Random vs fixed effects in ANOVA

[thrillhouse86]

I am having a lot of trouble conceptually understanding the idea of a random effect in ANOVA testing - more specifically identifying whether a factor is random or fixed

Thanks,
Thrillhouse86

 

[diggy]

Say you were interested in comparing the densities between different foods, and you took samples from a bunch of twizzlers, slices of bread, and banana splits.

The measured difference between the three groups is a fixed effect.

Now each subgroup will have its own random effect. Twizzlers are all about the same, so the variation in them will be small. Your breads may be a little different, but should be pretty close to the same, so you will have some variation within the breads. Likewise the banana split could have large variations, depending on who made it, and how much they like whip cream.

I think the general idea is that *between* different variables is "fixed", and *within* a type/factor is "random".

The scope of your analysis decides also determines what is fixed and what is random. If you are interested in comparing the group dessert to the group sandwich, then each dessert is now a random effect inside of the fixed dessert category.

I'm sure there is someone who can explain it better, but that's more or less how I think of it.

 

[thrillhouse86]

hmmm ... is it as simple as:
'between subject factors' = fixed variable
'within subject factors' = random variable
?