learned from my eye doctor

[cellio]

Another thing I learned today: those vision tests where you read letters off a chart on the wall? Put a great big "SWAG" around any results you get from that. Lighting in the room, brightness of the bulb in the projector, focus, and a bunch of other factors can affect your score -- and these all vary from one exam room to another. Bugger! Here I thought it meant something when I did, or didn't, get the same results on successive visits!

Oh, and some letters are harder than others -- a fact I have noted before -- and sometimes the person administering the test will fudge your score based on that. So if you got the "Q" (which looks an awful lot like an "O") that might count for more than correctly getting the "L" (which is pretty unambiguous). Or not. It depends on the technician, or maybe on the phase of the moon. This, at least, is something where any given practice could set policies; I'll have to ask about that.

(Why yes, I do read my chart while waiting for the doctor. I think this is perfectly appropriate.)

Comments

[estherchaya.livejournal.com]

I had an opthamologist for years who eventually figured out that he needed to use the randomized eye charts (which didn't exist when I was a kid) because I had them memorized and I found it difficult to figure out if I was really *seeing* "FZBDE" or if I just knew those were the letters so it worked. FZBDE is my favorite line. Closely followed by AEOPTF.

Er. I'm a nerd, aren't I?

[cellio]

I remember having to repeatedly explain to one person that for a valid test, he had to either change charts or do the eyes in the order I said -- because if I read the chart with the better eye I would remember that data long enough to not be sure if I was really seeing it with the weaker eye.

eye charts

[cvirtue.livejournal.com]

it's the lunar tides!

[(Anonymous)]

How many points do they give you for being able to read the chart properly? :)


*snicker*

- Inkhorn

[cellio]

The scoring system is a little opaque. Here's how it works, as much as I've figured out: each line on the chart corresponds to a base value -- 20/20, 20/30, 20/40, and so on up to the really big E, which I think is 20/200. (Miss that one and you're legally blind.) Most people don't read exactly N lines, so you'll end up with something where, say, you read everything on the 20/40 line correctly and you also got one from the 20/30 line. They will note this as "20/40 + 1", usually. Or if you got everything on the 20/40 line except one, that'll be "20/40 - 1". Fine so far.

I've never seen numbers greater than 2 in the +/- even though smaller lines contain 5 or 6 (or maybe more, but I'm done by then) letters. And what happens if you missed one on the 20/40 line but got two right on the 20/30 line? They might note it as "20/40 + 1", averaging it out, or they might adjust the number based on what the missed letter was -- maybe they cut you slack for a Q ("we'll give him 20/40 + 2" but not for an L ("he should have gotten that so we'll say 20/40").

Now none of this matters until you have to apply the numbers. The cutoff for a driver's license, for example, is something like 20/60 -- do they really mean 20/60, or is "20/60 plus one but minus one" ok? The corner cases matter, and the algorithm is insufficiently specified. And what about areas where you really want high visual accuity, like pilots' licenses? (I think you have to hit 20/20 there.) There will be a lot more people with borderline scores there, I suspect, and there's no consistent way to resolve them.

It doesn't matter for me personally, but nondeterministic systems bother me on principle. :-)