 |
 |
Seaint from my Verizon Wireless Phone ----- Reply message ----- From: "MCLAREN, Donald" <[log in to unmask]> To: <[log in to unmask]> Subject: [SPM] 2-group t-test stat problem Date: Tue, Aug 30, 2011 5:09 pm There is a common misperception that one sample t-tests translate into group differences; however, there are numerous examples of that not being the case. Here are some illustrative examples: Take the case where mean activity in group A is .5 and the T-statistic is 20 and the mean activity in group B is .5 and the T-statistic is 2. For the one-sample T-test you will find the there is highly significant activity in group A, but not group B. There will be no difference at the group level. Take the case where mean activity in group A is 5 and the T-statistic is 2.99 and the mean activity in group B is 1 and the T-statistic is 3.1. For the one-sample T-test you will find the there is significant activity in group B, but not group A for a threshold of T=3. This gives the impresssion that group B is greater than group A. When you do a two-sample t-test, you find out that group A is greater than group B. In summary, you need to be very cautious about inferring group differences from T-statistics from one-sample t-tests. In the absence of a group difference, I'm not sure if you can find a group difference. Best Regards, Donald McLaren ================= D.G. McLaren, Ph.D. Postdoctoral Research Fellow, GRECC, Bedford VA Research Fellow, Department of Neurology, Massachusetts General Hospital and Harvard Medical School Office: (773) 406-2464 ===================== This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is intended only for the use of the individual or entity named above. If the reader of the e-mail is not the intended recipient or the employee or agent responsible for delivering it to the intended recipient, you are hereby notified that you are in possession of confidential and privileged information. Any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you have received this e-mail unintentionally, please immediately notify the sender via telephone at (773) 406-2464 or email. On Tue, Aug 30, 2011 at 4:41 PM, Doug Terry <[log in to unmask]> wrote: Hi Folks, I'm analyzing 2 groups (n=20 for each group) using the 2-group t-test option and finding no differences between the groups at FWE 0.05/FWE 0.10, and very minimal differences at p=0.001 uncorrected. I've looked at each subjects' data to make sure the pre-processed scans are okay, and that the group contrasts are correct. The activation patterns for each group separately (1 group t test) show activation in the hypothesized regions. Visually, the two groups have similar activation patterns, but are by no means "the same", as the clinical group shows clusters of activation at the single group level (FWE 0.05) that are not shown in the control group. Obviously the differences have to be robust/reliable enough to survive the statistical contrast, but I was wondering if there is anything else I can do to elucidate differences between these groups? All suggestions welcome. Thanks, Doug