I have measured variables of two groups of age, divided in three different housing conditions (2X3 Factors), through 3 days of tests. I have realized a RP-MANOVA with SPSS, but I cant find the right way to obtain the p values for a DAY*AGE*HOUSING interaction. For report the results, are the confidence intervals enough?
Do you have replications ? if not, it is impossible to have the 3way P.
Are you using GLM or the old manova ?
Send me the spss dataset at [email protected] (putting fictitious data in place of your results) and the sintax, I will look at the problem in next days.
Confidence intervals for effect sizes beat p values every time. What do p values really mean? Check http://einarbthorsteinsson.blogspot.com.au/2010/09/publish-or-perish.html
Well your first problem is that you don't have a 2x3 but perhaps a2(age)x3(housing)x3(days). If you enter the data into the GLM repeated measures analysis the MANOVA table generated will give you the p values. If you don't have 18 columns of data required for a 2x3x3 then we need to know which of these variables is not actually a repeated measure (my guess is days). If this is a "between subjects" factor (ie you have a column coded day 1,2,3) then it needs to be entered as such in the "repeated Measures" window in SPSS.
Of course Einar is right, the confidence intervals are a more appropriate way to go.
Perhaps unavoidable, but why is the age measure dichotomized? Aren't you potentially discarding valuable information in a multiple regression approach?
Walter, this is because we are using rats, and the homogeneity of the groups allows us to establish this dichotomization based in the time of the birth of all animals respect the time of the exposition to the conditions. One of them are being exposed in an early stage of the development, and the other group has being exposed in a more advance stage. So, the variable age reflects it, and not precisely the real age of the animal.
Ya lo comprendo; se puede classificar o llamar el variable/concepto: edad o nivel de desarrollo a la exposicion...me parece interesante su estudio.
A! Gracias! Nuestro grupo va a presentar los resultados de esta investigación en el Annual Meeting of Society for Neuroscience en San Diego, Ca, en noviembre de este año. Estamos utilizando este estudio para comparar los efectos entre grupos y a lo largo del tiempo en variables de memoria espacial y memoria de trabajo espacial.
- Badges
- Science topic