I was asked a question about the correlation coefficient. On calculators, this is the r-value that may pop up when you have done a regression. If this value is close to -1 or 1, the data points in question provide a good fit to the proposed equation. Actually, this r-value only applies to linear regressions, it is only defined there.

Someone said that that was not correct, seeing as the r-value also occurs on logistic or exponential regression. Actually, these are also linear regressions, they are transposed into linear relationships before regression analysis is carried out.

Anyway, now that calculators are too old, and Excel is too expensive, how do we carry out such a regression with the correlation coefficient also calculated? The answer (again) is GeoGebra. OK, to be fair, OpenOffice could probably just as well be used for this, but since I am a GeoGebra fan…

In the video below I enter some data points, do the regression and find the correlation coefficient. The menus are in Norwegian, but I am sure you can find the right commands in whatever language you want. You can easily change language in GeoGebra from the settings menu.

GeoGebra – Korrelasjonskoeffisient from Øistein Gjøvik on Vimeo.