Monday, March 28, 2005

Carroll Analyses of 50 CHC tests: Rnd 4-Two hypothesized higher-stratum models

Time to move to a higher level!

I will be attending the 2005 NASP conference this week. Although I don’t have time to interpret/discuss the results in detail, I decided that I would post the 2nd- and 3rd-order factor structures identified in the Carroll Analyses of 50 CHC-designed tests (see 3-17, 3-18, & 3-26-05 posts) with little comment—folks can chew on the results, and fellow NASPites might want to corner and chat with me about the results in Atlanta.

Click here and you can view/download/print the entire pdf file…which includes the descriptions of the tests used, the first-order results (already discussed) and two equally plausible higher-order structural models. These are then followed by two pages that provide background information regarding the terms (based on the dual cognitive processing model of information processing--with major distinctions between controlled vs automatic processing and process vs product-dominant abilities) suggested for one of the plausible models.

The second plausible structure, where the factors are labeled as per the Berlin BIS model refers to Berlin Model of Intelligence Structure, a model I believe that American psychologists need to become more familiar with. The Berlin BIS is a faceted model of intelligence (see SuB & Beauducel, 2005). Briefly, a facet model, which has its roots in the work of Guttman and the Radex models, attempts to describe/classify cognitive tasks with regard to several characteristics (typically mental operations and content—test format/stimuli). In contrast to the more-or-less discredited Guilford SOI model, each of the 12 BIS facets (4 operations x 3 content facets) do not have the status of ability factors, but instead, are used to classify performance on measures in each cell.

As you can see, the second alternative BIS higher-order structure suggests that CHC theorists may benefit from incorporating information regarding test content (figural, verbal, numerical) in possible “tweaks” to the CHC model framework. Collectively, both higher-order structures presented in these analyses suggest possible intermediate level stratum that should be entertained in CHC-driven research, particular if the results can be replicated in additional data sets.

Enjoy. I hope that some of the regular IQ blogsters will take a stab at commenting on these two plausible higher-order CHC frameworks. Much to ponder.

3 comments:

Joel Schneider said...

In what sense are the Gs and Naming Facility tasks "automatic?" They seem to involve a fair amount of controlled processing to me.

Kevin said...

Good question. I think it is largely a matter of releative degrees of controlled or automatic processing.

Gs and NA tasks, given that they are timed, place a premium on efficient automatized performance. Also, by definition, most all Gs tasks are comprised of items of easy to moderate (at best) difficulties, items that almost everyone would get correct if there were no time limits. The lack of difficult items, which is what is found on "level" measures (vs "rate" or "fluency" measures), is probably the key to considering these tasks as "more" automatic......what do others think?

John Garruto said...

I should read the .pdf file before commenting. Then again, there's probably a lot of things I should do, but don't.

Wouldn't others think there's an inherent difference between Gs and NA tasks? I think that's what moved our conceptualizaton of NA as a subset of Glr, rather than Gs.

Oh it involves speed, but Gs tasks are usually always novel. Our every day routines do not have us searching for identical numbers or that the number two had this symbol (those who have been triennialed to death excluded of course.)

I think there is key difference between cognitive fluency and Gs (and it's not just the motor component...I think the academic fluency measures of the WJ-III are truly not Gs measures.)

I guess long story short-one seems to be highly dependent on rate of recall from LTM while the other is a pure rate of cognitive processing speed. Perhaps the latter is needed for the former, but I do not think the converse is necessarily true.

Probably best not to ask what I think ;-)