Which statistical method can be used to find the relationships between independent and dependent variables instead of regression? And their advantages over regression. It should be mentioned I am not able to do regression due to less number of observations that I have 5 independent variables 3 categorical and two continuous.
$\begingroup$
$\endgroup$
2
-
$\begingroup$ re: "I am not able to do regression due to less number of observations that I have 5 independent variables": does this mean you have more predictors than you have samples? So, your sample size is less than $5$? $\endgroup$– MacroJun 2, 2012 at 15:45
-
$\begingroup$ @macro: my observation is 21 and my variables are 5.my complete question is my previous question stats.stackexchange.com/questions/29612/… $\endgroup$– roseJun 3, 2012 at 2:37
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
In that case, the best approach is to do a linear model with the 5 variables. If you know R, you put them in a data.frame that you call data, the "independent" variable in the first column and the other ones after, the individual observation in each line and you do:
anova(lm(data))
To me that's the best use you can make of your precious data points. If you have too few, no statistical method will save you.
