Tuesday, December 23, 2008

IQ Scholar Spotlight: Dr. Linda Gottfredson

Another IQ Scholar Spotlight - Dr. Linda Gottfredson.

Dr. Gottfredson organized (and presented) at the recent 2009 (Dec) ISIR conference a symposium called:  Causal models that integrate literacy, g, and health outcomes:  A practical guide to more effective disease prevention and health promotion?  She is a highly regarded intelligence scholar who has a wide array of interests in the field of intelligence including (from her web page):
  • Intelligence, health and everyday life
  • Intelligence and social inequality
  • Employment testing and job aptitude demands
  • Affirmative action and multicultural diversity
  • Career development and vocational counseling
The diversity of her interests is reflected in a wide range of publications (which can be found at her web page).  For example, her 1997 Intelligence article "Why g matters:  The complexity of everyday life" is one of the top 10 cited articles in the journal Intelligence.  She was also the lead author of an important 1977 Intelligence editorial (Mainstream science on intelligence: An Editoral with 52 signatories, history, and bibliogrpahy) in response to the controversial Bell Curve book.  Here most recent "in press" Intelligence publication is Arden, R., Gottfredson, L. S., Miller, G., & Pierce, A. (in press). Intelligence and semen quality are positively correlated.

I found the measurement/prediction issues raised at her symposium very interesting.  The bottom line is that in the field of health care/literacy, doctors expect (hope? pray?) for 100% compliance in patient follow-through in treatment recommendations, the taking of prescriptions, etc.  So...a central question is how to ascertain which patients need more assistance in understanding their health care.  How can we predict which patients will need additional or special instructions and/or follow-up?  Of course, as we all in the field of individual difference measurement know, the best available measures in intelligence can only explain up to 50+% of the variance of any outcome or dependent variable.  An interesting dilema.

I had the opportunity to chat with Dr. Gottfredson about this issue.  One of her interests is in finding (or developing) brief, ecologically valid measures of g to screen patients.  She articulated the need for measures that tap a patients ability to handle complex information processing (high g tests) but that do NOT appear to look like intelligence measures, seem more "real world" in terms of ecological validity, and that would be easy to administer and score.

I shared with her some unpublished g-factor loadings of all the WJ III tests (when cognitive and achievement tests are combined together).  These g-loadings were calculated at different age groups using principal components analysis.  A summary of the table I provided Dr. Gottfredson can be found by clicking here.  Of interest (in our discussions) was a test like Understanding Directions....a test where a subject follows an increasingly long and complex set of simple directions (e.g., point to.....now point to.....now point to....,then.....,but first......).  As can be seen in the attached table, it is a high g test that would appear to have ecological and face validity for this purpose.  I believe it is a high g test due to the complexity of language-based working memory demands placed on subjects.  This discussion (and material) is presented here to stimulate thought and discussion.  Readers not familiar with the task demands of the WJ III tests should click here.  [Conflict of interest - I'm a coauthor fo the WJ III].

Technorati Tags: , , , , , , , , , , , , , ,


Anonymous said...

It's ironic that you'd espouse her "diversity of interests" range in light of allegations of her racially insensitive financial benefactor, i.e., The Pioneer Fund. http://mediamatters.org/items/200410250007

M said...

The Pioneer Fund has a very high standard of output - for instance the Twins Raised Apart study.

Liberal creationists don't like having their bubbles burst I guess.