# Can you compare two different F-ratios from two different one way ANOVA tests?

#### thename1000

##### New Member
I have one data set and two tests:

The first compares aging (0-500 days) per unit in inventory as the dependent variable and product type (type 1-5) as the independent variable. My results are:

Where P1-5 are product 1, product 2, etc.

From this I can see that the sample means vary a good amount, the F ratio of 108.38 is rather large (which is evidence I think that these independent variables have an impact on aging, and at the bottom you can see that P2-P4 and P3-P5 have similar 'aging' times but the rest vary a good amount.

The second comparison is aging (same data as above) vs the independent variable of 10 sales regions.

^the confidence interval tests is cut off in the screenshot.

My main questions are:

1.) For the second test, the F ratio of 26.4 and the low p value seems to indicate that the aging is influenced somewhat by what sales region it is in, right?

2.) The first test of product vs aging has a F ratio of 108.38, the second test of sales region vs aging has a F ratio of 26.4. Am I able to say that (since both have the same dependent variable), the larger f ratio of product has MORE of an impact on aging than sales region, although both clearly impact aging?

Thanks so much! If you have any other random comments on how to interpret my data please let me know.

FYI my data looks like this, where each row is a unit in inventory:

But to do the ANOVA, I separated it into two tables where one is product + aging, and the other is sales region + aging.

#### hlsmith

##### Less is more. Stay pure. Stay poor.
TLR - are the models nested?

#### Miner

##### TS Contributor
You would be better off running a 2-way ANOVA on product type, sales region and the interaction between the two. Then you can check the relative contribution of each factor.

#### thename1000

##### New Member
Thanks everyone.

1.) Please forgive me but I don't know what nested means in this context. To be clear the data all together looks like this:

But to run the ANOVA in StatTools I had to convert it look like this (and the same thing for regions): which I got from this youtube video (my data is the same before the conversion as his):

2.) I can't figure out how to run a 2 way ANOVA with this because the data is not balanced (I get a 'balanced experiment error'). There are a different number of products (more of product 1 than product 2) and a different number of region (more of region 1 than region 2).

But in a general sense, are the two different F ratios comparable to each other?

#### Karabiner

##### TS Contributor
I guess that balanced experiment error perhaps means that in some regions some products are missing. Can that be the case?

#### Miner

##### TS Contributor
2.) I can't figure out how to run a 2 way ANOVA with this because the data is not balanced (I get a 'balanced experiment error'). There are a different number of products (more of product 1 than product 2) and a different number of region (more of region 1 than region 2).

But in a general sense, are the two different F ratios comparable to each other?
You may be able to run a general linear model (GLM) to get around the balanced experiment error. Comparing F-ratios may give you a general idea of relative strength, but is not a good approach.