stats help please - f ratio

[OlliePsych]

---

Members don't see this ad.

I'd like to see if anyone can offer some help with this stats question. One of my professors told me that I needed to do an f ratio, but I'm a little confused about how to go about this. Essentially, I am trying to determine if there is greater variance in one group of people on a select variable than the other group. The project is a little removed in time from when started (and the talk of the f ratio) so I'm struggling to determine how to do the f ratio. This will probably be a duh moment once someone explains, but I've done a google search and haven't really found what I'm looking for.

Thanks!

 

[edieb]

I'd like to see if anyone can offer some help with this stats question. One of my professors told me that I needed to do an f ratio, but I'm a little confused about how to go about this. Essentially, I am trying to determine if there is greater variance in one group of people on a select variable than the other group. The project is a little removed in time from when started (and the talk of the f ratio) so I'm struggling to determine how to do the f ratio. This will probably be a duh moment once someone explains, but I've done a google search and haven't really found what I'm looking for.

Thanks!

An F-Ratio compares the Mean of Squares between your groups (e.g., treatment variance) over the Mean of Squares withing your groups (e.g., within groups variance). If your treatment/Independent variable produces an effect, the MSB will be greater than the Mean Square within and your F Ratio (Mean Square Between/Mean Sq Within) will be > 1.0

 

[ClinicalTrainee]

Are you asking how to do this by hand, or how to do this in a statistical software package? It doesn't sound like you actually want a traditional F-test (as aptly described by edieb), but that you want the homogeneity of variance test that Jon Snow described. The Levene's test is a test of homogeneity, that will tell you if one group has more variability on a DV than another group.

You can ask for the Levene's (or it's friend, the Brown-Forsythe, which is the same concept but uses the median for calculation intsead of the mean) in SPSS using the One-Way command, I believe (in the menu system, Analyze-Compare Means to One-way ANOVA), and selecting the homogeneity of variance box. I don't have SPSS in front of me, so I don't remember if you can also do it from the Univariate menu. Otherwise the wikipedia link has the formula to calculate by hand.