - they're just names.).

This equation accurately describes the trendline.

Like in correlation, regression also provides an r measure, describing how well the model fits the data, except we use upper-case R to show it's regression. The addition of the equation describing the line adds a lot to simple regression - it allows you to predict the value of one variable from another. Remember that R^2 (r squared) tells you just how accurate these predictions are likely to be.

analyze > regression > linear