Posted by genebujold on October 11, 2003, at 22:03:51
In reply to Re: All seriousness aside.......Caution Advised, posted by yabba on October 11, 2003, at 13:47:33
> An interesting thing that I noticed was that when you look at the colored graph you can see that although there is a bunch of red women's dots on the right side in the women's area but there is also red dots over the entire graph even though the blue men's dots are mostly on the left except for a couple of small clusters. Not that the test is exactly all that scientific though.
I think it's terrific that you noticed the spread. While the majority of respondants clear fall into one category or another (bimodal response), you can see that there is indeed variation in the responses.
Although I am a man, and feel like a man, I also feel I have a strong nurturing side, a fact which others have told me as well (along with other comments such as "you know, you scare me - you think just like my wife").
It's not surprising then that if the results were plotted on a 1 (man) to 10 (woman) scale, I scored at the 7 level (more feminine in my thinking than a man).
Still, I AM a man, hence the following analytical comments:
Actually, it's quite scientific. If they'd asked 100 questions chosen at random from a variety of different tests, and 20 of them produced significant results with the ability to identify the sex of the individual, then the test was revised to include just those 20 questions and the results plotted over some 15 Million respondants, you'd have a very good test!
Which is precisely what they did.
In fact, this is pretty much how all predictive analysis tests are created (Strong-Campbell Interest Survey, etc.), through the generation of numerous questions, and the weeding out of questions that produce little correlation in the responses (and subsequent ability to measure whatever they're trying to measure).
Obviously, if you're looking to measure just two modalities (man vs woman), you don't need a very long test. One question that would certainly be a strong predictor would be "What's your bra size: A) Flat as a board, B) A-cup, C) B-Cup, D) C-cup or larger). Correllating that with the individual's age would make it a strong predictor for everyone over the age of 18.
The fact that the questions they chose are somewhat off the wall merely reflects the fact that they're not a government-funded research organization, and can have a little fun with the test!
Which was the whole point of why I suggested it in the first place, so...
poster:genebujold
thread:54234
URL: http://www.dr-bob.org/babble/20031010/msgs/268496.html