Timeline for Rationale behind an “all-or-nothing” design for a test for the purposes of diagnosis and epidemology
Current License: CC BY-SA 3.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 22, 2017 at 17:08 | vote | accept | Wrzlprmft | ||
| Feb 22, 2017 at 17:07 | comment | added | Wrzlprmft | You are right, “this test is flawed” was a bit going too far. Rather, it was: “was this possible effect taken into account?” Anyway, I am not complaining. | |
| Feb 22, 2017 at 16:12 | comment | added | Arnon Weinberg♦ | Your question was "What are the justifications for such a test design", not "Is this test design flawed". Had your question been "Is this test design flawed", then it would likely have been Closed for being Primarily Opinion Based, which is outside the purview of this forum. I suspect you think this scale should measure something different than what the authors think it should measure. | |
| Feb 22, 2017 at 9:26 | comment | added | Wrzlprmft | beyond the scope of this question – well, it is more or less my central question. However, I am open to the possibility that the answer to this question is that the test is actually flawed in this respect. | |
| Feb 20, 2017 at 21:13 | history | edited | Arnon Weinberg♦ | CC BY-SA 3.0 |
Updated re clarification from OP
|
| Feb 20, 2017 at 20:01 | comment | added | Arnon Weinberg♦ | Yes, absolutely. As I said, that is a discussion beyond the scope of this question (though the post I linked to does discuss this issue - for example, your assertion that autism and mathematics are separate constructs is just as contentious as a scale that measures mathematics when it purports to measure autism!). Anyways, it sounds like you have a good grasp of the issues already. | |
| Feb 20, 2017 at 19:29 | comment | added | Wrzlprmft | Regarding the subject of validation: I think I understand this. My problem is rather that the test has properties that will make it succeed validation irrespective of the desired qualities. As a simple and blatant example, consider a world where 80 % of all mathematicians are autistic and vice versa and consider a test that simply asks the subject whether they are mathematicians. This test is obviously flawed, yet it will have pass a basic validation. | |
| Feb 20, 2017 at 17:05 | comment | added | Arnon Weinberg♦ | Re telling us about reality, this is the purpose of validation (a topic far too large for this post, but see cogsci.stackexchange.com/questions/9610/… for more detail). The validation process confirmed that the amplification is both expected and realized, and does in fact tell us what we want to know from this scale about reality - ie, that mathematicians do have higher than average autistic tendencies. It's possible that you think this scale should measure something different than what the authors think it should measure. | |
| Feb 20, 2017 at 17:00 | comment | added | Arnon Weinberg♦ | @Wrzlprmft Yes, it is a forced-choice (ipsative) test, the authors themselves state this in the paper. A Likert-scale would have different scoring for slightly vs strongly (dis)agree. I was merely pointing out that scoring in such tests discourages comparisons between patients (compared to say, IQ tests), which may make them preferable for diagnostic tests that have potentially stigmatizing labels. The authors do not explain their decision one way or the other. | |
| Feb 20, 2017 at 11:35 | comment | added | Wrzlprmft | the original study confirms that mathematicians score significantly higher on this scale, as is to be expected – Sure they do, but that does not address my doubts, which (somewhat exaggerated) are: The way the test is designed, small tendencies are amplified. For example, a mathematician would will slightly agree with an “autistic” option, because they are primed that they are somewhat more autistic than average. Therefore a high AS score for mathematicians is expected from the test design and does not tell us anything about reality. | |
| Feb 20, 2017 at 11:25 | comment | added | Wrzlprmft | Unless I completely misunderstood the terminology, the AS test is not an ipsative test. Rather it is a Likert-scale test without a neutral option. I also fail to see what assessing pupils and beating previous scores has to do with the AS test. | |
| Feb 20, 2017 at 7:21 | history | answered | Arnon Weinberg♦ | CC BY-SA 3.0 |