Is it possible or meaningful to run an ANOVA on only two groups?

#1
Hi,
we have been given a bunch of homework questions and i can't seem to work this one out.

"Is it possible or meaningful to run an ANOVA on only two groups?"

I've tried to google it numerous times but nothing good that is directly targeted against this pops up. I understand that some might be hesitant to reply when its regarding someones homework, but i just need a helping hand pointing me in the right direction, maybe some cues or sharing of ideas.

Thanks beforehand!
 
#2
Make up some data in two groups . Try an anova, and a t test with equal variance. Look at the p values.
As an extra, square the t value and compare with the F value.
 

noetsi

Fortran must die
#3
No if its a dependent variable. Two groups I assume refers to a binary dependent variable (where the variable has only two levels such as yes or no to some question or a gender with male or female). ANOVA was created for interval level dependent variables. If by two groups you mean a predictor variable with two levels you can use dummy variables. Look up assumptions of ANOVA to see a discussion of this, and in particular the scale of the dependent variable.

This assume two groups refers to what I suggested. It could mean lots of things.
 

Dragan

Super Moderator
#4
It really does not matter - it is still an Analysis of Variance, regardless. More broadly, it can be mathemically shown that a t-test is subsumbed as an Analysis of Varirace; which is sumbsumbed under Canonical Correlation....In short, it' all correlation analysis.