While I try to find out if we're correlated positively, or whether our relationship is linear , I find that you are testing me to a critical point. And there I used to think that you were normally distributing me.
Try not to sum the squares of errors, and think of the least squares. Since when did you have to question your hypothesis to find out whether we were linear or not. So what happens when you find the t value? and what if my t-critical value is less than your t-value? will you reject me? I'm scared that you might have to additionally test our fit by conducting an f-statistic, because you might find an SSE ( an error sum of squares)
Oh so you're only 95% confident? This confidence interval is not that important, what's important is the 0.05 significance value- Yes it's significant ...because with this significance, you get your degrees of freedom..
See now that you forward selected, and really looked backward just to find what's best.. you find that my R^2 value is high, and that my Cp is low.. This means that i'm good enough of a variable to consider...
You better not test me again.. because I'm a full, not a reduced model.
Posts
