I’m making a score sheet to rank applicants based on 5 characteristics (from 1-4)
Is it inherently better, to make the ‘best’ score a 1, or is it better to make the ‘best’ score 4?
Or does it really not matter?
If it matters, can you explain why one system is better than the other, so I can articulate it well to my colleagues?
My first thought is that if you make 1=worst 4=best and then something comes along which exceeds all your expectations you can say “Wow, this one deserves a 5!” and then when something even BETTER shows up you can say “OMG this one is a 6!”. But if you make 1=best and 4=worse then then you end up saying “Wow, this one deserves a 0!” followed by “OMG this one is a -1!”. IMHO, the former makes sense and the latter doesn’t.
Having the best score be a 4 rather than a 1 is better just because most people intuitively read a higher number as being the better one on such scales. It makes scoring easier and less error prone. It also makes it easier for people that only causally glance at your results to interpret them correctly.
Right. Just like in golf!
Mathematically, there is no difference whatsoever. In every other way, there is no difference whatsoever.
Not necessarily. We also say this one ranks first or number one, this one is number two.
Is this for ranking people, or for scoring them? That is to say, is each person scored individually such that different people could get the same number, or are they compared to each other to see who’s the best, second-best, and so on? If it’s a score, then most people probably expect the high numbers to be good, and if it’s a rank, then most people probably expect the low numbers to be good.
Moderator Action
While you may be able to find some factual cites for this sort of thing (maybe a psychological study about which way people prefer rankings or some such), I think that this is probably going to end up more a matter of opinion than anything else.
Moving thread from General Questions to In My Humble Opinion.
One thing to note is that you should always have an even number of possible scores. It’s too easy to just choose average if there are an odd number. With an even number of choices you have to decide if the person is on the high side of average or the low side.
You could always make the best a 2.5, then treat it like a bell curve.
I’d say rate low to high, but rank, high to low. In my experience, that seems to be both the most intuitive and the most functional way of doing it.
And for those confused by what I mean, I just mean for a given criterion, if someone is bad at it, give a rating of 1, if someone is good, give a rating of 4. Then, after rating all the criteria, apply a weighting value to each one, sum them up to get a cumulative score, and then the give the highest cumulative score a rank of 1, and the lowest a rank of 4.
With that, obviously you’ll have as many ranks as candidates, but for ratings, that’ll take some consideration. If there’s not enough, you may not have the resolution you need, but if there’s too many, you end up making meaningless distinctions. Generally, I think a rating scale of 5 is ideal, with the understanding that not being a 5 isn’t bad, it just means well above average, where 3 is average or satisfactory. 4 would also work, but I don’t see much reason in deliberately excluding an average/satisfactory rating.