Firstly, apologies for an uninteresting title, and whether this is absolutely newbie maths. I'm a programmer, and when it comes to maths and me - the wheel is going, but the hamster is dead.
On To the question. We use a third-party review site for our products. I aggregate this data through their API. Customers can rate our products as 1, 2, 4, or 5. They don't do the number "3" (this is a concept of the 3rd-party independent reviewing company, I did not introduce this).
If I average these scores using the normal total over count method, is it skewed by the fact that there is no, and never well be, a rating of 3?
Once again, apologies if this is an idiotic, base-level, GCSE/math-101 question, but I'd like to be sure.
This does make a difference, I do not know how these scores come into play, is it the higher the better? or is $3$ effectively like $0$? as in $4,5$ are $+1,+2$ "points" and $1,2$ are $-2,-1$ "points"?
Either way, if you have a customer that is indifferent, i.e. they would pick $3$, then they may do one of the following:
Pick $2$
Pick $4$
No review.
The first two are bad in the "points" scenario, as you are not getting an accurate representation, since people are choosing on a whim.
But not picking at all in the points scenario is fine, since it is effectively neutral.
If it is the higher the better, then all of these cases are bad, as you either get an inaccurate representation, by losing $3$'s or by getting $4$ when you should get $3$, and of course getting $2$ when you should get $3$.
But overall, if your sample sizes are large enough, at least in the points method, there will not be much variation.