I need help with t-test
Registered User
Joined: Mar 2009
Posts: 2,887
Likes: 0
Originally Posted by magician,Dec 1 2009, 11:12 PM
Perhaps a feeling similar to those experienced by people who come here seeking legal advice, medical advice, career advice, romantic advice, . . . .
wait.... what???
Registered User
Joined: Jul 2001
Posts: 6,592
Likes: 0
From: Yorba Linda, CA
Originally Posted by tarheel91,Dec 1 2009, 08:14 PM
You're going to want to subtract the data from one product from the data from the other (for each person).
Originally Posted by tarheel91,Dec 1 2009, 08:14 PM
Then, do a T Interval test and create a range of 2-3 standard deviations in both directions for the difference.
Originally Posted by tarheel91,Dec 1 2009, 08:14 PM
If 0 isn't included in the range, one product is clearly better than the other. If it is, you can't conclude anything meaningful from the data.
Registered User
Joined: Aug 2008
Posts: 862
Likes: 0
Originally Posted by magician,Dec 2 2009, 12:41 AM
Has it been established that the purpose of the test is to determine whether there's a statistically significant difference in the average time to complete the task using one product instead of the other?
The number of standard deviations to use will, of course, depend on the confidence level of the test.
At the specified level of confidence. A different level of confidence could lead to a different conclusion.
The number of standard deviations to use will, of course, depend on the confidence level of the test.
At the specified level of confidence. A different level of confidence could lead to a different conclusion.
Registered User
Joined: Jul 2001
Posts: 6,592
Likes: 0
From: Yorba Linda, CA
Originally Posted by tarheel91,Dec 1 2009, 10:04 PM
There's not much else you can do with a t test here. I thought it was fairly obvious that he's trying to compare the 2 products performance. Thus, finding a statistically significant difference is probably the point of all of this. I realize the standard deviations depend on the confidence level, but do you really think someone who doesn't know how to do a t test knows what a confidence level is? Most confidence levels (outside of engineering) are somewhere between .05 and .01, so I said 2-3 just to help him out a bit. If this isn't a stats class, we don't have to be too technical.
Oh, well.
(At least we didn't tip the degrees-of-freedom hurdle, eh?)
Registered User
Joined: Aug 2008
Posts: 862
Likes: 0
Originally Posted by magician,Dec 2 2009, 02:40 AM
I was hoping to get him to say all this, not you. Clearly, you know this stuff. It would have been nice to determine if OP had a clue what he was trying to establish. If he's been instructed to use a t-test, he must have been given some guidelines on why he's doing it.
Oh, well.
(At least we didn't tip the degrees-of-freedom hurdle, eh?)
Oh, well.
(At least we didn't tip the degrees-of-freedom hurdle, eh?)
Fortunately for him, degrees of freedom isn't too difficult to understand, and he can get his answer with a simple googling of it.