nemorosa wrote:
Silly questions like:
"I have difficulty figuring out cultural and social rules."
and
"I don't realise when other people are bored by a conversation."
If you have problems in these areas you are unlikely to see it and therefore can't answer the questions.
It's a good point - but on the flip side, it's almost impossible to find observation-based signs that someone has trouble with cultural rules or conversational cues. I'm hoping that enough people have a rough idea when things are a bit out of synch that the questions have worth, and that those who don't will be picked up by other questions in the test.
Just another limitation of multiple choice, unfortunately...
Zur-Darkstar wrote:
I like that test. I have most of these issues. To be more scientifically valid, it needs some control questions that a person with AS would probably answer "no" to.
Good idea - I'll try to do that a bit more in version 2.
cyberdad wrote:
Zak....One small suggestion...try and develop a more even spread of weighted averages across more indicators so you are not skewing the scores on specific traits. Otherwise you are weighting the Likert scores on specific traits > others.
That may explain the cluster of 72s....
Some of the weighting in the test is actually deliberate - some of these traits would be much more likely to be possessed by an autistic person than an NT, so only a low score is needed to "pass". Other traits could more easily apply to NT's, so a very high score is needed to pass. In that sense it's a bit different to the philosphy of target-based data mining (such as business KPI's).
The 72's are only really a problem if non-AS people can obtain them - if it's all AS people getting the high scores, that means the test is working...
bigdango wrote:
I did the test and pressed skip when it wanted to publish to my profile.
10 minutes later i get a msg - 'ahh so you have ass burgers?'
Moral of the story - never trust a facebook app!
Unfortunately whenever you use any app on facebook, your status messages will say "<person> has started using <appname>" - so even though the results don't get published when you press skip, someone looking through your status messages might see that you've taken the test and draw their own conclusions...
I don't really know a way around this one.