# How to calculate relative keystroke proficiency

Discussion in 'Physics & Math' started by Jennifer Murphy, Mar 20, 2017.

1. ### Jennifer MurphyRegistered Senior Member

Messages:
176
It seems to me that the true "proficiency" of each key that a person types on a keyboard is not accurately measured by just the percentage of correct keystrokes. I can guarantee close to 100% accuracy by taking as much time as I need to find the right key. The standard typing test practice of subtracting errors from overall words per minute (WPM) is also a very crude measure. It seems to me that to get an accurate proficiency rating for each key, I would need the percentage correct and the time interval from the previous keystroke.

Suppose I have a program that will give me a string of text to type and will record both the accuracy of each keystroke and the time interval from the previous keystroke (in milliseconds). And suppose I get this data for 6 of the keys:
Code:
C/R     C          D    E    F    G    H    I
5  Key            1    2    3    4    5    6
6  %Correct     0.99 0.99 0.95 0.95 0.90 0.90
7  Interval (ms) 200  300  200  300  200  300
8  WPM            60   40   60   40   60   40
9  Proficiency   ???  ???  ???  ???  ???  ???

How do I combine rows 7 and 8 to get a relative proficiency for each key (row 9)?

Is key #3 (Col F) with an average accuracy of 0.95 and an average interval of 200 ms (60 WPM) more or less proficient that key #2 (Col E) with a higher proficiency (0.99) but a slower speed (300 ms = 40 WPM)?

Any suggestions?

to hide all adverts.
3. ### DaveC426913Valued Senior Member

Messages:
8,213
I don't understand your logic here. Of course you can get 100% if you take your time. Which is why WPM is the second factor in calc'ing proficiency.

Since the only criteria are accuracy and overall speed, I don't see how/why current tests would be considered crude. And therefore don't follow why you see a need to change the criteria.

to hide all adverts.
5. ### Jennifer MurphyRegistered Senior Member

Messages:
176
I see your point, I think. I guess it depends on how WPM is calculated. In the example I provided, the WPM row was calculated directly from the average time intervals.
Code:
wpm = 1200 / interval
So it did not take into account the accuracy, just the raw speed.

to hide all adverts.
7. ### DaveC426913Valued Senior Member

Messages:
8,213
How can you have a WPM calculated on a single key? WPM is, by definition, an aggregate value of the strike times of disparate keys.

8. ### Jennifer MurphyRegistered Senior Member

Messages:
176
I suppose, if you are limited to a pedantic, absolutist definition, then you can't. But where do you draw the line? If a text sample never uses the Z key, does my result not qualify for a WPM rating? What if it only uses 20 letters or 15 or 10? What's the difference?

Typing programs routinely calculate WPM ratings on even very early drills when only 2 keys, f & j, have been introduced. A typing student, after typing a string like, "fjf jfj ffj j jf j jfjf ffjf..." will get a WPM rating. It's a handy way to tell the user how they are doing using a metric that is well understood. The program could go into some arcane discussion about characters/second (CPS), but WPM is automatically understood and even if it applies to just 2 letters, it gives the user a sense of how they are doing in relation to their goal of learning to type 40 WPM.

Anyway, why do you have your knickers in a twist about this trivia? Could it be that you don't have an answer to the question I actually asked?

9. ### exchemistValued Senior Member

Messages:
7,055
I'm not an expert on this, but since you have put it in the public domain for comment I will try to address it.

It doesn't look like trivia to me either. I cannot see the value in attempting to define a proficiency for hitting an individual key, when the thing that limits proficiency is the proportion of mis-hits, which naturally tends to rise as a function of the speed of typing.

If you focus on some notional concept of proficiency of hitting a single key, how do you assess the proportion of mis-hits? I suppose you can do it if you use a standardised piece of text and count the number of times each letter was supposed to be hit but was not, or was hit when it was not supposed to be. Every time you get one of those, you must get another at the same time, on whatever key was hit by mistake. I do not see what advantage this has: it just seems to me to cause confusion, compared to simple assessments of total mis-hits and speed.

I can see it is useful to provide a breakdown of which keys are most commonly mis-hit, to help the typist work out where they need to improve, but trying to define a "proficiency" at the level of individuals keys looks to me like a wild goose chase.

But now, looking at your exchange with Dave, I see I have fallen into the same trap, which is the one of the Irishman saying, "If I were you, I would not start from here.", rather than answering your question. So I'll get out of this thread. Sorry for the intrusion.

Please Register or Log in to view the hidden image!

Last edited: Mar 26, 2017
10. ### DaveC426913Valued Senior Member

Messages:
8,213
Well, no I don't have an answer, because I can't figure out the rationale. I don't consider it trivial; it's kind of the crux.

But, granted, you didn't ask for advice on your logic, so I'll recuse myself.

11. ### DaveC426913Valued Senior Member

Messages:
8,213
I like that. That'll come in very handy here on SciFo.

Anyway, sorry Jennifer that I couldn't help. Best of luck.