If you are talking about a point (sharp) null hypothesis, true
(unless, of course, you have the entire population of scores on hand), -- but
testing point null hypotheses is pretty silly anyway. If we must test null
hypotheses, they should be range (loose) null hypothesis - that is
close second.
Sent from my BlackBerry® wireless device from U.S. Cellular
-Original Message-
From: Marc Carter
Date: Thu, 16 Jun 2011 14:18:45
To: Teaching in the Psychological Sciences (TIPS)
Reply-To: "Teaching in the Psychological Sciences (TIPS)"
Subject: RE: [tips] Nul
:pcbernha...@frostburg.edu]
Sent: Thursday, June 16, 2011 2:52 PM
To: Teaching in the Psychological Sciences (TIPS)
Subject: Re: [tips] Null hypothesis
Unfortunately, it seems to me that textbooks are drifting all over the place on
this issue.
Our currently selected textbook for statistics uses &q
Unfortunately, it seems to me that textbooks are drifting all over the place on
this issue.
Our currently selected textbook for statistics uses "retain the null
hypothesis" in the case where it is not rejected. I've decided not to argue
with the textbook, because I feel I have bigger fish to f
t: RE: [tips] Null hypothesis
I teach my students there are two possible outcomes to a null hypothesis test:
reject or fail to reject.
That said, if you have power in the area of .99, then you might have good
reason to believe the null is true. But you still don't "accept" it.
I teach my students there are two possible outcomes to a null hypothesis test:
reject or fail to reject.
That said, if you have power in the area of .99, then you might have good
reason to believe the null is true. But you still don't "accept" it.
Bayesians, as I am learning, might have a whol