Rdos blog

It is claimed in publications about the validity of the AQ test by Simon-Baron Cohen that 80% of diagnosed AS will score in the Autistic range while only 2% of controls will. This seems to indicate a very reliable test with good discriminitative power.

The cutoff to “very likely Aspie” in Aspie-quiz was originally set in version 7 to provide similar properties for diagnosed AS/HFA/PDD as the AQ test (80% would get their diagnosis confirmed). The control groups in Aspie-quiz however has much higher amount of “very likely Aspie” than the AQ tests 2% (the average is around 16%). This seems to indicate much worse discriminitative power of Aspie-quiz.

To evaluate relative discriminative power between the AQ test and Aspie-quiz it is necesary to administrer both tests to the same population. This has been done twice in Aspie-quiz. The first time was in version R4 (an early experimental version) and the second was in version F1 (the first final version).

Results from R4:

• 81% of diagnosed AS/HFA score above the cutoff in the AQ test compared to 75% in Aspie-quiz (6% difference)
• 66% of all males score above the cutoff in the AQ test compared to 58% in Aspie-quiz (8% difference)
• 50% of all females score above the cutoff in the AQ test compared to 43% in Aspie-quiz (7% difference)

Results from F1:

• 59% of diagnosed AS/HFA/PDD score above the cutoff in the AQ test compared to 70% in Aspie-quiz (11% difference)
• 42% of all males score above the cutoff in the AQ test compared to 46% in Aspie-quiz (4% difference)
• 45% of all females score above the cutoff in the AQ test compared to 50% in Aspie-quiz (5% difference)
• 16% in the NT control group score above the cuttoff in the AQ test compared to 19% in Aspie-quiz (3% difference)

In version R4, the AQ test consistently gave higher scores in all groups (but primarily in the whole group and the male group). In version F1, Aspie-quiz consistently gave higher scores in all groups (but primarily in the diagnosed and female group and to a lesser extent in the whole group and the control group). These findings seems to show that Aspie-quiz has slightly higher discriminative power than the AQ test.

The correlation between score difference in Aspie-quiz and AQ score is 0.83 in both R4 and F1.

The AQ test was developped specifically with the DSM definition of AS in mind, while the intention of Aspie-quiz is to identify positive, autistic personality-traits in adults. We would thus expect the AQ test to have much better discriminative power for AS/HFA than Aspie-quiz, but this doesn’t seem to be the case. Why is this? A hint is that six questions in the AQ test has no relevance whatsoever for Autism.

These questions are:

• If I try to imagine something, I find it very easy to create a picture in my mind.
• When I’m reading a story, I can easily imagine what the characters might look like.
• I find making up stories easy.
• I am not very good at remembering phone numbers.
• I don’t usually notice small changes in a situation or a person’s appearance.
• I am not very good at remembering people’s date of birth.

Some of the above issues were early stereotypes put forward by some autism-researchers that Autistics would lack imagination, which of course is not at all true. Another 2-3 questions related to pretending also have very weak relevance.

When 12% of the questions in the AQ test lacks relevance for Autism, it is not that strange that Aspie-quiz can do a better job at discriminating AS/HFA.

The AQ test also focuses only on six of the twelve groups in Aspie-quiz’ spider-diagram. These are:

• Neurotypical social
• Neurotypical communication
• Aspie compulsion
• Aspie ability
• Neurotypical ability
• Aspie perception (one question only)

It does not have any relevant questions for:

• Aspie activity
• Aspie hunting
• Aspie communication
• Neurotypical compulsion
• Neurotypical hunting
• Neurotypical perception

Here are two pretty revealing questions about friendships:

1. Do you prefer to have friends of the opposite gender?
2. Are friends of the same gender important to you?

One would expect these to be opposites, and they are. Their linkage to Aspie and/or neurotypical scores seems less obvious. Without deeper knowledge about this, it would be hard to believe any of them are relevant to being Aspie or neurotypical. The fact is that 1 is related to Aspie score and 2 is related to neurotypical score.

The traits related to a preference to have friends of the same gender are in the neurotypical compulsion group, and especially Do you prefer the company of those of the same generation as yourself? So, neurotypicals seems to have a preference to socialize in same-age groups and primarily with people of the same gender when it comes to friendships and with the opposite gender for relationships. This is hardly news as these traits cannot be missed in society.

The question about a preference to have friends of the opposite gender has different relations. It is related to:

1. Have you experienced stronger than normal attachments to certain people?
2. Do you tend to lean towards your partner when you are at a restaurant or party?
3. Do you tend to become obsessed with a potential partner and cannot let go of him/her?

These are the partner-obsession traits previously analysed in Aspie relationships. These traits group in the environmental group and in the Aspie communication group. They are related to environmental traits like depression because they give big problems in a society that think friendships are between peers of the same gender. The relation to Aspie communication is probably the natural connection.

These results have implications about how the natural social environment of Aspies work. If it is not based on same-gender, age-matched friendships, how does it work? A good guess might be that friendships does not even exist in the natural Aspie environment. What does exist is group-bonding, and the bonds primarily seem to exist between people of the opposite gender. A good guess would also be that what looks like opposite gender friendships naturally develop into relationships over time. This is why this trait is related to relationship questions. The previous finding that being in love in more than one person at the time is an Aspie relationship-type-of-trait fits nicely into the picture, because this indeed indicates a group-bonded social context rather than a monogamous, male-female social context.

Other indicative traits are these:

1. Do you prefer to only meet people you know, one-on-one, or in small, familiar groups?
2. Do you feel uncomfortable with strangers?

The above traits also indicate a preference for being with the same people all the time, and being reluctant to socialize with strangers. However, they do not point in the direction of a single male-female social context which exist in other primate species. These traits taken together point to a closed group social environment with no pair-bonding within.

As with the relationship traits, the correlation with Aspie score is weak, and thus this way of functioning cannot be generalized to the whole group.

A fundamental property of questions in Aspie-quiz is how they are related to each others. There is a general tendency for questions that have good relevance for Aspie/neurotypical scores (have a high correlation with scores) to also have high average correlation with other questions. This is shown in the following diagram which plots score correlation (x-axis) versus average correlation with other questions (y-axis) for 1,370 questions:

This property is very important for grouping questions and finding clusters in the dataset. As can also be seen above, if a question have high correlation with score, it will inevitably be correlated to about anything you want to ask. This is a common mistake in psychiatric research, and especially in research that includes Aspie traits. In fact, it is possible to group Aspie traits in about any collection of syndromes if the only criteria for traits to belong together  is that they are correlated to each others.

Another way to show this is by plotting score correlation (x-axis) versus highest correlation with all other questions (y-axis):

A similar thing appears. Questions with high score correlation will seldom have low correlation with all other questions.

These findings have important implications for finding clusters. Instead of finding clusters by only looking at correlations, which will give too small clusters if questions have low score correlation and too large clusters if questions have high score correlation, another method is needed. What is needed is to compare correlations with score correlation. So, when presenting correlated questions to a specific question, one should not use a absolute correlation cutoff, but rather a relative cutoff. Using 80% of score correlation as a cutoff usually works quite well and gives good clusters regardless of a questions score correlation.

This also works in the reverse way. If a set of questions is put together, and it is found that they are highly dependent on each others (without being identical), this is an indication that these questions are measuring Aspie traits.

I recently researched love and relationships in Aspie-quiz. First I added a number of typical romantic questions from a swedish quiz posted on aspergerforum.se. These questions didn’t go very well and showed mixed relevance. While the traditional romantic questions seemed at least weakly linked, they showed no consistent linkage with the Aspie/neurotypical dimension. Instead, some questions were linked to neurotypical compulsion (“Do you enjoy a good love story?”) while others seemed linked to Aspie communication (“Do you tend to lean towards your partner when you are at a restaurant or party?”).

Something needed to be done to resolve this. Therefore I added another nine questions from older versions of Aspie-quiz that seemed to be related to the topic of love and relationships. This seemed to clear up the issue.

The primarily linked questions now are these (intercorrelations are in  range 0.4-0.55 which is much higher than between romantic questions that have 0.2-0.3):

1. Is it harder for you than for others to get over a failed relationship?
2. Have you experienced stronger than normal attachments to certain people?
3. Do you tend to become obsessed with a potential partner and cannot let go of him/her?
4. Do you refuse to give up on a relationship or potential relationship that others would not bother with?

Here 2 and 3 seems to be the primary traits, while 1 and 4 are problems these causes in a non-Aspie adapted world. Both 2 and 3 have simlar relevance (a little above 0.4 correlation with Aspie-score), and are best related to each others and to the problems 1 and 4. This means that Aspie attachment to people is mostly related to obsessions with a potential partner (partner obsession) rather than to other compulsive traits. One can imagine this behavior can lead to stalking when combined with other traits.

Now, using these now established Aspie-traits one can look at linkage for other traits in the romance quiz. First, let’s look at the trait above characterized as linked to Aspie communication (“Do you tend to lean towards your partner when you are at a restaurant or party?”). It is linked to the following traits:

1. Do you refuse to give up on a relationship or potential relationship that others would not bother with?
2. Do you save love-letters?
3. Is it harder for you than for others to get over a failed relationship?
4. Have you experienced stronger than normal attachments to certain people?
5. Do you tend to become obsessed with a potential partner and cannot let go of him/her?

It seems evident that this trait is related to the partner obsession trait from above.  It also shows linkage to one of the core romantic traits (2).

Finally, looking at the core romantic traits, it is evident that these clusters are also related to partner obsession. The romantic cluster looks like this:

1. Do you save love-letters?
2. Do you enjoy a good love story?
3. Have you and a partner ever had a song of your own?
4. Is it harder for you than for others to get over a failed relationship?
5. Do you refuse to give up on a relationship or potential relationship that others would not bother with?
6. Do you tend to become obsessed with a potential partner and cannot let go of him/her?
7. Do you tend to lean towards your partner when you are at a restaurant or party?
8. Do you hope to meet the “right one”?
9. Do you believe in fate when it comes to love?

This seems to show that romantic traits really can be said to be Aspie in nature, but because Aspies have so many failures in relationships many of the individual traits shows no significant correlation to Aspie score (and in some cases shows correlation to neurotypical score). To research it, linkage studies between traits are needed, and partner obsession must be taken into account. Romantic love in fact seems to be a form of partner obsession.

Another (odd) cluster is this:

1. Have you been in love with more than one person at the same time?
2. Have you experienced stronger than normal attachments to certain people?

Contrary to the traditional view of romantic love, partner obsession doesn’t seem to be a monogamous thing. Here being in love with several people at the same time is shown to be related to partner obsession.

There are some other clusters related to sexuality as well:

1. If you were single, would you find casual sex (one-night stands) rewarding?
2. Do you have compulsive sexual behavior, e.g. spend too much time on sex or switch sexual partner frequently?

Finding these traits positively correlated is rather strange because 1 is correlated to neurotypical score while 2 is correlated to Aspie score. However, here it is also necesary to look at opportunity. Aspies seldom can persue casual sex and this is pretty much why they answer this question negatively. It is an issue of opportunity. The second question doesn’t need a real partner, but can be persued with pornography.

A final cluster shows how these traits are related to other sexuality issues in Aspie-quiz:

1. Do you have compulsive sexual behavior, e.g. spend too much time on sex or switch sexual partner frequently?
2. Do you have unusual sexual preferences?
3. Have you have had long-lasting urges to take revenge?
4. Are you hypo- or hypersensitive to physical pain, or even enjoy some types of pain?
5. Do you tend to say things that are considered socially inappropriate?

Compulsive sexual behavior is related to unusual sexual preferences. It is also related to an urge to take revenge (this could be the missing trait that creates the stalker).

So what does this mean for the average Aspie? Not much as the correlation with Aspie score is weak. What it does show is that romatic traits must have been part of the original Aspie profile before these traits diluted into the general population. It also shows that the romantic traits did not evolve in a simple male-female monogamous environment.