Gifted Issues Discussion homepage Forums Identification, Testing & Assessment Large discrepancy in test results

Yes.

With regard to the NNAT-2, I am not aware of the specific news coverage mentioned upthread, but I do know that the author has been involved in a fair amount of litigation arguing that the use of other instruments, with heavier language-loads, or more subjective criteria (e.g., teacher rating scales), unfairly discriminates against children from cultural-linguistic minorities. I mention this to give some perspective on the thinking behind its design. (I actually sat at the breakfast table with him once, while he was discussing some of his advocacy experiences with another distinguished professional in the field. Yes, I was eavesdropping!) The test is not without its flaws (as with any instrument), but it is used in GT admissions because its history is less discriminatory toward diverse populations.

In the case of the OP's child, the lack of scantron experience would probably be the place to start. If it turns out that there are visual issues, one accommodation would be to have the student mark the responses directly on the test booklet, then have a proctor transfer the responses to the scantron sheet after the test. You may also be able to have them check the original test booklet, to see if there were bubbling errors, as suggested by several PPs.

Thank you for all of the responses...good info and food for thought. I am going to ask his current school if it could give DS some type of timed test with a scantron and ask someone to watch him as he fills it out (I bet the school would do this for me). I will research a bit more on strategies to help students that may have lower processing speed. No teacher has ever mentioned him struggling to complete work in a timely manner, but sometimes, DS in his quest to finish quickly (homework, for ex), he does not do a thorough job reading instructions, etc. Thank you again!

aeh, you mention that the NNAT has a history of being more fair to diverse populations, but there is a study to suggest that english language learners scored more poorly on it than other tests. The study also found that the NNAT overidentified both extreme high and low scorers on the test. http://gcq.sagepub.com/content/52/4/275

This study was done on the original NNAT, not the NNAT2. I'm curious if the development of the new version has eliminated some of those flaws. The counselor at my son's school also mentioned that the NNAT2 was a more difficult test than the cogat, which seems strange given that they are both normed to an average of 100. His district is diverse (we live in Texas), so I'm wondering if that might have something to do with her impression, if indeed the NNAT2 underidentifies minority students in comparison to the cogat.

Sorry to derail the thread a bit, though it does point to the fact that these tests are imperfect. I would tend to trust the results of an individually administered IQ test over a 30 minute group test, even if the group test is more recent.

Appleton, I agree...again, realizing I am not objective about my own child, but that the likelihood of results from of an individually administered comprehensive test being completely 'off' is less then the likelihood of results from a 30 minute group administered test being completely accurate:). He took the NNAT2. Honesty, I don't think our district knows 1/100th of what the folks on this board do about tests. For example, when registering DS for the district's testing, the coordinator only referred to it as the 'GATE test'. I finally asked 'what test will he take'? and the response was 'it's the GATE test'...I responded that there were a number of tests that could be used (NNAT, OLSAT, etc.) and all I wanted to know was when to drop him off and pick him up and some tests are longer than others...geesh!

A drop of basically 4 standard deviations (assuming sigma=15) just does not seem plausible without something else going on. This isn't regression to the mean as much a 'reflection' about the mean almost as though someone was reading from the wrong end of the chart. I would follow this up IIWY

This raises a good point. One tricky thing about computer-read forms... within each section, some are numbered horizontally (across in rows) while other computer-read forms may be numbered vertically (downward in columns). Therefore it can be easy to mismark if one is focused on the questions and not being mindful about the answer sheet. Stray marks and incomplete erasures can also be problematic.

I think the OP means that her son scored in the 69th percentile,not a 69 SAS. The SAS for 69th percentile is somewhere between 105-110, which is around 2 SD different from the GAI of 135.

Good point, I just realised this and just logged on to delete my post but was too late - whoops

First, of course, individually-administered tests are better measures of cognitive ability for pretty much anyone. They are more nuanced, cover a more diverse range of ability-types, are more adaptive, and allow for querying when children give atypical responses.

Second, as to the study you linked, that study compared the NNAT to two other instruments administered as group cognitive assessments. As it happens, it was authored by the author of one of those tests (coincidentally, the test that the study found to be the "best" was the one by the author). Not knocking the research, but giving some perspective. The Ravens, which was found not to be properly centered (the mean is off), is a very old test, so that's not a surprising finding. The CogAT has a much larger standardization sample than the NNAT, so it is also not surprising that it would have more accurate norms. Note that the article is about comparing three nonverbal assessments. We can only see the abstract before the paywall, so I don't know what's in the rest of the article, but the title suggests that the comparisons were between the CogAT6 Nonverbal and the other two tests. This is not how schools commonly use the CogAT.

This study comparing the CogAT6 to the NNAT-2 found that the NNAT-2 had slightly less of a gap for Hispanics and ELLs than the CogAT Composite, with gaps significantly affecting GATE identification for Asians and ELLs. The CogAT Nonverbal was slightly better than the NNAT-2. They also found that follow-up WISCIVs matched the Composite scores the best, which also makes sense, since the domain overlap should be better: http://gcq.sagepub.com/content/57/2/101.abstract

The history of schools attempting to use figural tests for GATE screening reflects lawsuits in California back in the 80s and 90s regarding overrepresentation in special education of CLD/minority populations. Courts decided that the face cultural load of verbal IQ tests discriminated against diverse populations. This was the impetus for research and design into updated nonverbal tests. Item analysis since then has found that there is actually negligible cultural bias in most of the gold-standard cognitive assessments (though, naturally, LEP factors continue to affect performance on verbal tasks). So yes, it is quite possible that the CogAT6 composite does not discriminate against diverse populations any more than nonverbal instruments of equal psychometric robustness. One can make an argument that nonverbal assessments discriminate against verbally-skewed students of all CL backgrounds.

However, to my knowledge, the history of GATE screening with and without nonverbal instruments appears to support their use as part of the process, in terms of the actual shift in diversity of students accepted to programs. Whether use of these instruments continues to be necessary to level the playing field for CLDs is an open question, especially as newer tests come out, and CLD factors are given more attention in test design and item tryout.

Bottom line, no one score can be used to describe a whole person. Obvious, right?

Sorry, especially slow today, what does CLD stand for?