Contingency Table questions

RwM

New Member
#1
I have a large data set, n=2556. I have two Types, Natural (n=720), Treatment (n=1836) and 3 levels, "Not Present", >0 & <20, and >20.

What I'm trying to determine: is there a significant difference between the Types?

I determined the row frequencies, and set up a 2 x 3 contingency table.
G test 18.86, df 2, p-value 0.00008. Reject the Null hypothesis based on the p-value. (No relationship between Group type and classification level)

(I have never worked with contingency tables before, so any pointers gladly received!)

Now to the questions: Does it matter that there is such a large difference in the number of species by group?

Do I also need to determine phi?
 

vinux

Dark Knight
#2
I have a large data set, n=2556. I have two Types, Natural (n=720), Treatment (n=1836) and 3 levels, "Not Present", >0 & <20, and >20.

What I'm trying to determine: is there a significant difference between the Types?

I determined the row frequencies, and set up a 2 x 3 contingency table.
G test 18.86, df 2, p-value 0.00008. Reject the Null hypothesis based on the p-value. (No relationship between Group type and classification level)

(I have never worked with contingency tables before, so any pointers gladly received!)

Now to the questions: Does it matter that there is such a large difference in the number of species by group?

Do I also need to determine phi?
In your case it doesn't matter that the difference in the number of species by group.
I guess you were testing the proportion of the levels are same. You could use the terminology like "test for homogeneity (Check this link to verify your understanding http://stattrek.com/AP-Statistics-4/Homogeneity.aspx?Tutorial=Stat ; it doesn't change your result)

I guess finally you concluded that there is a relation exists between group and levels.