Diagnosis Questionnaire Proposal

[redcalx]

Hi all, I'm considering creating a questionnaire for helping with self-diagnosis of position on the autism spectrum, similar to the Baron-Cohen questionnaire some of you may have seen but with some improvements

Comments

[mercurygrrl]

http://www.rdos.net/eng/

http://www.rdos.net/eng/aspeval/

[redcalx]

Thanks, that's pretty much what I was hoping to do. I've sent an email to the site owner to see if I can run a CF algorithm on their data. Looking through the site it looks like they have calculated correlation coefficients between each question and whether the respondant has autism, aspergers or ADHD. The sort of CF algorithm I'm thinking of should be even more accurate, it also finds common combinations (or clusters) of answers - if so each cluster would indicate a different type of autism/asperger's perhaps with different underlying causes, e.g. there might be clusters for certain genetic variations, and other clusters for environment induced autism. It won't tell us what the underlying causes are but it will indicate if more than one cluster exists.

[intralimina]

Three questions (not meaning to sound obnoxious, just wearing my scientist/peer reviewer hat and curious):

1. How are you planning to come up with your questions for the instrument? In other words, how are you determining what is / is not a possible indicator of autisticness?

2. How are you justifying the scale of "autisticness?" Like, what measurement are you using for a "aspergers" vs. an "autistic?" This question I'm particularly interested in because of the lack of good measurement or consensus in general around this topic.

3. Have you considered thinking about this instrument in the other direction and using it to ask, "what sorts of things may be an actual indicator of an ASD?" i.e. consider this to be an exploratory study not a confirmatory one?

[redcalx]

Good questions. The technique I'm intending to use is a gradient decent based matrix factorisation. If you imagine a matrix(table) with the questions along the top and the respondents down the side, each square in the matrix represents an answer to a question, this includes the question "Where are you on the autism spectrum". Any unanswered questions are represented by empty squares

[redcalx]

I didn't answer question 2 very well. It may turn out to be rather crude to place people on a single autism axis. The number of vector pairs - and which questions those pairs have high correlation with - will indicate if this is so. It may be that we identify two or more signicant groups of answers and that these hint at multiple autism axes. Thus it may be sensible to simply ask (A) Do you have autism [1=no, 5=yes, with no in between answers] and as a separate question (B) Do you have asperger's. We would of course expect these two questions to be correlated, and by how much will indicate where aspergers is, on average, on the autism axis, or if in fact it gets factored out into its own axis (or some combination of other axes).

[intralimina]

So is is fair to say you're exploring whether there are any differences in how people with one Dx vs. another answer the questions, rather than working off a hypothesis about whether and what these differences might be?

[foundunicorn]

I don't do well with the 1-5. I never can decide if some thing should be a 2 or 3 for example.
I would like it to be: yes-maybe/skipping-no