I am working on a project for my job that requires, in part, some regression analysis performed on a data set. Shockingly (to me), the R-Squared calculated when fitting a trendline to the data in Powerpoint was almost double the R-Squared for the same data points when I used Excel. Googling possible reasons why turned up nothing. Has anyone else had this experience? If not, what could possibly be causing this?
Under the category of human error, is it possible the ranges don’t include the same data points?
If it’s not human error, then it’s probably a bug in one of the Office products. Run it in R and see what pops out there.
Perhaps you are (unknowingly) fitting different types of curves to the data? For example, linear in PowerPoint and logarithmic in Excel.
Pardon a small hijack. Does R do the same analysis as SAS? I used SAS on a Vax mainframe at work a few years ago.
Yes, spend 3 months learning how to write the command line in R that will perform the trend analysis you want.