# Thread: figuring out a class evaluation score minus outlier

1. ## figuring out a class evaluation score minus outlier

Hi,
I have a some course evaluation data that I want to do a what-if analysis with (namely, "what would the median have been if one person didn't throw off the average with a ridiculously low score").

There were 5 options on the evaluation form (I don't know how each is weighted). Each of the top 3 grades were chosen by 2 students. The remaining student in the 7-person class chose the bottommost grade.

The median listed in the report is 2.57, with a standard deviation of 1.40.

Can I calculate what the mean would've been without that outlier with any precision even though I don't know weights used?

Thanks.

BB

2. BTW, I realize that the standard deviation is statistical measure, but I'm posting it here because I assume it is fairly trivial mathematically (which is why I hang my head in shame...).

3. Okay, I experimented a bit and discovered that if you weigh the grades whole numbers from 0 to 4, you get a median of 2.57.
So I guess a the more interesting question is, How do you determine how much variation there could be (i.e., if the weights had been different)?
I realize it's very unlikely they used different weights, but for the sake of conversation... Tx.

4. Do you have the actual data set?

If so just change the weights yourself.

5. ## figuring out the weights

Yes, I do. That's what I did.

I guess my 2nd question was kind of silly. I was just trying to think of a way to make my calculation bullet-proof (i.e., noone can nitpick and say, "Well, actually, that's not the weight we used, so your revised median is wrong."), but how much variation could there be in this situation? Grades have to be evenly distributed (if that's the right word) and I'm getting the same media using these weights, ergo I must be very close.

Thanks.

QUOTE=pickslides;615698]Do you have the actual data set?

If so just change the weights yourself.[/QUOTE]