A while back (2013) I was feeling frustrated and generated a questionnaire about autistic symptoms from the perspective of an autistic person. I called it the Communication and Coordination Questionnaire (CCQ). I posted one version on my website, got some data (mostly people from Wrong Planet), posted a revised version, left it for two years, and decided to take it down this June while cleaning house.
Both questionnaires were 15 questions, in both binary (yes/no) and 5-point Likert scale (1-disagree strongly - 5-agree strongly) versions (since I couldn't decide which was better).
Here are the two versions:
I only got one non-autistic person for the first version.
I'm too lazy to do an in depth analysis. Have some charts instead.
Version 1.0, binary (yes/no)
Version 1.0, Likert scale (1-disagree strongly - 5-agree strongly)
Version 1.1, binary (yes/no)
Version 1.1, Likert scale (0-disagree strongly - 4-agree strongly)
And here are the combined results for the questions that were the same in both versions:
Combined results, binary (yes/no):
Combined results, Likert scale (0-disagree strongly - 4-agree strongly):
To do a scale properly, first you need to pick a bunch of questions that address the issue at hand. Then you need to try them out on people. Then you do a factor analysis to see how many different concepts you are measuring, and how they relate to each other. For example, say you are measuring three different personality traits in one questionnaire. Items need to load on only one factor (rather than two or more, or none) - you don't want ambiguous questions, you want questions that separate the wheat from the chaff. Once you've done this, you weed out the bad questions, add some more, retest, reanalyze. Over and over again if need be to get a scale that measures what you want to measure, and not what you don't. Scales that were developed this way: the EPQ-R, various versions of the big five personality inventory, the Interpersonal Reactivity Index. Scales that were not developed this way: The Hartmann Boundary Questionnaire (Hartmann got as far as the first factor analysis and didn't go any further for some reason), the SQ, EQ and AQ (they've been analyzed but as far as I know nothing's been weeded out because it loaded on the wrong thing, even though the EQ contains some questions that appear to be more about the ability to communicate well than about empathy).
I was originally going to do all this but have lost the impetus. However I could gear back up again if there were a demand.
I haven't checked for sex or age differences. I figure I don't have enough data for that. But I could do that.
Should I do more work on this?
If I were going to develop this further, I'd add questions on schizoid and schizotypal personality disorders to see if I could distinguish them from autism (since I think they get lumped in with ASDs these days). Also, there's recent research showing that late talkers and regular talkers are different between the ears, and questions to separate those groups might be useful, too. Any other factors to include?
To what extent would it be important to separate out subtypes, and to what extent would it be better to have a general autism scale? (Are there good scales out there that people can do that actually look at at disability and not just autistic-like traits?)
How long or short should it be?
Binary or Likert scale?
I posted this, more or less, on Wrong Planet, and one person answered the poll saying I should develop this further. No one commented. I'm not going to do anything more for now, but may be inspired down the road.