So, the gist of it is, from testing between doing the work by hand, and doing it on the calculator, I'm getting incorrect results on the calculator (checked with my teacher.) I am also getting ridiculous answers, like getting more residuals than there are x and y values (or getting less, seems I get 8 either way for some reason.)
I'm going to run a Regression tTest (TEST -> t -> REG) where Ho = 0 and Ha < 0
X list will be in List 1, the Y list will be in list 2, and goes as follows: n=19
After running the test, everything except the standard error is correct.
se = 37.8785798
Actual se = 3.457162419 (Formula used is (StdDev of residuals) / (StdDev of X multiplied by sqrt(n-1)
Now the residuals are really strange for me and my teacher. Assuming we have been talking about the residuals the entire time, the "Save Res: List" option, and I haven't completely missed something about statistics.
The residuals from the calculator regtTest go as follows:
That's it. It only covers the first 8 values, and incorrectly as well. From what I know, there should be residual for all 19 x and y pairs, and they should add up to 0. This sums up to 285.88
The actual residuals are:
etc till you have all 19
I did discover one thin just now typing this up. I can get the actual residuals by going to the setup screen and having the calculator save all residuals from a regression test (and linear regression tTest) into a chosen list. This means I have to set the "save Reg: List" option to "none" so I don't get the wrong one.
Anyway, please tell me where I am wrong, thanks.
Edit: I did this with the calculator before and after updating the OS to the latest version, same results.
It's also very late at night when I am writing this, sorry if I made some unseen blunder.
I had something I felt like I wanted to edit into here, but I can't remember it now.
Edited by Crazierinzane, 02 May 2013 - 04:09 AM.