P VALUES ARE THE DEATH OF ME

McPass you definiton is just brilliant. Makes me see that what I just wrote is wrong…

mcpass Wrote: ------------------------------------------------------- > Think of p-value as a score on a horrible looking > girl. YOU WANT TO REJECT THAT ONE. REJECT IT AND > YOU’RE HAPPY. > > p value is the score your friends give the girl. > You want them to give it a low score so you can > reject her on that basis. Otherwise you cannot > reject her and she’ll want happy times. > > The critical value (10%, 5%, 1%) is the ‘okay’ > score. If she scores below it, you can reject her > and look for a better one. If your friends say her > score is above that value (say, they give her a > p-value of 12%) you can NOT reject. > > If you cannot reject you’re screwed. You don’t > want that. Low p-values is what you want to see. Just to further clarify this bit of brilliance, when you say “The critical value (10%, 5%, 1%) is the ‘okay’ score” you really mean the level of signifigance, correct?

*in the terminator’s voice when speaking to miles dyson*: listen to me very carefully picture the probability distribution curve. the very center is 0. your test statistic plots somewhere on the x-axis, let’s say it’s 5. the area underneath the curve AFTER 5 is the p-value-- it’s a probability figure. let’s say your level of significance is 5% (one sided) then that’s the area of the curve after the critical value (the value you get from the chart in the back of the book, to put it simply). if your p-value is smaller than that then you know that your test statistic is further away from 0 than the critical value so the area underneath the curve will be smaller. so a smaller p-value will make your level of significance (alpha) lower and your conclusion of rejecting the null stronger

“Get all heteroskatastic on her” ------------------------------------- OMG…i laughed my balls off for 10 minutes straight after readin this…smarshy…if you EVER decide on an alternative career as a stand up/improv artist… it’ll be my honor to be present on your first show!!!

You guys are the best… my quant is solid now…

Smarshy Wrote: ------------------------------------------------------- > mcpass Wrote: > -------------------------------------------------- > ----- > > Think of p-value as a score on a horrible > looking > > girl. YOU WANT TO REJECT THAT ONE. REJECT IT > AND > > YOU’RE HAPPY. > > > > p value is the score your friends give the > girl. > > You want them to give it a low score so you can > > reject her on that basis. Otherwise you cannot > > reject her and she’ll want happy times. > > > > The critical value (10%, 5%, 1%) is the ‘okay’ > > score. If she scores below it, you can reject > her > > and look for a better one. If your friends say > her > > score is above that value (say, they give her a > > p-value of 12%) you can NOT reject. > > > > If you cannot reject you’re screwed. You don’t > > want that. Low p-values is what you want to > see. > > > Just to further clarify this bit of brilliance, > when you say “The critical value (10%, 5%, 1%) is > the ‘okay’ score” you really mean the level of > signifigance, correct? Alcohol. A sober day will get you and your friends to use 10%. P-value of 0.06 doesn’t cut it on those days. Once you start drinking you may get flexible more and more. P-value of 0.06 may suddenly not lead to rejection anymore. And when your ‘okay score’ (or significance or whatever you want to call it) is down to 1% you can only reject the real ugly once with p-values below 0.01. Waking up is the real disappointment those days.

Wow, it’s like you know me.

We’ve all had our Wish-I-Had-Rejected-This-P-Value-Morning.

p is for probability. It’s the probability that you’d get that value if the null was true. I’ve always understood hypothesis tests in terms of p values rather than the other way round. I suppose it’s just how your brain is wired.