|
Hi All: Can someone ask me this simple question: If the assumption of normality between my dependent variables is not met, which is the analog counterpart of a Repeated Measures ANOVA (RMA)?. Or this test is strong enough to overcome that stump?.
Thanks in advance...
-G-
-- |
|
Administrator
|
I can virtually guarantee that your dependent variable is not normally distributed. As George Box said, “…the statistician knows…that in nature there never was a normal distribution, there never was a straight line, yet with normal and linear assumptions, known to be false, he can often derive results which match, to a useful approximation, those found in the real world.” (JASA, 1976, Vol. 71, 791-799) For others to judge whether the approximation is good enough to be useful, you need to give more information about your dependent variable. Is the shape of its distribution similar at each level of the repeated measures factor? Are means and SDs are useful and trustworthy as descriptive statistics? What is the sample size?
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
Gerónimo, Bruce and other friends,
He could also think about the magnitude of the violation? Need to be thinking about how the distribution is not normal. Normality is not a feature of the distribution dichotomy, but a gradient. best regards Luciano -- ______________________ Luciano Basso Lab of Motor Behaviour - LACOM EEFE-USP-Brazil ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
|
In reply to this post by Bruce Weaver
Dependent variable: Time in minutes
Factor: 3 treatments in 4 different doses
Sample: 4 mouses per treatment/dose
I've been reading and found that RMA are strong enough to withstand the hit of non-normallity among the dependent variable.
Thanks! -- |
|
Administrator
|
If that is time to complete some task, it is quite possible to have some very long times (e.g., if an animal gets distracted from the task). In his chapter on transformations, Dave Howell (Statistical Methods for Psychology) gives an example quite similar to this, and suggests that using the reciprocal of the times (i.e., the speed) can be advantageous. His example: Times: 10 11 13 14 15 45 450 Reciprocals: .100 .091 .077 .071 .067 .022 .002 He then observes that "the differences among the longer times are much reduced from what they were in the original units", and that "the outliers will have considerably less effect on the size of the standard deviation than they had before the transformation." HTH.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
In reply to this post by Gerónimo Maldonado
The question is whether or not the residuals show a pattern such that their distribution looks definitely non-normal. Art Kendall Social Research Consultants On 10/18/2010 11:30 AM, Gerónimo Maldonado wrote: ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
Art Kendall
Social Research Consultants |
| Free forum by Nabble | Edit this page |
