Hi, this is a good question to discuss with the colleagues !!!
I think a good idea to start is to test the ANOVA's assumptions:
1 - The normal distribution using the Shapiro-Wilk test and
2 - The homogeneity of variances using the Levene test.
If you have some treatments with such a small sample size
probably the best idea should be to use the non parametric
ANOVA called Kruskal-Wallis test.
Good luck and contact me if you need any help, OK ?
Voltolini
_____________________________________________
Prof. J. C. VOLTOLINI
Grupo de Estudos em Ecologia de Mamiferos - ECOMAM
Universidade de Taubate (UNITAU)
Departamento de Biologia
Taubate, SP, Brasil. CEP 12030-010
Tel: 0XX12-2254165 (lab.), 2254277 (secret. depto.)
FAX: 12 - 2322947
E-Mail: [log in to unmask]
_____________________________________________
----- Original Message -----
From: Galina Glazko <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, November 27, 2001 5:06 PM
Subject: ANOVA sample size
Previously I never did ANOVA analysis and now I am
confused with it. Could someone help me with advise about
individual groups (factor levels) sample size in ANOVA?
My data set contains 73 items, but some factor levels contain
only 3 observation. Is it justified - to conduct such an analysis?
What is the rule here? Thank you in advance Galina
|