Originally Posted by aeh
Originally Posted by geofizz
Originally Posted by aeh
Originally Posted by geofizz
In my kids, IQ and grade level equivalents have no relationship, while the scaled scores of index scores correlate with one another, at least within a standard deviation or so. Grade level equivalents appear to be produced by a random number generator.

Attempting to make correlations with grade level equivalents in our case are a fruitless endeavor.

Not a random number generator. Regression & curve smoothing. But still with little relationship to real-world instructional levels. Which is why the APA and NASP specifically dis-recommend the use of grade equivalents in assessment reporting and interpretation.

I know this. I should have maybe worded it "as if generated..." The GE does not correlate well across subtests, nor does the same scaled score equate the same number of grades ahead for a given subtests across the years (so a 130 in kindergarten is very different than in 4th). DS has identical scaled scores on different subtests, yet the GE is wildly different. Looking at his GE's on the latest round of WJ (which, granted, will qualify him for special ed), range from -2 to +14 of his current grade level, which the SS's range from low end of average to +78 points above the lowest score.
Clear now.

The scaled/standard/all z-score-derived scores correlate with each other because all the psychometric work goes into making sure they do. That grade equivalent tables exist at all is an unfortunate concession to age/grade-locked teacher-thinking, which is an outgrowth of our warehoused approach to formal education.

I think grade equivalent scales and mental ages are as important as z-scores. Suppose the distribution of intelligence were much more compressed than it actually is, so that only 1% of 10-year-olds were as smart as the average 11-year-old. You could still create IQ scores with mean 100 and standard deviation of 15, but the need to make special accommodations for children with IQ of 130 would not be so urgent, because they would not be that much smarter than the average child their age. When, in reality, 10yo children with IQ of 130 are about as smart as average 13-year-olds, this suggests the need for accommodations.