Watchman

Gaelic

TRAITORS

Medicine?

IAEP 

International Assessment of Education Progress

• The percent of Chinese students who correctly answered IAEP math questions was almost twice ours. 
• Mozambique's natives' average IAEP Math score of 423 equals zero intelligence. 
• The comparison between China and the US is 80 percent versus 40 percent rather than 561 versus 494.
• Our 13 year olds scored lower than Spain's.
• Correlation between IQ and IAEP is an r-squared of 0.74.

The unbelievably low score of our 13 year olds on IAEP math is almost too difficult to explain.  In order to understand why we, at 494, would have scored so much lower than any other industrialized nation, 45 points lower than Switzerland, 51 points lower than Taiwan, and 39 points lower than Russia, we must delve into territory now claimed to be that only for racists.

But delve we must.

Any way you measure education spending, we already outspend the above nations by 2 to 6 times, so it cannot be blamed on lack of dollars, or lack of concern.  If we presume that the 3 percent of the students in the US who took the test were Asians who scored in the range of their brethren in Taiwan or Korea (542 to 545), that the 11% who are Hispanics scored in the range of their brethren in Spain, and that the 73% who are Whites scored in the range of their brethren in Switzerland, then the only way to explain our extremely dismal score is if the 13% who're blacks scored 209.

As bad as we know blacks perform academically, a score 218 points lower than Mozambique and 324 points lower than Russia is an impossibly low score.  How can it be explained?  In the worst scenario, assuming American Whites score in the range of Russia, Asians score in the same range as American Whites, and Hispanics score in the range of Brazil, then blacks score 308:

The problem is that neither Asian students, Hispanic students nor White students could possibly score so much lower than their brethren in Japan and Europe and Spain.  The worst case scenario would be a compromise between the two, for an average for Asians of 538, for Hispanics of 469, and for Whites of 536.  This is extremely conservative, because it estimates that Asians in the US score 7 points lower than Korea, Hispanics score 26 points lower than Spain, and Whites score 3 points lower than Switzerland and 9 points lower than Korea.  But for the US to have the low score of 494 would require that American blacks score 270, which is 127 points lower than Mozambique and 268 points lower than American Whites:

An r-squared of 0.7415 was achieved when Professor Lynn's IQ of nations was used to correlate IAEP scores to IQ's, proof that his estimates for national IQ are very close to what the IAEP measures.

His estimates for IQ's were then compared to a linear Excel scatter plot and adjusted to achieve an r-squared close to 1.0.  Doing this required IQ's for some countries to be decreased by: 1 point for the UK, 4 points for the US and Portugal, 5 points for Spain, and 6 points for Brazil.  iow, the actual performance of students on this test was significantly lower than his estimates for IQ would indicate they should be.  This could be due to some of his data being out of date, thus failing to reflect dramatic decreases in academic performance in all of these formerly great nations.  His overestimate for the UK might be due strictly to his being from the UK.  Other standardized tests confirm that his estimates for Spain, Brazil, and Portugal are a bit high.

IQ's for other countries were increased by: 1 point for Hong Kong and Korea, 2 points for Canada and Slovenia, 3 points for France, 4 points for Mozambique, 5 points for Switzerland, 8 points for Russia, and 12 points for China.  The 1 and 2 point differences for Hong Kong, Korea, Canada, and Slovenia are not significant and could disappear in other tests, or in other years of the same test.  The four point underestimate for Mozambique indicates either that their schools are doing a much better job than anyone realized, or there was a flaw in the way they handled the test or the data.  Performance of Switzerland on other tests indicates that their IQ has increased significantly in recent years, or has always been underestimated.

Which leaves us wondering about his whopping 8 and 12 point underestimate for Russia and China?  This is most likely left over from cold war propaganda. As a scientist, having lived in and done business in both Russia and China, I'm aware of the magnitude of this propaganda, and so suspect that this is where his error comes from.  I also know that both countries have accomplished amazing things which require IQ's higher than the mere technological-development IQ's of the US and Europe.  Both countries could have fabricated this data or manipulated the student statistical subset, but my bet is that the data is correct and the IQs of both countries are that high.  After all, who lives in Hong Kong where Professor Lynn DOES estimate the IQ to be 107, just one IQ point lower, other than the same Chinese who also live in China? Furthermore, both PISA and TIMSS confirm precisely his estimate for Hong Kong of 107.

Russia has pockets of excellence which leave us in the dirt, indicating that their IQ's are very close to China's, if not higher.  Which leaves us with our nation's "capitol", Washington, DC, where up to 81% of the population are the lowest scoring blacks on the universe, with IQ's 4 points lower than the average black American and 10 points lower than Mozambique.  What exactly HAVE they accomplished, what could they possibly accomplish, and why are my fellow Americans so STUPID that they think they need to elect a black president?

Each 3 point decrease in IAEP scores equals a 1 IQ point decrease in IQ.

Each 2.2 point decrease in NAEP scores equals a 1 IQ point decrease in IQ.

Direct from the NAEP Web Site

ABUSING SCALED SCORES

The objectivity and amazing integrity of international studies like IAEP, TIMSS, and PISA can be completely undermined with the use of scaled scores in the hands of educrats with an agenda, as the following table illustrates.  In 1991, 80.2 of Chinese 13 year olds, 55.3% of ours, and 28.3% of Mzambique's correctly answered the IAEP math problems and received scaled scores of 561, 494, and 427 respectively:

What these scaled scores conceal is that IF these were all four part multiple choice questions, then just guessing at the answers would yield 25% correct.  So an IAEP score of 427 in a student body in which only 28.3% answered correctly is very close to zero intelligence and math ability.  In addition, numerous known test taking strategies could increase the percent correct to 30%, or even more, when guessing at questions to which you know that you don't  know the answer.

The following graph shows the linear relationship between scaled scores and the uncorrected percentage correct where each 1% increase in the number of correct answers equals a 2.6 point increase in the scaled score:

The lowest one percentile of students in China, Switzerland, Quebec, and Saskatchewan scored considerably higher than the average Mozambique student.  But half of the 95th percentile of Mozambique students correctly answered the questions, suggesting the 4% who're Portuguese (and whose genetic brethren back in Portugal answered 48.3% of the questions correctly and had an IAEP score of 483) and the 2% who're Indians who score slightly higher, at 490), also took this test.  This would mean that the IAEP scores of the indigenous niggers was lower  than 427:

(.04 x 483) + (.02 x 490) + (.94 x X) = 427

X = actual IAEP score of indigenous niggers = 423

The fact that so many of their cousins in neighboring South Africa scored lower than if they'd just guessed on so many similar math problems gives validity to the report that the lowest one percentile of Mozambique's students answered only 11.5% of the questions correctly.  Where the use of scaled scores makes it appear that the comparison between China and Mozambique is 561 vs 427, the actual comparison is 80.2% correct versus 0.0% correct.  Where the standard deviation of Mozambique in general is 18, the actual standard deviation of the native population is zero, as their actual performance on the test is less than zero.

Such a racist, invidious, misleading use of scaled scores is unfair both to students who performed well and the students whose complete lack of math skills has been concealed and ignored.

The SAT Math Equivalent (SATME) Based on NAEP:IAEP Crosslink Study

• One SATME point equals:
  • Two TIMSS Geometry points.
  • Two percent correct on Figure 13.
  • One third of an IAEP Math point.
  • A 1% increase in men teachers.
• An average SATME score of 420 equals zero percent correct.

Table S23 Mathematics proficiency scores for 13-year-olds in countries and public school 8th-grade students in states, calculated using the equi-percentile linking method, according to Beaton and Gonzales, by country (1991) and state (1990) provides the opportunity to create an "SAT Math Equivalent" (SATME) to grade each country based on 12th grade SAT Math score of each state.  The IAEP:NAEP curve has a linear correlation with actual NAEP scores of public schools by state of r-squared = 0.9363, which provides confidence in the accuracy of this correlation.  But when compared to the NAEP scores of non-public schools by state, r-squared decreases to 0.6583, which raises questions about how accurate the IAEP:NAEP will be with other countries; or with a grade level which represents an age difference of 4-5 years; or with 12th grade TIMSS scores by country; or with the percent of correct answers by country on the TIMSS test subjects for which scores are available.

The problem is an inconsistent deviation between the NAEP scores of public and non-public schools.  The difference in states like North Dakota, Iowa, Minnesota, Nebraska, Massachusettes, and Rhode Island is only 6-12 points, but the difference in states like Texas, Georgia, California, and New Mexico is 20-31 points.  The reason that all states fall into two distinct classes like this is unclear, but it does explain the poor correlation with Beaton/Gonzales.  Since the difference in average math scores between blacks and whites of 28.9 NAEP points is equivalent to 110 SAT Math points, the 31 point difference between the public and non-public schools of Texas is the equivalent of 118 SAT Math points.  In other words, the difference in math skills within one state between public and non-public schools is as big as the difference in SAT math scores between engineering and education majors.

The nonpublic schools in California, South Carolina, New Mexico, Massachusetts, New York, Louisiana, and Rhode Island DO score considerably higher than the public schools of their state, they score lower than the public schools in North Dakota, Iowa, Minnesota, Missouri, Montana, and Nebraska:

Based on this, it would be expected that the NAEP which tests 13 year olds and SAT Math for 12th graders would not show very high correlation, but r-squared with 12th grade SAT Math scores by state is a surprising 0.8483.  This demonstrates that *within* the US there is a high consistency between states between these two grade levels.  In other words, there is little change in state ranking from 8th to 12th grade in NAEP scores.  The SATME is created from a linear extrapolation of the IAEP:NAEP data and assigns an SAT Math score to each country which is linearly proportional to its IAEP score.  Taiwan, the highest scoring "state" with an IAEP score of 296.7, is assigned an SATME of 555, and Jordan, the lowest scoring "state" with an IAEP of 236.1, is assigned an SATME of 445.  R-squared for SATME and 8th grade TIMSS Math scores by country shows the same low correlation which non-public schools show, or 0.5287.  But r-squared between the TIMSS Geometry scores of the 16 countries whose 12th graders participated in TIMSS and their SATME grade is 0.8128, which is equivalent correlation to IAEP:NAEP to TIMSS Geometry (0.8483).  This is evidence that SATME is an accurate way to grade the average math skills of students in each participating country.

The international TIMSS study provides the ability to correlate the percent of correct answers by country to each country's TIMSS score, which in turn enables a correlation to be made between the percent of correct answers in TIMSS math and SAT Math scores.  As would be expected, there is a close correlation between TIMSS Geometry scores and the percent of correct answers on TIMSS Geometry questions.  But there is also a high degree of correlation with probability and statistics questions, and an even higher correlation with calculus questions.  This suggests either that geometry is an important foundational skill for advanced math skills, or that those countries whose schools are good at teaching geometry are also good at teaching other math skills. 

The average r-squared for the 7 math items which show the highest degree of correlation is 0.6675.  You can see from the graphs that SATME predicts the percent of a country's students who can correctly answer Geometry Item J11 to within plus or minus 6.4%, and for Probability & Statistics Item I05 to plus or minus 10%, which is sufficient accuracy for a correlation to SAT Math.  The SATME grade crosses zero percent correct at an average of 404 points, and it crosses 20% correct at 437 points.  If half of these questions were five-answer multiple choice question, and if half of them require a direct answer, then a student who just guessed at these math questions would receive an SATME grade of 420 points.  In other words, these TIMSS questions show that an SAT Math score of 420 is equivalent to zero math knowledge.  Each 1% increase in the percent of correct answers raises the SATME grade by an average of 3 points, so the upper limit of the SATME grade at 100% correct is 720 points:

Copyright @ 2007 by Fathers' Manifesto & Christian Party