Autistic Spectrum Components 2.0

Post Reply New Topic

My results:


# Repetitive or restricted Behaviours and Interests (RBI) - Stereotyped, repetitive behaviours and interests
# Social Impairment (SI) - Social understanding
# Language problems (L) - Speech, words and sentences
# Planning, Organization and Concentration problems (POC) - Cognitive skills related to being able to plan, organise and stay focused
# Imaging and Recall problems (IR) - Visualisation, imagination and remembering past events
# Reasoning and Problem solving problems (RP) - Cognitive skills related to rational deduction and working things out
# Sensory problems (S) - Impact of senses
# Motor problems (M) - Control of own movement

RBI =6.5
SI=7.75
L=7.75
POC=6.75
IR=7.5
RP=2.75
S=4.75
M=6

I think this test could do much better if it was based on collected data and not just (mostly) arbitrary groupings of traits. While I do agree that some groupings are sound and consistent (from evidence in Aspie-quiz), others are collections of largely unrelated traits. Before one can put up these groupings, extensive research into trait correlations are needed, unless one believes in the arbitrary groupings in DSM.

Having said that, I really like the concept, and given the minimal data-collection this far, the results are not that bad.

I would like to ask you about the following:

1. Why don't you use the results (and most importantly) correlations in Aspie-quiz for creating groups? You can both study the official evaluation http://www.rdos.net/eng/aspeval and get access to SQL data from my database, provided you give me credit for collecting it if you publish it, and that you don't give it out to others without my permission.

2. Can I use your PHP-code for creating these diagrams for presenting my own groupings in this way in Aspie-quiz?

I wouldn't mind to collaborate with you on this issue.

I must say that a majority of my time on Aspie-quiz (which certainly translates to thousands of hours) have been trying to group traits together based on correlation. I might not have reached the optimal point yet, but at least I think the groupings are quite good at the moment. The group-group correlation report http://www.rdos.net/eng/aspeval/group.htm would also tell a lot about where on the circle to place things. Most related groups should be placed as close togehter as possible. I suppose one can solve this in a objective, matematical way inputing only the correlation matrix.

I agree with you about the grouping and said before I would want to split them up and have as many as 12 individual components. In actual fact on a radar plot it is possible to compare related things without having to formally group them simply by having them side by side and also it is useful to compare 'opposite' traits. The whole geometrical analysis it a key feature of this concept. You don't have to have established theories about groupings, so long as we improve the accuracy and objectivity of the component tests it should show them in the presentation. Someone suggested using statistical qualifiers to train the data. I plan on having open source front ends for distribution. It is important that the data be collated centrally as with other major phych tests.

I think it is a danger zone if we start basing theories on theories, generalisations on generalisations. We need to pick fairly established traits for the components. This will give us a good head start.

I'm happy collaborate, and would prefer if it is many people who are collaborating. It is my goal to contact various experts in the various component fields involved. The most important thing to note is this is not an Aspie test. It is to show where anyone (possibly 'NT') lies on the autistic spectrum (ASD). It comes from the position that the Asperger's, NLD, PDD-NOS, HFA, Kanner's, etc all have arbitrary cut offs. There is will be a separate abilities test and then and function calculation. This all to be deliberated. Function is always going to be an estimate, but I think you agree the cut off 70 IQ points is not a good assessment of function. The abilities test is not an 'autistic abilities' test but a general abilities test. There are several that are in widespread use.

# Repetitive or restricted Behaviours and Interests (RBI) - Stereotyped, repetitive behaviours and interests
# Social Impairment (SI) - Social understanding
# Language problems (L) - Speech, words and sentences
# Planning, Organization and Concentration problems (POC) - Cognitive skills related to being able to plan, organise and stay focused
# Imaging and Recall problems (IR) - Visualisation, imagination and remembering past events
# Reasoning and Problem solving problems (RP) - Cognitive skills related to rational deduction and working things out
# Sensory problems (S) - Impact of senses
# Motor problems (M) - Control of own movement

RBI =8.5
SI=8
L=5.5
POC=8.25
IR=4
RP=7.75
S=4.75
M=6.75

Very high functioning, which is what I suspected. My gut is that a lot of my life has meant that I've got very good coping skills for a lot of this stuff (my mum's an English teacher, for goodnessake, no wonder I've little difficulty expressing myself in written form and have a large vocabulary).



RBI =7.25
SI=6
L=3
POC=5.25
IR=3.75
RP=4.25
S=3
M=4

Yes! Yesyesyesyesyes! Thank you! I've seen so much "we're all so different," and anti-NT stuff on this forum, it's so refreshing to see someone understand that this is a continuous spectrum, reaching the whole way from NT through low functioning Autism. It would be interesting to try and figure out where the cutoffs are, but my gut is we might even be better served by an understanding of the spectrum where there are significant overlaps between each category, where someone could be understood to fit in under more than one bracket. For instance, I feel like I'm definitely AS, but very, very close to NT, my wife's definitely NT but with some AS traits.

The number of individual groups is arbitrary, because there does not exist any natural limits between groups. I suppose an optimation criteria could be to keep groups as clean as possible. I've used this in my groupings.



You cannot have more than two dimensions at a time in a polar plot, which is a shortcoming. If you have 12 groups you cannot compare them all at the same time in a polar plot.



OK, good that you will keep it open source. The Aspie-quiz evaluation is actually also open source and can be downloaded with SVN http://rdos.net/vc/viewvc.cgi/trunk/aspie-quiz/. However, the raw-data is not open source as I want to have a little control over it.



Actually, there is no theory and little subjectivity in Aspie-quiz. The score weights are fully automated and generated by factor-analysis of previous results. The grouping have been manual, but the reports on the groupings (and which question a group belongs to) are based on correlation and thus people's answers.



Despite it's name, neither is Aspie-quiz. Aspie-quiz have evolved towards a neurodiversity test. The original intention was to develop an Aspie-only test, but it soon became apparant that ASDs could not be separated from other conditions like ADD/ADHD, Tourette, dyslexia and much more. OTOH, factor-analysis and score distributions have proved that neurodiversity is indeed a valid concept that has support in the 45,000 answers collected. This is so because factor-analysis always yields two major, largely complementary axises, and score distrubtions show clear signs of two overlapping bell-curves.




Exactly, but ASDs also have arbitrary cut offs against ADHD, TS, dyslexia, dyscalculia, bipolar and at least partly schizophrenia. IOW, ASD is also an arbitrary condition. You need to include all neurodiversity conditions as well to get a sharp cut off.



I think it is very hard to assess IQ online. I've tried, and it was mostly a failure. You need to calibrate you're scores against large IQ test, which I have no idea how to do.

You'd, wish

No, there is not a continous spectrum all the way from ASDs to NT. There is a continious spectrum between various neurodiversity conditions, but NT doesn't lie close to any of these, but rather represents the lack of neurodiversity-traits.

How, then, do we explain the diversity of those on the ASD spectrum who do begin to approach NT? I'm using myself as an example, here. It's possible that my results are simply skewed by upbringing (as I've said on a previous thread somewhere, ask me these questions 10 or 15 years ago, and my Aspie scores would be much, much higher), but it seems like many of these traits, since we're already accepting then can be varying in severity, would logically reach the point where they crossed the line into NT, but still be there. What would we say to someone who is right on the borderline on each of the traits, or is into AS territory on one or two, and a lot closer to "normal" on the other ones? Do we just pretend these people don't exist?

I tried to above. There is not a single border between ASD and NTs, but multiple, like ADHD-NT, AS-NT, Bipolar-NT, dyslexia-NT, and so on. This is also reflected in the radar plots in this thread. NTs will generally score low on ALL dimensions, and this cannot be assigned as an extra dimension.



There do exist an intermediate group in Aspie-quiz, but histograms of score distributions show that there is also a local minima between neurodiversity and NT. That's why I hypothesize that it is the summation of two independent bell-curves, which is the two factors that factor analysis yields. Also, more than 2/3s of all variation in answers could be attributed to these two factors. The remaining 1/3 is mostly environmental. IOW, almost all of human diversity is related either to NT or neurodiversity traits.

On second thoughts, I think it would be far better to present results in two radar plots. One plot would contain "positive" NT traits and the other "positive" Aspie/neurodiversity traits. By "positive" I mean that somebody has some traits (stims, focuses narrowly, have a certain set of communication style or certain talents). I think it is a bad idea to show lack of traits.

Alternatively, one might show NT traits on one side of the radar plot and Aspie/neurodiversity-traits on the other.

Ah, I understand. So rather than my suggestion that someone who might be over the line in one or two aspects would be borderline, we say they're borderline on a more specific disorder? Am I reading that right?

Oh, agreed without question on that score, I think I'm looking at this from a different POV, where we might take these different aspects, plot out total area on the radar, and use that as a way to figure out where someone fits on the overall spectrum. From what you're saying, that's either impractical or simply outright wrong.

Er... I was going to say I'm not convinced, but actually I'm just not convinced what you meant by that.
I do wonder, mostly because of the already low rate of diagnosis of those with Asperger's, whether there are more borderline people out there than is immediately apparent. If we assume the high estimate of those undiagnosed, that's 50% of the Asperger's population, who presumably are getting by to some degree or another (I'm guessing that it's the higher-functioning population who tend not to get a diagnosis). How much moreso, therefore, those who're on the continuum but move into NT territory. Those on the Autism Spectrum amount to less than 1% of the population, so naturally in a random sample of the population, borderline cases will be little more than a statistical blip, and so are unlikely to show up at all in a self-selected group like the ones you'll be getting on online quizzes.

# Repetitive or restricted Behaviours and Interests (RBI) - Stereotyped, repetitive behaviours and interests
# Social Impairment (SI) - Social understanding
# Language problems (L) - Speech, words and sentences
# Planning, Organization and Concentration problems (POC) - Cognitive skills related to being able to plan, organise and stay focused
# Imaging and Recall problems (IR) - Visualisation, imagination and remembering past events
# Reasoning and Problem solving problems (RP) - Cognitive skills related to rational deduction and working things out
# Sensory problems (S) - Impact of senses
# Motor problems (M) - Control of own movement

RBI =9
SI=7.75
L=6.5
POC=9
IR=4.5
RP=5.75
S=5.25
M=4.5

Yes, down to any groupings you might want to do, and regardless if you use DSM. the neurodiversity concept or something else.



No, it is only wrong if you use NT as one of the aspects. You must use the magnitude of different aspects to mean degree of NT. Then it would be correct.



I think 1% is way, way to low. Some indications from the NT-control group in Aspie-quiz stable 2 gives 15% "Aspies", 24% borderline and 61% NTs. Of course, the percentage for the borderline group is arbitrary since it depends on where you put the cutoffs, but if you assign half of it to the Aspie-group, you end up with 27% Aspies. OTOH, people that suspect they are autistic will be more likely to do the test. A better control group would be the Norwegian psychology-student group that were told to do the test as favor. Their average Aspie-score is only 38, about half of that of the usual NT-control group. A fair guess would be that a random sample would yield at least 10% and possibly as much as 25% with higher Aspie-score than NT score.

Another way to look at it is to compare to the results of the AQ test. I featured the AQ test in a previous version, and over 600 people did both tests. From this information, I calculated conversion formulas between these tests. Using those, it turned out that the NT control group in Aspie-quiz largely had similar scores as Simon-Baron Cohens control group.



But Aspie-quiz doesn't have a random sample. In most versions, the Aspie group is the largest, but usually they are pretty equal. That's because the test is mostly taken by people that suspect they are autistics, and because it mostly runs on autism-community sites.

Ah, it was never my intention to suggest that, though. Clearly not expressing myself as well as I'd like. I was simply trying to say that, overall, the Autism Spectrum is a continuous line from low functioning autism, through high functioning, onto Asperger's, and then into NT as one continuous line (as I visualise it, anyway).

Well, I'm not sure if there are any firm numbers for Asperger's, at least none that I've seen. Best Autism guesses right now are around 1 in 150 children, if we accept the theory that the number is increasing (IMO this is because of genetic self-selection, but I accept this is far from proven), I'm assuming perhaps some of those included in that number are Asperger's, not high-functioning autism (assuming you believe there's a difference) and perhaps a similar number of definitive Asperger's (OK, so 1.5%?), then an increasing number of borderline cases in a bell curve? It's hard to guess without any firm number of diagnoses available. But 15% seems to me to be so high as to defy credulity.

Exactly.

See, even then you're self-selecting. If we assume that Asperger's just doesn't exist in those with an IQ below 100, and that 100 is a rough average along a bell-curve for IQs (real-world seems to come out with an average anywhere from 95-105), then half your population won't have it straight out of the gate, and that half will not be at a university, and so wouldn't be likely to take our student's test. You've doubled the numbers right there. These people are also less likely to be on the internet and, I'd guess, even less likely to be online taking tests about ASDs.

The more I think about it, the less I think it's possible to come up with concrete numbers without asking doctors to report when they make diagnoses, and get numbers from central government. Anything else is just hoping.

Do you have a link to that study? I ought to read over it. I'm totally just speculating here.
How many NT controls were there, and were they just random people off the internet, did they get redirected from this site, or?...

Guess that just answered my above question. I still think that the numbers you'll likely get from this is probably skewed high.
Why were we talking about this, again?

Anyway, I think if I've got an overarching point, it's that I don't believe that any of these tests are a good way of trying to figure out the prevailance of ASDs in the general population. What I do think they're useful for is as a self-diagnostic tool amongst those who suspect they have ASDs, and also as a better way to understand the disorders we have. The test featured here excels in that latter part, or at least will once it's refined.

I've already thought about this and this is idea is flawed. Some abilities are actually disabilities depending on the context or point of view such a hypersensitivity to sound. The whole point of ASC is it does not specifically record abilities but rather ASC components. These components have to be established enough to merit going on there, though these are still fairly broad areas. I'm not aiming to prove a specific theory pertaining to autism, and won’t set up the test accordingly.

It should show ADHD, dyslexia, dysphasia traits although the objective is to map everything relative to the autistic spectrum components. I myself am diagnosed with ASD co-morbid with ADHD and have serious executive dysfunction as well as dyslexia and other things. But really there is not clear interface between any of them. They do fall under ASC.

The abilities section is merely an additional possibly optional section. It is just something that may be used to cross reference with the ASC. I’m not a big fan of IQ test either. There is more general psychometric tests that I have done that are more broader than simply an IQ test, but they are best done with supervision as you said. I recon it is probably not necessarily to have the abilities test in there as it is up to health professionals if they want to cross-reference psychometric tests with the ASC. It was merely a response to people who said ASC didn’t show the ‘positive’ aspects of ASD. But really although I hope it will improve nuero-diversity, it’s purpose is primarily for analysis and not campaigning on nuero-diversity.

The aim is to start from a little bias as possible. Improve the objectivity of the questions. By objectivity I mean one person taking the test, will as close as possible to another person taking the test in how they would interpret questions. So definitely practical sections will play a part as well as improved questioning techniques. Then will use vector mapping and statistical training, etc to look for patterns in the data. Although the data will be collected centrally, anyone would be free to do their own analysis of that data. I’m interested in doing that myself, naturally.