Relationship investigation with large dataset
- From: "amin" <ruzbahan@xxxxxxxxx>
- Date: 9 Aug 2005 12:54:11 -0700
Hi All,
I want to investigate the relationship between 6 independent variables
and one dependent variable.
I have designed some experiments to collect data points for
relationship investigation.
Values of these 6 variables are determined by counting some features in
each subject.
I can say that independent variables are nominal but I can't argue that
they are interval.
The variables have many levels, say more than 10. The number of
collected data points is about 680.
This number is large I think, and the result of every hypothesis test
is highly significant.
I'm thinking about dividing data points into 30 random sets containing
30 data points while these sets may have common data
points. In each set I can use multivariate ANOVA (I dont know if it is
usable for my variables because I cant argue that they
are of interval scale.).
Each ANOVA test has 6 p-values, say ANOVA_P_VALUE (I have 6 independent
variables and one dependent one).
The p-values of 30 multivariate ANOVA tests are used in a T-test to
investigate wether the p-values of ANOVA's are less than
0.05 or not. That is, in the T-test I want to know if
ANOVA_P_VALUE<0.05.
1) Can I use ANOVA, regarding that I can't say the scales are interval?
2) Is this method reasonable? (splitting large data set into small
subsets). Is is Cross-Validation?
3) Is there any better solution to my problem?
I'm not a statistics man, so I apologize if I can't use terms properly.
Thanks in advance,
--Amin (ruzbahan@xxxxxxxxx)
.
- Follow-Ups:
- Re: Relationship investigation with large dataset
- From: Richard Ulrich
- Re: Relationship investigation with large dataset
- From: Phil Sherrod
- Re: Relationship investigation with large dataset
- Prev by Date: Sampling (???) Statistics
- Next by Date: Re: Relationship investigation with large dataset
- Previous by thread: Sampling (???) Statistics
- Next by thread: Re: Relationship investigation with large dataset
- Index(es):
Relevant Pages
|