In response to the previous post, Steve Sailer suggested the slight but steady rise over time in the percentage of GSS respondents deemed to have “good” comprehension of the interview questions asked them could be an illustration of the Flynn effect at work. The following graph quantifies the rise. The value displayed is derived by taking the percentage of respondents judged to have “poor” comprehension and subtracting it from the percentage assessed as having good comprehension. The middling assessment (“fair”) is left neutral:
There’s a bit of apparently random variation from year to year (most conspicuously the ‘spike’ in 1985), but the general trend is upward until 2006, the first year those not fluent in English were included in the survey. The slide beginning after 2004 might be a result of the Spanish version of the Wordsum test being more difficult than its English equivalent, but it is more likely simply a consequence of the fact that among people residing in the US, thoseable to converse in the Anglo-Frisian subgroup are sharper than those who unable to .
Steve also wrote the following:
It would be interesting to see who is overrated and underrated relative to their test scores. My impression is that people in New York City seem mentally quicker than people elsewhere. Some of that is actually mean IQ difference, but some of it is different affect.
To calculate predicted comprehension by geography, I took the percentages of each of the five intelligence categories (wordsum scores of 0-3, 4-5, 6, 7-8, and 9-10) by region and determined what each region’s mean comprehension would be if that region’s five groups showed good comprehension at the national rate. The following table shows how respondents in each region were assessed by interviewers relative to what their wordsum scores alone predicted their aggregate assessments would be. Positive values indicate respondents being assessed as having better comprehension than wordsum performance predicts they should have (they appear sharper than they actually are); negative values indicate respondents being assessed as having worse comprehension than wordsum predicts (they appear duller than they actually are):
|East North Central (IL, IN, MI, OH, WI)||1.1|
|New England (CT, ME, MA, NH, RI, VT)||0.6|
|Pacific (AK, CA, HA, OR, WA)||0.4|
|Mountain (AZ, CO, ID, MT, NV, NM, UT, WY)||0.3|
|South Atlantic (DE, DC, FL, GA, MD, NC, SC, VA, WV)||0.2|
|West North Central (KS, IA, MN, MO, NE, ND, SD)||0.1|
|Middle Atlantic (NJ, NY, PA)||(0.3)|
|East South Central (AL, KY, MI, TN)||(1.1)|
|West South Central (AR, LA, OK, TX)||(1.3)|
The variances are extremely modest, to the extent that the table should be taken very lightly. There is no way to tell if individual states within each of the nine regions are representated proportionally to their respective populations. Further, the interviewers presumably tend to be from the areas they are conducted the surveys in, so should conceivably already have already factored in any artificial increase or decrease in perceived intelligence by differences in affect.
Despite all this, the results still strike me as having some face validity. When I come across someone with a distinct upper midwestern accent, it creates a bit of a competency halo, in the same way British accents do. Southern drawls, on the other hand, have the opposite effect. The Boston affect does it for me (though it also makes me wary–must be too much Simpsonian influence growing up), yet so does the rapidity of the New Yorker’s word delivery. Consequently, the Middle Atlantic’s position feels the most out of place. But I have NYC in mind, to the exclusion of the bucolic expanse between Pittsburgh and Philadelphia.
GSS variables used: WORDSUM, COMPREND, REGION