As mentioned in 1, 2, 3... not the best place to start? (Part 2A), I decided to make a basic number sense tester. It's inspired by the one found here - but with a few things I thought should be changed.
So without further ado, here it is. Note that pressing Space Bar (when you're not in the middle of a test) lets you fiddle around with some options if you want to.
A few basic modifications I made...
(1) more dots. The other number sense tester could only go so far; this one can show up to 500 of each colour, allowing extremely small ratios to be tested. If you want to limit yourself to 20 anyways, press space bar and edit the "max. number of dots of each colour" field. For most purposes I've found you don't really need more than 100.
(2) different background colour. The original uses #666666, or 40% gray (from 0% black to 100% white). But with this, the yellow stands out too much - some very basic testing showed that of all my incorrect answers, about 80% would be because I saw too much yellow. Conversely, if the background were white, the blue would stand out too much - 75% of my incorrect answers were because of seeing too much blue. The colour I ended up choosing - #AAAAAA (67% gray) was the best I could find to show blue and yellow fairly equally.
(3) ratios. Simple - just divide blue dots by yellow dots (or the other way around, depending on which has more)
(4) calculating skill level. This was the hardest part - to find a way to take all the results and calculate about where your skill level is. Much thanks to the Mathcamp alumni mailing list for helping me figure out how to go about calculating that.
(5) providing tests based on skill level. Whatever your skill level is, you'll be given a couple questions just above it and a couple just below it, so you're always at a place you're comfortable with.
(6) options. Press space bar to change things... maybe you'll discover something by tweaking the options a bit. Can you get a better score if you have fewer (or more) dots? Does the relative size of the dots make a difference? Try it out for yourself.
This was mostly just for fun, so chances are I won't come out with an updated version, even if I do find ways to make it better... but if you have any comments or suggestions, I'd love to hear them!
This is cool! One suggestion I'd make is to have the viewing screen be smaller; because the observation time is so short, I only have time to look at a small portion of the screen, which means that I am very susceptible to times when the distribution is even moderately far from uniform. Also, (though this may be a personal thing), I still find that the yellow dots stick out more for me than the blue. Maybe make the blue lighter rather than the background? Anyway, this is pretty cool. Good job! --Yuval
ReplyDeletehaha, there are actually soooo many variables in this that could affect the result... background colour, viewing size, circle size, circle variation, number of circles, length of time displayed... and I just keep discovering more. Who knows, maybe people can tell red and green apart better than blue and yellow. To make this test really rigorous, you'd have to do tons of tests measuring the effect of each variable. Thanks for the suggestions though - I'll keep those in mind in case I do upgrade it!
ReplyDelete