The Commonwealth of Puerto Rico is a Spanish-speaking territory of the United States. Puerto Ricans are United States citizens—they can freely migrate between the island and the states, join the military, or even run for president. But they can’t vote for president, because the territory is not a U.S. state. In three referendums from 1967 to 1998, Puerto Rican voters rejected both political independence and U.S. statehood. However, in November 2012 a solid majority (61.3%) voted in favor of statehood. This kind of political nudging could quite possibly result in Puerto Rico becoming the 51st U.S. state … but only Congress and the president have authority over the matter, and analysts agree that approval is unlikely. This particular referendum also left off the option traditionally most favored by Puerto Ricans: continued commonwealth status. Many islanders appear to feel that statehood offers few additional benefits over citizenship; a majority of Puerto Ricans already live on the U.S. mainland (5 million vs. 3.7 million).

From the earliest days of intelligence testing, social scientists have taken a special interest in U.S. Hispanics. Proportionate to their numbers, it’s possible that more tests have been given to Hispanics than to blacks. But this special attention has also lacked focus. African-American test results have been subject to meticulous cataloging, synthesis and analysis (Shuey, 1966; Jensen, 1998; Jencks & Phillips, 1998) leading to somewhat of a consensus on the size and shape of the black-white cognitive performance gap. Yet there has not been a similar effort to process the disparate and voluminous literature on the abilities of U.S. Hispanics. Therefore there is less knowledge and consensus about the historical and contemporary test performance of Hispanic minorities.

Most of the U.S. Hispanic population is Mexican American (63%). Puerto Ricans are the second largest Hispanic minority (9.2% … or 15.3% including the Commonwealth). This post represents the first effort to comprehensively summarize the abilities of one of these two important American minority groups. Here I describe and analyze the results from over 70 studies that have measured the abilities of Puerto Ricans.

                 SECTION I: INTELLIGENCE TEST STUDIES IN PUERTO RICO                 

Table I summarizes the data from 26 different studies of intelligence in Puerto Rico. The eight columns describe the year the test was administered (Admin), the age range of the study subjects (Age), the number of subjects tested (N), the name of the test (Test), the “Flynn Correction” (FC), the Intelligence Quotient (IQ), the Standard Deviation (SD), and the study citation with PDF (Reference).

The Flynn Effect is the systematic inflation of IQ scores over time. Scores increase, on average, by 0.3 points every year, so that older test norms will give higher scores than newer test norms. To correct for outdated comparison norms, I adjust scores downward by 0.3 points for every year between the test norms and the year the test was administered for a given study.

The alternating bands of color represent different study samples. Most often rows sharing a color band are data for the same group on different tests, but it can also be the same group at different ages. Different references can also share a color band if they report data for the same people (or at least a substantially overlapping group of people). Values are not repeated in the same band; if the sample size is 100 in the first row, then that is the implied N for the row underneath.

Table I. IQ test scores in Puerto Rico

Admin	Age	N	Test	FC	IQ	SD	Reference
1921	8—9	1815	Holley		97		Monroe, 1921
1925	9—13	847	Pinter		102.2		Columbia University, 1926
1928	5—9	852	Detroit	1.8	84.2	14.1	Vincenty, 1930
		684	Pinter	1.5	81.5	14.6
		547	Dearborn	2.1	80.9	12.1
1954	6—16	128	WISC	1.5	86.5	21.6	Roca, 1955
1954	5—17	104	SBIS	5.1	90.6	18.2	Roca, 1955
1962	18—54	1500	GATB		82.5	17	US Employment Service, 1965
1965	7	236	I-AT		85		Manuel, 1966
1965	15—64	1110	EIWA	3.6	83.1	16	Green & Martinez, 1967
1971	Adult	253	ACB		77.4	14.7	Maier, 1972
1975	5—8	210	EIWN	1.5	83.8	12.5	Fuertes et al, 1976
1977	5—12	2400	CPM	0.6	80.6		Kahn et al, 1977
1977	9—16	2911	SPM		84.6		Kahn et al, 1977
1979	5—15	100	EIWN	9	100.9	14.6	Prewitt-Diaz & Gaisel, 1980
1976	6—8	50	McCarthy(S)		69		Shellenberger, 1982
1982	3—18	1488	TVIP	0.3	82.6		Dunn et al, 1986
1984	11—13	50	EIWN-R	0.6	91.2	14.5	Rodriguez et al, 1985
1986	8—10	34	WISC	11.1	98.6	12.7	Matias Carrelo & Zaidspiner, 1990
			WISC-R	3.6	101.1	12.9
1986	7—15	11	EIWN-R	1.5	86	9.4	Rossello et al, 1988
			DAM		83.7	12
1988	7—10	30	EIWN-R	1.8	109	15.7	Vazquez, 1989
			KABC(S)	1.5	100.9	15.3
1988	37	12	EIWA	9.3	80.6	8.2	Ortiz Colon et al, 1993
1990	19	30	EIWA	6.9	114.2	5.4	Andrue et al, 1991
			WAIS-R	2.7	96.1	11.1
		60	APM	1.5	105.4	7.5
1991	11—14	68	SPM	3.6	91.4		Alvarez et al, 1994
1997	9	108	EIWN-R-PR	1.5	108.8	11.8	Busigo et al, 1998
2001	4	152	WLPB-R(S)	3	82.7	12.7	Paez et al, 2007
2004	5	11	KBIT	4.2	98.9	11.4	Anderson & Souto, 2005
2005	6—17	72	EIWN-IV		96.2	17	Montes et al, 2010
2007	16	30	EIWN-R	7.5	95.5	13.3	Pons et al, 2008
			EIWA-III		105.8	9.3
			SPM	8.4	84.6
2008	11—13	40	EIWN-R-PR	4.8	101.7	12.3	Torres Diaz et al, 2009
			EIWN-IV	0.9	91.5	8

__________________________________________

■ Ia. Test Standardizations

All of Richard Lynn’s books on national IQ scores cite two studies for the intelligence of Puerto Rico: the standardization of the colored (CPM) and standard (SPM) forms of the Raven Progressive Matrices. The numbers for each test are cited from different Raven test manuals, and are said to be from different data collections (1975 for the CPM and 1977 for the SPM). I don’t know if Lynn’s manuals actually say this, but my own Raven manual includes the norm tables from both studies, calls them both 1977 standardizations, and gives them the same reference: Kahn et al (1977). This reference is unpublished and not held by any library, so I tracked down one of the authors, who generously mailed me a photocopy of the original manuscript.

In 1975, Congress enacted PL 94-142 mandating special education services for children with disabilities. School districts would typically determine eligibility for these services through a combined use of intelligence and achievement tests. As a US territory Puerto Rico was subject to the same law, and so the PR Department of Education commissioned the standardization of the Progressive Matrices to have local test norms for this purpose.

These Raven standardizations are the largest and best data source for Puerto Rico. The Standard Progressive Matrices—a 60 item nonverbal abstract reasoning test—was given to 2,911 8.5-16 year-old schoolchildren. The Colored Progressive Matrices—a 36 item version designed for younger children—was given to 2,400 5-11.5 year-old schoolchildren. Compared with the UK norms, I calculate an average IQ of 82.8 on the two tests. This is virtually identical to the average IQ calculated and reported by Richard Lynn for the two tests (83.5, which is rounded up to 84).

Four other studies have sample sizes of over 1,000. The second largest study, by Walter S. Monroe (1921), is also the oldest published study of Puerto Rican ability. Monroe gave 1,815 second and third grade San Juan public schoolchildren the Holley Picture Completion Test and compared them with 1,560 US schoolchildren. I calculate an IQ of 97 in relation to the US children. Monroe took this similarity in outcomes as a credit to the test instrument, but these children were probably an elite sample. In 1921 a much smaller percentage of Puerto Ricans attended school than was the case in America, and I do not attempt to determine the magnitude of or correct for this sampling issue in any of the studies here.

The third largest study, organized by the Department of Labor in the early 1960s (US Employment Service, 1965), is a standardization of the General Aptitude Test Battery on a broad sample of the adult working population, ages 18-54. These workers had the same amount of education as their US counterparts and no one was tested who had less than a sixth grade education, making this an elite sample. The test, which includes verbal and nonverbal subtests, was translated into Spanish, but according to the report, not otherwise changed or adapted. The IQ of these 1500 working adults was 82.5 on the General Learning Ability subtest.

The fourth largest study was the standardization of the Spanish language version of the Peabody Picture Vocabulary Test-Revised (called the TVIP: Test de Vocabulario en Imágenes Peabody); this popular point-at-the-picture-that-represents-the-word-I-say vocabulary test was translated and normed separately on populations in Spain, Mexico, and Puerto Rico (Dunn et al, 1986). The scores in Mexico and Puerto Rico were virtually the same, but were almost one standard deviation lower than the scores from Madrid. Scaled on American norms, the IQ of this sample of 1,488 3-18 year-olds was 82.6.

Lloyd M. Dunn, the eminent Canadian-American psychologist who created the PPVT, also wrote an English language companion manuscript for the TVIP that started a relatively small and now forgotten academic controversy that resembles the backlashes against “How much can we boost IQ and scholastic achievement” and The Bell Curve. Dunn’s manuscript, which was intended for educators using the TVIP with US schoolchildren, is a summary of Hispanic performance of cognitive tests. Dunn reviews the literature, which shows a reliable test score gap between Hispanics and non-Hispanic whites, and critically evaluates popular explanations for this disparity, such as socioeconomic status and bilingualism. Influenced by Arthur Jensen’s scholarship, Dunn finds poor evidence for these theories and concludes that inherited racial differences probably have a lot to do with the lower scores of Mexicans and Puerto Ricans:

While it is a very delicate and controversial topic, race, as a contributing factor, cannot be ignored … most Mexican immigrants to the U.S. are brown-skinned people, a mix of American Indian and Spanish blood, while many Puerto Ricans are dark-skinned, a mix of Spanish, black, and some Indian. Blacks and American Indians have repeatedly scored about 15 IQ points behind Anglos and Orientals on individual tests of intelligence …

For the reasons just cited … it would be simplistic and irresponsible to contend that the 10- to 12-point IQ differential is due exclusively, or even primarily, to bilingualism, as has often been argued, or even to all environmental influences combined, including cultural incompatibility … my best tentative estimate is that about half of the IQ difference between Puerto Rican or Mexican schoolchildren and Anglos is due to genes that influence scholastic aptitude, the other half to environment. (Dunn, 1988, p. 73)

The inevitable academic blowback against Dunn’s manuscript appeared in a special issue of the Hispanic Journal of Behavioral Sciences (September 1988, Volume 10, Number 3), where six social scientists take turns accusing Dunn of “misinformation,” “racism,” and “blaming the victim.” In his sheepish response, Dunn remarks that the publisher of the PPVT had stopped distributing his companion manuscript to educators after a major psychological organization moved to blacklist them. Shaken by the professional consequences of his integrity, Dunn apologizes for this portion of his manuscript, calling it insensitive and a “tactical error,” and then ceremonially retracts it from the public record. Perhaps it isn’t so surprising that there have been few subsequent academic review papers on Hispanic IQ. Omission and pretense are accurately called good manners in the context of everyday social interactions—it is equally accurate to call them fraud in the context of academic discourse.

The fifth largest study was the Puerto Rican translation and standardization of the Wechsler Adult Intelligence Scale (the EIWA or Escala de Inteligencia Wechsler para Adultos). Green & Martinez (1967) tested a representative sample of 1,110 adults, ages 15-64. Scaling the raw scores on the American norms I get an IQ of 83.1. Lopez and Romero (1988) make their own statistical adjustments to compare the two tests, and this method gives a similar IQ score (80.7). A problem with comparing these tests is that the WAIS was not only translated, but most of the items were changed in order to make the test much easier (Melendez, 1994). Two of the subtests—digit span and object assembly—were unmodified and can be directly scored on US norms, giving us an IQ of 78.5.

Several other early studies had sample sizes that were close to one thousand. In the 1920s the Teachers College of Columbia University and the Puerto Rican Department of Education collaborated to survey public schoolchildren with various kinds of achievement tests. 980 children, ages 9-14, were given the Pinter Non-language Mental Ability Test. Performance on these tests was very similar to an American comparison group. In fact, IQ scores were higher for Puerto Ricans in grades 3-5 before dropping below average in the 6th grade and falling to 93.3 by 7th grade. The researchers concluded: “the results of the tests … confirm … that the Porto Rican people reveal as much sheer intellectual ability as do the people who now make up that great urban mass of humanity in the towns and cities of continental United States” (Columbia University, 1926, p. 149).

The results of Monroe (1921) and Columbia University (1926) might suggest that the kinds of Puerto Rican children who attended school in the 1920s had comparable scores to the children that attended school in the United States in the 1920s. Except yet another survey of the ‘sheer intellectual ability’ of public schoolchildren was conducted in 1928 and these children did not compare as favorably with that great urban mass of humanity. Nestor Vincenty (1930) gave several popular non-language intelligence tests to a representative sample of some 852 schoolchildren, ages 5-9, from across the island. Their IQ was 82.5.

Later efforts by the PR Department of Education to translate various US intelligence tests are described by Pablo Roca (1955). The Wechsler Intelligence Scale for Children was translated into Spanish and a number of simple substitutions were made for culturally specific items (e.g. “In the Analogies test we changed question number 5, In what way are a peach and a plum alike, to … an orange and a banana.”) The IQ of 128 elementary schoolchildren on the translated WISC was 86.5. Roca also describes modifications to the translated Stanford-Binet after several experiments with some 900 children. 104 children tested on the final version exhibited an IQ of 90.6. Norms were also collected for the Goodenough Draw-a-Man (a nonverbal intelligence test) with over 2,000 children, ages 6-10, but for some reason Roca does not report this IQ, only stating with a discouraged vagueness “Puerto Rican children scored lower than American children.” Roca summarizes the findings by going beyond his evidence and asserting that lower scores are an inevitable by-product of cross-cultural testing: “There is no doubt that no matter how well an intelligence scale is adapted from one culture to another, there are cultural differences, which will make the children from the second culture score lower than those from the first” (Roca, 1955, p. 151).

Roca’s explanation for the low performance of Puerto Ricans looks less convincing in light of the Inter-American Tests, which might glibly be described as tests designed by Puerto Ricans that nevertheless give Americans much higher scores.

The Inter-American Tests were first developed in Puerto Rico in the 1940s in an attempt to build a test from the ground floor that is culturally appropriate and linguistically interchangeable for English-speaking Americans and Spanish-speaking Puerto Ricans. The tests were jointly designed and modified over time in a collaborative effort by American and Puerto Rican educators. However Puerto Rican schoolchildren still scored one standard deviation lower than American schoolchildren (Manuel, 1966).

__________________________________________

■ Ib. Experimental Studies

The low scores of Puerto Rican recruits were a problem for the US military during the Vietnam War. One military study found that an intensive 6-week English language-training course did little to improve scores on the Army Classification Battery, outside of narrow gains on the verbal sub-test. Their average IQ was 77.4, with few reaching the qualifying score of 90 (1972 Maier).

Children born to women with and without a history of oral contraceptive usage were tested with the Puerto Rican translation of the Wechsler Intelligence Scale for Children, which I just described (called the EIWN: Escala de Inteligencia Wechsler Para Niños). There were no significant differences between the two groups of children; their IQ was 83.8 (Fuertes et al, 1976).

Yet another study with the same WISC translation found an IQ of 100.9 for 100 elementary schoolchildren. Prewitt-Diaz & Gaisel (1980) suggest the low scores found by Roca were a fluke, and that the translated test (which was already being widely used) is actually appropriate for Puerto Rican children.

More recent studies with smaller numbers of children have used translations of newer versions of the Wechsler tests. While earlier papers explicitly scaled the scores on US norms, we should assume the later papers are reporting Puerto Rico norms unless otherwise indicated. A possible exception is Matias Carrelo & Zaidspiner (1990), who claim to test a small group of children using the WISC and WISC-Revised, rather than the EIWN and the (not yet available) EIWN-Revised. It is not clear if their sample (which includes both advantaged and disadvantaged children) somehow speak proficient English, or if the authors are using their own translation, but the resulting IQ is 99.9.

Several studies compare the performance of Puerto Rican islanders with Puerto Ricans living on the US mainland (Shellenberger, 1982; Rodriguez et al, 1985; Paez et al, 2007). I will describe the results of these tests in Section IIf.

In two studies the participants were given the Raven matrices in addition to the translated Wechsler tests. Pons et al (2008) gave the Spanish language versions of the WISC-R and the WAIS-III to a sample of 30 16-year-olds. They scored 95.5 and 105.8 on theses tests, but on the Standard Progressive Matrices they exhibited an IQ of 84.6, which is the same score attained by the SPM norm sample in 1977. Andrue et al (1991) tested Puerto Rican University students on the translated WAIS, on the standard English language WAIS-R, and on the Advanced Progressive Matrices. They scored 114.2 on the WAIS translation, 98.9 on the non-verbal subtests of the WAIS-R, and 105.4 on the APM. In other words, the reported IQ scores from the local test norms are about 1 standard deviation higher than the IQs from tests scored on US norms.

In another study (Alvarez et al, 1994) the SPM was given to 68 schoolchildren in an experimental educational program. Their IQ was 91.4.

An exception to low scores is a small study by Anderson & Souto (2005), who tested 11 San Juan Head Start children on a nonverbal subtest of the Kaufman Basic Intelligence Test. These 4 year olds had an IQ of 98.9.

__________________________________________

■ Ic. Estimated IQ for Puerto Rico

Excluding the 6 studies reporting IQ scores from local norms, and the elite sample of University students, this leaves 19 studies which, I believe, are adequate for estimating a national IQ score: Monroe (1921), Columbia University (1926), Vincenty (1930), Roca (1955), US Employment Service (1965), Manuel (1966), Green & Martinez (1967), Maier (1972), Fuertes et al (1976), Kahn et al (1977), Prewitt-Diaz & Gaisel (1980), Shellenberger (1982), Rodriguez et al (1985), Dunn et al (1986), Matias Carrelo & Zaidspiner (1990), Ortiz Colon et al (1993), Alvarez et al (1994), Anderson & Souto (2005), and Pons et al (2008).

This does not mean that I find all these scores equally attractive. Monroe (1921) seems like a questionably representative sample, but the IQ scores from the next four largest standardizations are nearly identical: General Aptitude Test Battery = 82.5; Wechsler Adult Intelligence Scale = 83.1; Raven Progressive Matrices = 82.8; Peabody Picture Vocabulary Test = 82.6.

This would give us a plausible estimate, but I feel it’s more neutral to include all reasonable samples. The median IQ from 19 studies is 84.6.

__________________________________________
__________________________________________

                 SECTION II: INTELLIGENCE TEST STUDIES OF PUERTO RICAN AMERICANS                  

Table II summarizes the data from 45 different studies of ethnic Puerto Ricans living on the mainland United States. The earliest research, from the 1920s, was conducted in Hawaii, which did not become a state until 1959.

Table II. IQ test scores of mainland Puerto Ricans

Admin	Age	N	Test	FC	IQ	SD	Reference
1923	12	23	NIT		72.7	20.8	Murdoch, 1925
			Army Beta		70	18.3
		40	Various		72.4		Porteus & Babcock, 1926
	6—18	276	Binet		70		Porteus, 1932
		241	PMT		76
1924	10—14	45	Unique		78.9		Smith, 1942
1935	9—14	240	AIPT		86.8		Armstrong et al, 1935
1938	10—15	198	Unique		76		Smith, 1942
1944	10—15	235	Pinter		83.7	15.4	Darcy, 1952
1950	11—15	108	CCF		81		Anastasi & Cordova, 1953
1951	5	50	DAM	7.8	87.9	21.2	Anastasi & deJesus, 1953
1953	25	47	Bellevue		85.6	19.5	Brown, 1960
1955	16	116	Lorge		78.6		Morrison, 1958
1955	7	60	CMMS		88.2		Morrison, 1958
1956	6—15	71	WISC	2.1	80.5	16.4	Talerico & Brown, 1963
1959	10	2868	Otis		85	11.9	Moriber, 1961
1960	Adult	49	WAIS	1.5	74.3	13.3	Shor, 1961
1965	7—18	4995	Unique		83.7		Coleman et al, 1966
1967	4	60	SBIS	9	86.6	10.8	Hertzig et al, 1968
1969	6	45	SBIS	9.6	90.6	13.6	Hertzig et al, 1971
1970	6—16	116	WISC	6.3	84.4	11.2	Thomas et al, 1971
1970	5—6	210	Vane	0.6	98	12.5	Walsh & D’Angelo, 1971
1973	7	50	WISC	7.5	77.6		Orsini, 1974
1974	7	123	WISC	7.5	74.8	10.4	Marmorale & Brown, 1975
1976	9	77	WISC	9.3	77.1	11.8	Marmorale & Brown, 1981
1975	9—13	25	WISC		85		Boone & Montare, 1976
1976	2—3	339	CIIS+SBIS	5.1	89.4	12.8	Morris, 1977
		508	Lorge	5.2	74.8		Sowell, 1978
1976	5—9	50	McCarthy(S)		70	14.9	Shellenberger, 1982
		30	McCarthy	1.2	81.3	12	Shellenberger & Lachterman, 1979
			WISC-R	0.6	76.2
1979	15—22	268	AFQT		88		Malloy, 2014a
1980	8	20	WISC-R	1.8	91.2		Oplesch & Genshaft, 1981
			EIWN	7.5	86.7
1980	18	249	Unique		89.8		Sahai, 1989
1984	6	123	PPVT	5.7	55.3		Hakuta & Diaz, 1985
			CPM	1.5	86.5
1984	12—14	80	EIWN-R	0.6	84.6	15.9	Rodriguez et al, 1985
1985	6	30	KABC	0.6	85.1		Glutting, 1986
1985	6	27	KABC	0.6	92.6		Glutting, 1986
1986	6—11	194	PPVT	6.3	58.6		Hakuta, 1987
			CPM	2.1	84.7
1988	6	30	W-J(S)	1.8	88.4	11.9	Prewitt Diaz & Riveria, 1989
			CMMS	4.8	84.7	6.3
			EIWN-R	1.8	92.3	11.9
1988	9—13	34	SPM	2.7	89.1	12.2	Loyola & McBride, 1991
			PPVT-R	2.1	49.2	15.8
			TVIP	0.6	87.6	15.6
1989	10—12	32	W-J(S)	2.1	91.7	9.1	Rodriguez & Prewitt Diaz, 1990
			EIWN-R	2.1	89.8	5.9
1989	8—16	20	WISC-R(S)		95.7	9.8	Tomayo, 1990
1990	9—18	334	SPM	3.3	74.7	10	Baudillo Mora de Jesus, 1990
1995	16	97	PPVT		92.9	14.6	Malloy, 2014b
1995	6-16	234	WISC-R	6.3	81.9		Leventhal et al, 2006
1995	18—89+	209	Wordsum		88.2	17.3	Malloy, 2014c
1997	6—10	108	WISC-R	6.9	94.6		Colon-Papazoglou, 1998
2001	4	310	WLPB-R	3	77.2	15.4	Paez et al, 2007
			WLPB-R(S)		75.7	13.5
2003	5	109	PPVT-III	1.8	87.4	12.8	Malloy, 2014d
2004	4	38	PPVT-III	2.1	75	17.7	Miccio et al, 2005
			TVIP	5.4	69.2	14.1

__________________________________________

■ IIa. Hawaii: 1920s-1930s

The United States annexed Hawaii in 1898, which is the same year that it acquired Puerto Rico from Spain. In 1899 Puerto Rican sugar plantations were devastated by hurricanes, and so hundreds of disrupted laborers migrated to the plantations of Hawaii.

Katherine Murdoch (1925) was the first to evaluate the children of these new migrants. In 1923 Murdoch administered the Army Beta and the National Intelligence Tests to 12-year-olds in Honolulu public schools (These children were all born in 1911, not long after the very first migrants arrived from Puerto Rico). The 23 Puerto Ricans had an IQ of 72.7 on the National Intelligence Test and 70 on the Army Beta. Murdoch lists the performance of 10 different ethnic groups. Filipinos were the only group with lower scores.

Stanley Porteus & Marjorie Babcock (1926, pp. 242-244) relay similar data from an unknown entity named Ernest B. Hoag. Hoag collected test data from nearly 5,000 Hawaiian public schoolchildren using the National Intelligence Test, the Morgan Test, and the Indiana Mental Survey. The 40 Puerto Rican children had an IQ of 72.4. Data is given for 12 different ethnic groups, and Puerto Ricans had the lowest average IQ. Further details about this study, like the ages of the children or the dates the tests were administered, appear to be lost to time, as I can’t locate any other published or unpublished sources for this data.

Stanley Porteus (1932) also reviews test results from schoolchildren referred to a Hawaiian psychological clinic. Porteus lists the average IQs of 12 different ethnic groups tested with a modified version of the Binet and with his own Porteus Maze Test. On the Binet 276 Puerto Rican children had an IQ of 70. Data is given for 12 different ethnic groups, and Puerto Ricans had the lowest IQ. The white children scored 97. 241 Puerto Ricans also had the lowest IQ on the Porteus Maze Test: 76. The white children had an IQ of 102.

In 1924, Stevenson Smith (1942) gave special mental tests to all the children in Hawaiian public schools between the ages of 10 and 15. In comparison to the white children, the Puerto Rican children had an IQ of 78.9. They had the worst performance of 12 different ethnic groups. In 1938 Smith administered the same tests to the new crop of public schoolchildren. This was one of the first discoveries of the Flynn Effect: IQ had risen by some 6 points in the intervening 14 years. But IQ had risen a little slower for the Puerto Rican children who now had an IQ of 76 relative to the white children.

__________________________________________

■ IIb. New York City: 1930s-1960

Significant waves of migration to the mainland United States first began after 1917 when the United States Congress approved the Jones-Shafroth Act, which granted U.S. citizenship to Puerto Ricans. New York City was the primary destination.

The first report on the intelligence of NYC Puerto Ricans became something of a target for local and subnational hostility. In 1935 the New York State Chamber of Commerce’s Special Committee on Immigration and Naturalization commissioned a study on the intelligence of Puerto Rican schoolchildren in East Harlem. The average IQ of 240 9-14 year-olds on the Army Individual Performance Test was 86.8; a comparison group of 400 white children scored 103.3 (Armstrong et al, 1935). This relatively low IQ score—and, probably more so, the report’s consequent recommendations against Puerto Rican citizenship and statehood—incited a backlash among local community educators and activists. The popular Commie Congressman Vito Marcantonio even ranted against the study before the House of Representatives (Thomas, 2010, p. 86). Academics, journalists, and government leaders in Puerto Rico also condemned the report—many offering parentheticals on the inferior intelligence of Americans. The New York Times reported:

A recent suggestion by a committee of the New York State Chamber of Commerce that statehood for Puerto Rico be held “in abeyance” because of the low intelligence indicated by a group of children … has angered political leaders and newspapers here …
[Commissioner of Agriculture] Mr. Menendez Ramos cites parts of a survey by… Columbia University… That … praised the intelligence of the island’s children, and said the level was higher than that of [American children] …
It is unfair and unscientific, Mr. Menendez Ramos asserts, to assume that the New York colony, composed of a working class faced with the difficulties of a new language and a new environment, represents the average of Puerto Rican mentality.
… The newspaper Mundo criticizes the report satirically … [suggesting] that the Chamber of Commerce study the excellent records of Puerto Rican students in American universities … the island’s college debaters have been uniformly victorious against American debaters (NYT Special Cable, 1936).

Harlem educator Leonard Covello organized a “Racial Committee” to critically evaluate the study. The group “argued that the Puerto Rican children’s poor performance was attributable entirely to their lack of familiarity with English” (Thomas, 2010, p. 86). Meanwhile, criticisms of the study from Puerto Rico’s white elite took on a slightly more insulted and chauvinistic tone (as already seen in the New York Times report). PR Assistant Commissioner of Education, Pedro Cebollero—apparently not convinced that the problem was “attributable entirely” to language differences—fumed that the children in the report were described as 76% colored, while the U.S. census described Puerto Rico as only 26% colored: “This fact is an evidence of [the report’s] absolute disregard of the principle of “representativeness”… Pinter points out that ‘all results show the negro decidedly inferior to the white on standard intelligence tests.’” (Cebollero, 1936, p. 5)

Of course, this is not to suggest that Cebollero’s race-based objection was totally invalid; I discuss the data on racial differences in Section IVa.

Subsequent studies also explored the language problem. In the 1940s Natalie T. Darcy (1952) gave Pinter intelligence tests to 235 bilingual Puerto Rican 5th and 6th graders. Their IQ was 83.7, but they performed better on the non-verbal than the verbal tests (79.6 vs. 87.8).

Anastasi & Cordova (1953) administered the non-verbal Cattell Culture Fair to 108 11-15 year old bilingual children. The children were split into two groups and either given the test with the instructions delivered in Spanish or English. The performance did not differ between the two groups; the paper reports a median IQ of 70 (my own conversion of the raw scores on the Cattell norms gives a mean IQ of 66). The authors state that the unusually large standard deviation of the CCF suggests that the score should be inflated to 81 to align with the results from more typical tests with a standard deviation of 15 or 16. I accept their correction: scores of about 81 are typical for Puerto Ricans during this place and time.

Anastasi & deJesus (1953) gave another non-language test, The Goodenough Draw-a-Man, to 50 5-year-old Puerto Ricans attending day care. The instructions were delivered in Spanish by the Puerto Rican examiner. The IQ of these young children was 87.9.

Brown (1960) and Talerico & Brown (1963) both report on the intelligence of Puerto Rican psychiatric patients at Mount Sinai Hospital. In the former report, 47 patients seen between 1947 and 1959 (average age 25) were given the Wechsler-Bellevue. The average IQ was 85.6. In the latter report, WISC data is reported for 71 children, ages 6-15, examined between 1952 and 1961. Their mean IQ was 80.5.

Magda Denes Shor (1961) looked at another kind of disadvantage. Shor administered the WAIS to 49 Puerto Rican men in NYC homeless shelters. Their IQ was 74.3. A comparison group of 50 white homeless men had an IQ of 96.5.

Puerto Rican migration to the mainland picked up momentum after WWII with the advent of cheap air travel (Table III). Between 1950 and 1960 the Puerto Rican population of NYC tripled: from almost 200,000 to over 600,000. The late 50s to mid-60s featured the three most intensive surveys of the Puerto Rican American school population ever conducted (Morrison, 1958; Moriber, 1961; & Coleman et al, 1966).

“The Puerto Rican Study” (Morrison, 1958) was one of the most lavishly funded educational studies of the time. Teams of scientists scrutinized the Puerto Rican school population of NYC between 1953-1957. Thousands of children were tested. Unfortunately, the final report only includes sample sizes and intelligence test data from two smaller experimental studies. The Lorge-Thorndike non-verbal intelligence test was given to 116 high school students who had been on the US mainland for various lengths of time. Their average IQ was 78.6. In another study, 60 first grade students with variable success in learning English were given the Columbia Mental Maturity Scale. Their average IQ was 88.2.

In contrast to Morrison (1958), a similar survey of Puerto Ricans in NYC public schools during 1959-1960 reports intelligence test data for numerous students (Moriber, 1961). 1,034 6th graders and 1,844 3rd graders were given the Otis Quick-Scoring Mental Ability Test. The mean IQ of all 2,868 students was 85.

__________________________________________
__________________________________________

■ IIc. Coleman and Beyond: 1960s-1980s

Studies of Puerto Rican Americans after 1960 begin to include some of the growing migrant populations outside of New York City. The largest and most representative sample of Puerto Rican Americans remains The Coleman Report (Coleman et al, 1966). James S. Coleman’s national survey includes scores from special mental ability tests for 4,995 Puerto Ricans in grades 1 through 12. In comparison to the white students this national sample of Puerto Ricans had a mean IQ of 83.7.

Several studies from this time period also looked at the IQ of younger pre-school age children. Hertzig et al (1968) gave the Stanford-Binet intelligence test to 60 Puerto Rican 3-year-olds from primarily working class backgrounds and to 116 white 3-year-olds from middle class backgrounds. Depending on their primary language, the PR children were tested with either English or Spanish versions of the test. The Puerto Rican children had an IQ of 86.6, the white children had an IQ of 113.4.

In another study, 339 low income children, ages 2-3 years, were recruited for testing from the Mount Sinai Pediatric Clinic (Morris, 1977). The children were either tested with the Cattell Infant Intelligence Scale, if they were 20-29 months old, or with the Stanford-Binet, if they were 30 months or older. Their mean IQ on these tests was 89.4.

Walsh & D’Angelo (1971) found higher scores for 210 5-6 year-olds enrolled in a summer head start program. These disadvantaged children had an IQ of 98 on the Vane Kindergarten Test.

The 3-year-old children from Hertzig et al (1968) were also tested at age 6 (Hertzig et al, 1971). Their IQ on the Stanford-Binet was 90.6, while the IQ of the middle class white children was 117.6. The size of the IQ gap between the two groups was virtually identical at age 6. Thomas et al (1971) appears to test these same children as well as their older siblings. The 6-16 year-olds in this sample had a mean IQ of 84.4 on the Wechsler Intelligence Scale for Children (WISC), although the study found that different examiners were associated with higher and lower scores.

A number of other experimental studies during this time period also used the WISC. Barbara Orsini (1974) tested 50 bilingual Puerto Rican 7 year-olds who were born and raised in Harlem. They had a WISC IQ of 77.6.

Marmorale & Brown (1975) tested 123 first graders from NYC public schools. These children had a WISC IQ of 74.8. Many of these children were followed up at the end of 3rd grade and tested again (Marmorale & Brown, 1981). Their IQ at follow-up was 77.1.

Boone & Montare (1976) tested a small group of 9-13 year-olds from Newark, New Jersey on the vocabulary subtest of the WISC. Their IQ on this subtest was 85.

Shellenberger & Lachterman (1979) tested 30 young children from Lancaster, Pennsylvania on the McCarthy Scales of Children’s Abilities as well as on the information and comprehension subtests of the WISC-Revised. On the McCarthy Scales their IQ was 81.3. On the WISC subtests their IQ was 76.2.

Oplesch & Genshaft (1981) tested a small group of bilingual 8 year-olds from a small city in northern Ohio on the WISC-Revised and on the EIWN (the Puerto Rican translation of the WISC). The children had an IQ of 86.7 on the Spanish-language WISC, and an IQ of 91.2 on the English-language WISC.

And finally, Maria D. Rodriguez (1985) tested 80 pupils who had recently migrated from Puerto Rico to Hartford, Connecticut on the Spanish language WISC-Revised (EIWN-R). Their IQ was 84.6.

In the late 1970s the popular economist and author, Thomas Sowell, in conjunction with the Urban Institute, collected over 70,000 historical intelligence test records from schools all over the country, for the purposes of comparing ethnic groups over time.

Table IV.

Tables IV and V are taken from Sowell (1978, p.214). Advantages of Sowell’s Urban Institute data are the large sample sizes and rare time trends for a number of ethnic groups that virtually disappeared from the IQ testing literature after the 1940s. Table IV shows test data for some 1,453 Puerto Ricans collected across 5 decades. Two features of the data in Table IV, however, render it unusable for my dataset. The two issues are only a problem in combination. The first issue is that Sowell’s numbers are the averaged data from multiple unknown IQ tests. The second issue is that the Flynn Effect (unknown to Sowell and most other researchers at the time) has artificially inflated the scores on tests over time. This means that Sowell’s time trends are nearly worthless, because scores are continually inflating on outdated test norms. We could otherwise correct for this (although not precisely, because the numbers are listed by decade instead of by year) if the data were all from a single test, but the norms for different tests were collected at very different times. A lesser compatibility issue is that Sowell reports the medians.

Table V.

However, this means I am able to use the data from Table V, which reports mean scores from a single test—the Lorge-Thorndike—which was standardized in 1954. So if we assume the tests were administered somewhat regularly over time, we can make a justifiable subtraction of 3.3 points for the 1960s (for 11 years of norm inflation: 1954-1965) and 6 points for the 1970s (20 years of norm inflation: 1954-1974). This gives us a Lorge-Thorndike IQ of 74.8 for 508 schoolchildren during the 1960s and 1970s.

In the future I will probably try and obtain the Urban Institute dataset, and maybe the proper Flynn Effect adjustments can be made for all 1,453 Puerto Ricans.

The final datapoint from the 1970s is my own analysis of the National Longitudinal Survey of Youth (Malloy, 2014a), which, you might remember from The Bell Curve, subjected a cohort of 15-22 year-olds to the Armed Forces Qualification Test. The IQ of 268 Puerto Ricans on the AFQT was 88.

In 1980, yet another national survey helmed by James S. Coleman, known as High School & Beyond, evaluated high school seniors on special mental ability tests. The IQ of 249 Puerto Rican 18 year-olds was 89.8 (Sahai, 1989).

Hakuta & Diaz (1985) and Hakuta (1987) report Colored Progressive Matrices data from Spanish-speaking children in a bilingual program in New Haven, Connecticut. The former study includes data for 123 6-year-olds, who had a CPM IQ of 86.5. The latter study has CPM data for 194 children, ages 6-11, from several different cohorts. The IQ from this sample was 84.7. All of these children were also given the PPVT, but the scores were in the 50s, indicating a very poor familiarity with the English language.

Joseph J. Glutting (1986) tested Delaware kindergarteners on the Kaufmann Assessment Battery for Children (K-ABC). The Puerto Rican children were divided into English- and Spanish-dominant language groups. The full-scale IQ of the English dominant group was 92.6. The non-verbal IQ of the Spanish-dominant group was 85.1.

Prewitt Diaz & Riveria (1989) tested 30 recently migrated 6-year-olds on the Spanish translation of the Woodcock-Johnson Psychoeducational Battery, on the Columbia Mental Maturity Scale, and on the EIWN-R (the Spanish WISC). IQ on the Spanish Woodcock was 88.4, on the Spanish WISC it was 92.3, and on the nonverbal CMMS it was 84.7.

Loyola & McBride (1991) gave several tests to 34 5th and 6th graders in bilingual classes in Philadelphia, Pennsylvania. On the Standard Progressive Matrices they had an IQ of 89.1. On the PPVT-R they had an IQ of 49.2, but when the items were translated into Spanish they scored 78.5. On the PR adaptation of the test—the TVIP—they had an IQ of 87.6.

Rodriguez & Prewitt Diaz (1990) tested 32 recently migrated 4th graders on the Spanish translation of the Woodcock-Johnson and on the EIWN-R. IQ on the Woodcock was 91.7 and 89.8 on the EIWN-R.

Finally, Tomayo (1990) tested 20 children drawn from Catholic schools in Chicago. He experimentally adapted the WISC-R vocabulary sub-test by substituting the English words with words that Spanish-speakers use with equal frequency. IQ on this experimental subtest was 95.7.

__________________________________________

■ IId. Recent Data: 1990s-2000s

Table II lists eight samples of Puerto Ricans tested since 1990. Rafael Baudillo Mora de Jesus (1990) gave the Standard Progressive Matrices to 334 children in the fifth, seventh, and ninth grades. Their average IQ was 74.7.

I analyzed data from the National Longitudinal Study of Adolescent Health (Malloy, 2014b). This is a representative survey of high school students in 1995. The Puerto Rican teens had a PPVT IQ of 92.9.

I also analyzed data from the General Social Survey, which has collected nationally representative social science data either annually or biennially since 1972. The GSS features a small vocabulary quiz—The Wordsum—that functions as an abbreviated IQ test. 209 Puerto Ricans of all adult ages have been tested since 1974. But the average person was a 35 year old, born in 1960 and tested in 1995. The mean IQ of Puerto Rican adults in the GSS is 88.2 (Malloy, 2014c).

In addition to the NLSY, the GSS, and Add Health, I analyzed the SWEEP dataset (Study of State-Wide Early Education Programs). This is combined data from two major studies on the effects of pre-k programs in 11 states. This included PPVT data for 109 Puerto Ricans who were born in 1998 and tested at age 5 in 2003. Their IQ was 87.4.

Leventhal et al (2006) analyzed data from the PHDCN (Project on Human Development in Chicago Neighborhoods). 209 Puerto Ricans, ages 6-16, were given the vocabulary subtest of the Wechsler Intelligence Scale for Children. Their IQ was 81.9.

Ailin Colon-Papazoglou (1998) looked at 108 bilingual Puerto Ricans who were born in the US and were attending grades 1-4 in a Brooklyn public school. On four sub-tests from the WISC-R they had an IQ of 94.6.

Paez et al (2007) likewise looked at 310 bilingual, mostly US born, Puerto Rican 4-year-olds attending pre-K programs in Massachusetts and Maryland. They were given the Woodcock Language Proficiency Battery-Revised, as well as the Spanish version of the same test. Their IQ on the WLPB-R was 77.2, while their IQ on the Spanish version was 75.7.

The final study for Puerto Rican Americans—Miccio et al (2005)—looked at 38 bilingual 4 year-olds attending Head Start programs in Pennsylvania. The children were given the PPVT as well as the Spanish translation of the same test (TVIP). Their IQ was 75 on the PPVT and 69.2 on the TVIP.

__________________________________________

■ IIe. Estimated IQ for Puerto Rican Americans

Simply eyeballing the 45 studies in Table II, we can readily see that not one study shows an average score of 100 or higher, and that most scores cluster between the upper 70s and the lower 90s. To get a better average we should exclude less representative studies. I’m excluding all of the Hawaii studies from the estimate, because this early sub-population had a noticeably lower IQ. Brown (1960), Talerico & Brown (1962), and Shor (1961) are out because they are disadvantaged samples. For all studies with overlapping participants, I’m selecting the reference with the largest sample size. I’m excluding averages from all Spanish normed tests like the TVIP and the EIWN, since the comparisons involve more assumptions. I’m also excluding the PPVT results from Hakuta & Diaz (1985), Hakuta (1987) and Loyola (1991), because they indicate near total English illiteracy for these samples. This leaves 30 out of 45 studies that I believe are adequate for estimating an intelligence score. The median of 30 studies gives us an IQ of 84.7 for Puerto Rican Americans.

The current average appears to be somewhat higher. The median IQ of 19 samples from the 1930s-1970s is 83.7. The median IQ of 14 samples from the 1980s-2000s is 87.4.

__________________________________________

■ IIf. Comparison Groups

Averaging the results from multiple studies suggests that the IQ of Puerto Ricans, both on the island and the mainland, has been about 1 standard deviation lower than the IQ of white Americans over time.

Most of the IQ values for Puerto Ricans in Tables I and II are in relation to the reference norms for various intelligence tests. One disadvantage of this method is that norms deteriorate over time, and the results are “Flynn Corrected” based on (possibly flawed) assumptions about how the reference population would perform on the test at the time of the study. Another issue is that norm groups are often built to reflect the average ability of the U.S. population. This means the Puerto Rican IQ scores are not strictly in relation to white Americans, but in relation to a population of whites, blacks, Asians, and Hispanics (including Puerto Ricans).

Another method of estimating IQ that avoids these pitfalls is to look at the studies that compare whites and Puerto Ricans directly.

Table VI includes 25 studies where tests have been administered to mainland Puerto Ricans and other groups at about the same time and/or location. Some are regional comparisons of, e.g., kids from the same schools or neighborhoods. Some are national surveys, where representative samples from diverse locations are given the same test during a specific time period.

In 19 samples Puerto Rican Americans were compared with whites, in 16 samples they were compared with African Americans, and in 3 samples they were compared with Puerto Rican islanders. The d column is the standardized mean difference between Puerto-Rican Americans and the comparison group. Negative values indicate higher scores for mainlanders.

Table VI. Direct IQ Comparisons: Mainlanders, Islanders, Whites, & Blacks

Mainlanders		Islanders				Whites				Blacks
N	IQ		N	IQ	d		N	IQ	d		N	IQ	d		Test	Reference
23	73						57	100	1.8						NIT	Murdoch, 1925
40	72.4						442	100.7	1.89						Various	Porteus, 1926
240	86.8						400	103.3	1.1						AIPT	Armstrong, 1935
45	78.9						337	100	1.41						Unique	Smith, 1942
198	76						773	100	1.6						Unique	Smith, 1942
50	95.7						50	102.3	.44		50	100.1	.29		DAM	Anastasi, 1953
2868	85										2,679	89.9	.33		Otis	Moriber, 1961
80	92.8										80	95.8	.20		Unique	Lesser, 1965
4995	83.7						>9999	100	1.09		>9999	84.6	.06		Unique	Coleman, 1966
60	95.6						116	122.4	1.79						SBIS	Hertzig, 1968
45	100.2						94	127.2	1.8						SBIS	Hertzig, 1971
50	85.1										50	93.9	.59		WISC	Orsini, 1974
123	82.3						82	108.3	1.73		61	94.1	.79		WISC	Marmorale, 1975
77	90.1						44	109.4	1.29		48	94.8	.31		WISC	Marmorale, 1981
25	94.7						52	102.8	.54		55	100.3	.37		WISC	Boone, 1976
339	94.5										136	97.8	.22		CIIS+SBIS	Morris, 1977
268	88						7027	100	.80		3022	84.3	-.25		AFQT	Malloy, 2014a
50	70		50	69	-.07										McCarthy(S)	Shellenberger, 1982
80	85.2		51	91.8	.44										EIWN-R	Rodriguez, 1985
27	93.2						65	103.9	.71		24	90	-.21		K-ABC	Glutting, 1986
249	92.8						9818	103	.68		1698	92.8	0		Unique	Sahai, 1989
97	97						4114	104.1	.47		1546	94.2	-.19		PPVT	Malloy, 2014b
234	88.2						361	97.7	.63		882	87.7	-.03		WISC-R	Leventhal, 2006
209	88.2						>9999	100	.79		3867	90.3	.14		Wordsum	Malloy, 2014c
109	89.2						1275	99.8	.71		607	88	-.08		PPVT	Malloy, 2014d
310	78.7		152	85.7	.47										WLPB-R(S)	Paez, 2007

__________________________________________

Three studies compare Puerto Rican islanders with Puerto Rican mainlanders. Shellenberger (1982) gave the Spanish language version of the McCarthy Scales of Children’s Abilities to 50 bilingual first- and second-graders in Lancaster, Pennsylvania and 50 matched children in Vista Hermosa, Puerto Rico. The mainlanders did about 1/10th of a standard deviation better (.07σ). The racial composition of the two samples was slightly different: 16% of the island children were described as black, while 100% of the U.S. children were described as white or brown.

Rodriguez et al (1985) gave the Spanish language version of the WISC-R to 51 adolescents in the Central Mountain region of PR, and 80 adolescents who had recently migrated to Hartford, Connecticut. The U.S. migrants had an IQ that was almost ½ of a standard deviation lower (.44σ). This suggests that migrants have lower initial IQ scores than those who remain on the island.

Paez et al (2007) gave the Spanish language version of the Woodcock Language Proficiency Battery-Revised to 319, mostly U.S born, 4 year-olds in Massachusetts and Maryland, and to 144 4 year-olds in Puerto Rico. Scores were nearly ½ standard deviation higher in Puerto Rico (.47σ). It doesn’t seem too shocking that U.S. born Puerto Ricans have lower Spanish language ability than children born in a Spanish speaking territory.

Sixteen of the studies directly compare African-Americans and Puerto Ricans. This data suggests that blacks had a higher IQ than Puerto Ricans until the late 1970s when the gap abruptly flipped. The giant national sample by Coleman et al (1966) and the giant NYC sample by Moriber (1961) both found higher IQs for blacks. But by the time of the National Longitudinal Survey of Youth in 1979, this trend had reversed. The 9 studies between 1951-1976 found an IQ advantage of 5.3 points for blacks (.35σ), but the 7 studies between 1979-2003 found an IQ advantage of 1.4 points for Puerto Ricans (0.08σ).

Eighteen of the studies directly compare whites and Puerto Ricans. The older Hawaiian studies show a distinctive gap, so once again those samples are omitted from the analysis. Duplicate samples are also omitted. The median white-Puerto Rican gap from the remaining 13 studies is .71σ, giving Puerto Rican Americans an IQ of 89.4.

This data also suggests that the white-Puerto Rican gap narrowed during the 1970s, just as the black-Puerto Rican gap was reversing. The Coleman Report showed a W-PR gap that exceeded 1 SD in the mid-1960s, but High School & Beyond showed a smaller gap by 1980. The seven studies from 1935-1979 give an IQ of 84, the six studies from 1985-2003 give an IQ of 90.

(With these studies we see a similar narrowing of the 1 SD black-white gap over the same brief time period. The six studies conducted from 1985-2003 give an IQ of 89.1 for blacks. Last year I also brought to light data that contradicts a narrowing of the black-white gap.)

The comparison group studies confirm that the Puerto Rican Americans IQ gap has been about 1 to 2/3 of a standard deviation over time. From the 1930s-1970s the gap was a little larger than 1 SD. Studies from the late 70s onward suggest a contemporary IQ that is between 87.4 and 90, but do not suggest a trend toward further narrowing.

I analyze these apparent trends over time in more depth in Section IVe with a larger panel of cohort-based samples.

__________________________________________
__________________________________________

                 SECTION III: MIGRATION, LANGUAGE, & ADJUSTMENT                  

This section explores how issues related to language and acculturation relate to test performance differences between whites and Puerto Ricans, and between mainlanders and islanders.

__________________________________________

■ IIIa. Nativity

For many Puerto Ricans English is a second language. Migrants come to the U.S. mainland at many different ages, and older children are less able to adapt to testing in another language. How much of the gap between whites and Puerto Ricans is accounted for by this migratory handicap? Recall that the “Racial Committee” of activists and educators who challenged the first study of intelligence in NYC “argued that the Puerto Rican children’s poor performance was attributable entirely to their lack of familiarity with English” (Thomas, 2010, p. 86).

In Thomas Sowell’s summary of the Urban Institute data (Sowell, 1978, pp. 214-215), he asks why Mexicans and Puerto Ricans had such stable deficits on intelligence tests over so many decades, in contrast to European immigrant groups who rapidly closed their own IQ gaps with the native population (… when such gaps even existed). He offers four possible explanations for this: 1) genetic differences, 2) low socioeconomic status, 3) enduring Spanish language usage in Hispanic homes, and 4) high return migration that inhibits acculturation. Thankfully, Sowell quickly assures us that “The genetic explanation is easily disposed of, however.” He offers no further defense of this statement but simply points to Table VII below, which is reproduced from Morrison (1958).

Table VII.

Time spent on the U.S. mainland and in U.S. schools is associated with improvements on even non-verbal tests. The children in Table VII were in the 10th grade, so the small group with the IQ of 93 had almost all of their education on the mainland. One more group from Morrison (1958, p. 179) provides a similar picture: an 8th grade cohort of children that were born and partially schooled on the island had a (non-verbal) IQ of:

73.6 if they attended mainland schools for 2 years.
79.9 if they attended mainland schools between 2-4 years.
83.2 if they attended mainland schools for more than 4 years.

Of course, the most schooling you can have on the mainland is all of it, and the most time you can spend on the mainland is your whole life. How do Puerto Ricans born and raised on the mainland perform? Morrison (1958, p 180) also reports on a cohort of fourth-graders. They had an IQ of:

76.3 if born on the island, and partly schooled on the U.S. mainland.
79.6 if born on the island and exclusively schooled on the U.S. mainland.
88.4 if born in the U.S. and exclusively schooled on the U.S. mainland.

So this gives us an upper bound figure of 88.4 for the limits of migration-related handicaps.

The study of 3rd- and 6th-grade NYC students by Moriber (1961), also includes data for nativity, and has an additional advantage of reported sample sizes. The IQ of the 1,357 Puerto Rican students born on the island was 82.3. The IQ of the 1,521 students born on the U.S. mainland was 87.

A third study with nativity data is Leventhal et al (2006). The average IQ of Puerto Ricans in this Chicago cohort was 81.9: the IQ of the 141 migrant children was 81, while the IQ of the 93 native children was 83.3.

A fourth study by Wen-Jui Han (2006) looks at the Early Childhood Longitudinal Survey—a nationally representative sample. Puerto Rican scores on special math and reading tests are reported for kindergarten-1st grade. 15 first generation children born (but not schooled) in Puerto Rico had an IQ of 85.3. 59 second generation children born on the mainland had an IQ of 89.5.

More generational data comes courtesy of the General Social Survey, which includes an abbreviated IQ test and asks respondents questions about themselves, their parents, and their grandparents:

Table VIII summarizes the data from five studies with birth origin comparisons. The average IQ of island-born migrants is 81.7. The average IQ of mainland born Puerto Ricans is 87.2. (This is close to the average of 86.1 from the two studies with strictly NYC born samples: Orsini, 1974; Colon-Papazoglou, 1998). The average gain from being born on the mainland is 5.6 IQ points. This shows us that the language difficulties and U.S. schooling deficits of first generation migrants can account for very little of the Puerto Rican test gap. A large IQ difference is observed for Puerto Ricans born and raised entirely on the mainland.

Table VIII. Mainlander IQ by Birth Origin

Island-born	Mainland-born	Difference	Reference
78	88.4	10.4	Morrison, 1958
82.3	87	5.3	Moriber, 1961
81	83.3	2.3	Leventhal, 2006
85.3	89.5	4.2	Han, 2006
82	88	6	Malloy, 2014c
(81.7)	(87.2)	(5.6)

__________________________________________

■ IIIb. Achievement Tests

The National Assessment of Educational Progress (NAEP) is a periodic survey of US student performance on various academic achievement tests. Representative samples of 1000s of students are tested in every state. These are achievement tests, not intelligence tests; primarily gauging learned information rather than reasoning ability. However, achievement and intelligence test performance have substantially overlapping etiologies and are highly correlated.

In 2003 and 2005 Puerto Rico was included in some of the student assessments. Mathematics tests were given to children in the 4th and 8th grade. Subtests include algebra, statistics, geometry, measurement, and number scales. It is not obvious from the island’s performance that Puerto Rican schools teach any of these things. Their average AQ (Achievement Quotient) was 63, some 2½ standard deviations lower than American students!

In contrast, the achievement test gaps for U.S. ethnic groups—sharing similar educational experiences—are about as wide as the corresponding IQ gaps. For example, averaging together the 2009 NAEP results on math and reading tests for the 4th, 8th, and 12th grades gives us an AQ of 86.8 for African-Americans (.88σ). The AQ of mainland Puerto Ricans, likewise resembling their IQ, is 89.7 (.69σ).

Unfortunately, NAEP sample sizes are not reported through the online Data Explorer. But abundant sample sizes are available for America’s favorite college entrance exam: the SAT Reasoning Test.

SAT scores were first disaggregated by race and ethnicity in 1976, so I’ve compiled all Math and Verbal scores for Puerto Rican Americans across the 38 years from 1976-2013. An Excel file of this data is available below in Supplement 1. My sources are as follows: Kobrin et al (2007) report sample sizes, means, and standard deviations for both whites and mainlanders for the 20 years 1987-2006. Sample sizes, means, and standard deviations for 2007-2013 were taken from annual SAT stat sheets released by the College Board. Mean scores for whites and Puerto Ricans for the years 1976-1984 are reported by O’Neill & Sepielli (1985). Means and standard deviations for the year 1985 are reported by Burton et al (1988). Burton also provided sample sizes for the years 1985, 1980, and 1976. Pennock-Roman (1993) provides standard deviations for the year 1980. And lastly, Duran et al (1985) report the sample size for Puerto Ricans in 1983.

Race/ethnicity data was not collected in 1986, so that gives us math and verbal scores for 37 years, sample sizes for 31 years, and complete standard deviations for 29 years.

As the name suggests, the SAT Reasoning Test (originally the Scholastic Aptitude Test) can be considered an IQ test, and there is empirical support for this characterization (Frey & Detterman, 2004). It is classified as an achievement test here for convenience. The SAT scores are mainly distinctive from the intelligence test scores (Table II) in that college-bound test-takers comprise an elite sample of the American and Puerto Rican American population. Otherwise, we might consider this better than the IQ data, given the vastly superior sample sizes: data is reported for 476,376 Puerto Rican test-takers in Supplement 1—although sample sizes are missing for 6 years, which indicates the true number is over 500,000.

             Supplement 1. White & Puerto Rican American SAT scores: 1976-2013             

Supplement 1 reports the yearly standardized mean differences between whites and Puerto Rican mainlanders on the SAT-Verbal (V d), on the SAT-Math (M d), and on the total SAT (d).

The mean 37 year Verbal difference is .775σ, the Math difference is .786σ, and the total difference is 0.784σ. This translates to an AQ score of 88.2.

Figure 1 shows the timeline for the white-Puerto Rican American gap. There was a gradual decrease in the gap from 1976-1990, and then there is a flat trendline for 1991-2013.

The early change in the gap was very modest: the average AQ for the first 10 years (1976-1985) was 87.7, while the average AQ for the last 10 years (2004-2013) was 88.6.

Figure 1. Standardized mean difference between whites &
Puerto Rican Americans on the SAT: 1976-2013

Every year the College Board has been releasing stat sheets for SAT scores in each individual US state. In 2012 and in 2013 the College Board released a “State Profile Report” for Puerto Rico. The 5,490 Puerto Rican test-takers sent most of their applications to mainland Universities. Their scores were virtually identical to those of mainland Puerto Ricans during the same years: The mainlander AQ was 88.6 and the islander AQ was 88.2. Scores are broken down by family income, and they do not exceed the US white average at any level: the 6% of Puerto Ricans with a family income over $200,000 have an AQ of 98.1.

__________________________________________

■ IIIc. Language

To what extent do language deficits contribute to the Puerto Rican ability gap? One method for disentangling this problem is to compare performances on tests that measure language ability and tests that don’t.

The verbal and nonverbal abilities of mainland Puerto Ricans are shown in Table IX.

Table IX. Puerto Rican Americans: Verbal & Nonverbal IQ Scores

Nonverbal	Verbal	Difference	Test	Reference
76			PMT	Porteus, 1932
79	78.8	.02	Unique	Smith, 1942
79.9	74.1	5.8	Unique	Smith, 1942
87.8	79.6	8.2	Pinter	Darcy, 1952
81			CCF	Anastasi & Cordova, 1953
87.9			DAM	Anastasi & deJesus, 1953
89.6	84.3	5.3	Bellevue	Brown, 1960
78.6			Lorge	Morrison, 1958
88.2			CMMS	Morrison, 1958
88.5	80	8.5	WISC	Talerico & Brown, 1963
84	83.3	.07	Unique	Coleman et al, 1966
86.5	79.1	7.4	WISC	Thomas et al, 1971
105	89.8	15.2	Vane	Walsh & D’Angelo, 1971
86.9	71.8	15.1	WISC	Orsini, 1974
83.5	69.8	13.7	WISC	Marmorale & Brown, 1975
85.1	78.5	6.6	WISC	Marmorale & Brown, 1981
	85		WISC	Boone & Montare, 1976
94.8	78.7	16.1	McCarthy	Shellenberger & Lachterman, 1979
	76.2		WISC-R
100.7	84.3	16.4	WISC-R	Oplesch & Genshaft, 1981
100.8	75.3	25.5	EIWN
94.5	89.2	5.3	Unique	Sahai, 1989
	55.3		PPVT	Hakuta & Diaz, 1985
86.5			CPM
85.1			KABC	Glutting, 1986
98			KABC	Glutting, 1986
	58.6		PPVT	Hakuta, 1987
84.7			CPM
84.7			CMMS	Prewitt Diaz & Riveria, 1989
105.5< /div>	79.1	26.4	EIWN-R
89.1			SPM	Loyola & McBride, 1991
	49.2		PPVT-R
	87.6		TVIP
101.8	82.1	19.7	EIWN-R	Rodriguez & Prewitt Diaz, 1990
	95.7		WISC-R(S)	Tomayo, 1990
74.7			SPM	Baudillo Mora de Jesus, 1990
	92.9		PPVT	Malloy, 2014b
	81.9		WISC-R	Leventhal et al, 2006
	88.2		Wordsum	Malloy, 2014c
99.7	89.5	10.2	WISC-R	Colon-Papazoglou, 1998
	77.2		WLPB-R	Paez et al, 2007
	75.7		WLPB-R(S)
	87.4		PPVT-III	Malloy, 2014d
	75		PPVT-III	Miccio et al, 2005
	69.2		TVIP

__________________________________________

The Wechsler intelligence tests include a Performance IQ, a Verbal IQ, and a Full Scale IQ. These three values are usually reported. For six different studies using the WISC, the average difference between the verbal and performance scales was 11.3 points. Average performance on the WISC (excluding the psychiatric sample) was 85.8, with a Performance IQ of 91.6, and a Verbal IQ of 80.

On three occasions mainlanders were given the Spanish version of the WISC, normed in Puerto Rico (the EIWN), and the verbal/performance difference on this test is 23.9 points. This appears to result from mainland children having weaker Spanish language skills, and seemingly also because the Performance sub-tests were simplified in the adaptation process to bring the PR scores closer to the WISC norms.

For representativeness I exclude all the same samples that I did in Section IIe. I also cut all the samples without clear verbal or nonverbal scores. This leaves 25 studies. Using medians: the average IQ is 85.2—the average nonverbal IQ is 87.8, and the average verbal IQ is 82.6. This is a difference of 5.2 points.

How has this changed over time? For 12 studies between 1944-1976, the average IQ was 85.2, with a nonverbal IQ of 87.8 and a verbal IQ of 82.6.

For 14 studies between 1980-2004, the average IQ was 88.9, with a nonverbal IQ of 90.4 and a verbal IQ of 87.4.

This represents an overall IQ gain of 3.7 points. This gain was 2.6 points on nonverbal tests and 4.8 points on verbal tests; so gains were almost twice as large on verbal tests. The verbal/nonverbal test score difference of 5.2 points has consequently shrunk to 3 points.

Other data suggest that the verbal ability deficit has disappeared. The rise in mainlander SAT scores from 1976-1990 was entirely on the verbal tests. During the first 10 years (1976-1985) the verbal AQ was 87.4 and the math AQ was 88. For the last 10 years (2004-2013) the verbal AQ had risen to 89.3, while the math AQ was still 88. So they didn’t just close the gap, they flipped it: they’re now performing higher on the verbal test than the nonverbal test. College-bound seniors are an elite population of Puerto Ricans, but the NAEP, which tests the general population of students, shows the same thing: The average gap on the math exam is even larger than the gap on the reading exam (.72 vs. .67).

If there is a remaining skew toward nonverbal tests, does this reflect residual problems with the English language or do Puerto Ricans simply have a distinctive profile of mental abilities? Some research suggests it is an ethnic pattern of cognitive strengths and weaknesses.

Lesser, Fifer, & Clark (1965) looked at the mental abilities of Chinese, black, Jewish, and Puerto Rican 7 year-olds in New York City. The researchers adapted their tests from the Hunter College Aptitude Scales for Gifted Children (this study is not included in Table II, because there are no comparison scores for a general sample of whites). The tests were translated into Spanish for the Puerto Rican children, but not otherwise modified. Figure 2 shows the performance of the four ethnic groups on the four subtests. Puerto Ricans had the worst overall performance on the tests but scored higher than blacks on the number and spatial scales.

Figure 2. Ability Patterns for Four Ethnic Groups

Lower class children from every ethnic group had lower scores, but had the same pattern of mental abilities as their co-ethnics from the middle class. Upper class and lower class Jewish children both performed best on the verbal scales and worst on the spatial scales. Upper class and lower class Chinese children shared the opposite pattern: worst on the verbal scales, best on the spatial scales.

Upper class and lower class Puerto Rican children both performed worst on the verbal and best on the spatial scales. Since the tests were in Spanish, this suggests that their relative deficit on verbal tests was not the consequence of a secondary language.

We also have some data to test whether a skew toward nonverbal cognitive abilities is a cultural trait: the verbal and nonverbal abilities of Puerto Rican islanders are shown in Table X.

Table X. Puerto Rico: Verbal & Nonverbal IQ Scores

Nonverbal	Verbal	Difference	Test	Reference
97			Holley	Monroe, 1921
90			Pinter	Columbia University, 1926
84.2			Detroit	Vincenty, 1930
81.5			Pinter
80.9			Dearborn
85.5	84.5	1	WISC	Roca, 1955
82.6	88.2	-5.6	GATB	US Employment Service, 1965
	85		I-AT	Manuel, 1966
86	87	-1	WAIS	Green & Martinez, 1967
83			ACTB	Maier, 1972
80.6			CPM	Kahn et al, 1977
84.6			SPM	Kahn et al, 1977
102.8	81.2	21.6	EIWN	Prewitt-Diaz & Gaisel, 1980
104.5	106.9	-2.4	McCarthy(S)	Shellenberger, 1982
	82.6		TVIP	Dunn et al, 1986
102.5	93.5	9	WISC	Matias Carrelo & Zaidspiner, 1990
103.6	98.1	5.5	WISC-R
89.4	84.9	4.5	EIWN-R	Rossello et al, 1988
83.7			DAM
104.7	111	-6.3	EIWN-R	Vazquez, 1989
80.6	77.7	2.9	EIWA	Ortiz Colon et al, 1993
111.8	114.7	-2.9	EIWA	Andrue et al, 1991
96.1			WAIS-R
105.4			APM
91.4			SPM	Alvarez et al, 1994
108	107.2	1.2	EIWN-R-PR	Busigo et al, 1998
	82.7		WLPB-R(S)	Paez et al, 2007
98.9			KBIT	Anderson & Souto, 2005
88.9	93	-4.1	EIWN-IV	Montes et al, 2010
94.5	95.7	-1.2	EIWN-R	Pons et al, 2008
104.5	106.2	-1.7	EIWA-III
84.6			SPM

__________________________________________

The average difference between nonverbal and verbal abilities is -1; or to put it differently, Puerto Rican islanders score 1 point higher on verbal tests.

To make a sample I can compare with U.S. norms, I exclude all of the same samples that I did in Section Ic. I also cut all the samples without clear verbal or nonverbal scores. This leaves 16 studies. The average verbal IQ for Puerto Rico is 85, and the average nonverbal IQ is 85.1. Virtually identical.

Puerto Ricans islanders do not show a relative weakness on verbal tests. The verbal deficit that mainlanders exhibit vis-a-vis their nonverbal test scores therefore appears to reflect differential adjustment to the English language. Calling this a ‘disadvantage’ requires some perspective since time spent on the mainland is associated with increases in IQ scores. If we accept the time trend, mainlanders even have higher verbal IQ scores than islanders. They’re profiting from the environment, but not as much as they would if they were fully assimilated to the language.

Section IIIa showed that mainland born Puerto Ricans have IQs that are 5.6 points higher than first generation migrants. It might be the case that the verbal deficit is strictly a first generation test pattern that is being averaged over the entire population. First generation migrants at least account for part of the difference: Han (2006) shows that verbal scores improve more for second generation children than nonverbal scores: Standardized math scores were .21σ higher for the mainland born children, but reading scores were .35σ higher. However, even the two studies with strictly mainland-born samples (Orsini, 1974; Colon-Papazoglou, 1998) both show large differences between the verbal and performance scales of the WISC.

But why would second- (much less third- or fourth-) generation Puerto Ricans have unique troubles with the English language? Recall that one of Thomas Sowell’s four possible explanations for stagnant Latino IQ scores was “enduring Spanish language usage in Hispanic homes.” At least half the world is bilingual, and the larger literature on bilingualism does not show a negative impact on IQ. Early studies of US immigrant groups like Jews and Germans did not show a negative impact from native language usage in the home. It was only the low scores of Hispanic bilinguals that even prompted a belief in the ills of dual language learning. Lloyd Dunn attempted to square the circle by arguing that high IQ is associated with successfully integrating two languages (e.g. with Europeans or East Asians), while low IQ bilingualism leads to illiteracy in two languages (Dunn, 1988, pp. 71-72).

Of course, objectively measuring ‘bilingualism’ makes a significant difference. Loyola & McBride (1991) tested mainlander children from Puerto Rican enclaves of Philadelphia. Spanish was the primary language in 96% of the homes, but 75% reported that their schools were a bilingual environment. This might make them bilingual on paper, but their IQ on the PPVT-R was 51.3, indicating almost no understanding of the English language. When they were given a Spanish translation of the same test their IQ was 80.6.

In contrast, Oplesch & Genshaft (1981) tested balanced bilingual children from northern Ohio on the English and Puerto Rican versions of the Wechsler Intelligence Scale. 80% of these children were born on the mainland, but Spanish was the primary language in 100% of their households. Nevertheless, the children were found to be equally fluent in English and Spanish through a language facility test. And their IQs were very close on the English and Spanish versions of the Wechsler: 91.2 on the WISC-R, and 86.7 on the EIWN. But even these balanced bilinguals had much higher scores on the performance scales.

Ailin Colon-Papazoglou (1998) tested mainland-born children from Brooklyn on the WISC-R. Spanish was the first language of 71.3% of this sample, but 78.7% reported English as their dominant language. English was the primary language in 72.2% of the homes, and 87% said that English was the primary language with their peers. Children were tested for English and Spanish language facility and divided into three groups: monolingual, marginally bilingual, and balanced bilingual. IQ was 93.9 for the monolingual group, 96.4 for the marginal bilinguals, and 94.9 for the balanced bilinguals. All groups had much higher scores on the performance scales.

In Section IVd, I will present evidence that the Puerto Rican IQ gap is apparent years before school entry. This would implicate the early home environment. How does first language relate to score gaps? Sathy et al (2006) present SAT Reasoning Test data from 2004-2006 for three different groups of Puerto Ricans: those who were raised only with the English language, those who were raised with both English and Spanish, and those who were raised only with Spanish. The AQ of the Spanish-speaking mainlanders was 86.4, the AQ of the bilingual mainlanders was 87.3, and the AQ of the English-speaking mainlanders was 91.6. The Spanish-speaking mainlanders did not show a difference between their verbal and math scores, but the bilingual and English-speaking groups showed a larger gap with whites on the math test (see Section IIIb).

The Add Health survey also asked about language. A large majority of mainlanders said that English was their primary household language (81%). The IQ of this group was 93.8. The 19% from Spanish-speaking homes had an IQ of 86.1.

Why do English-speaking Puerto Ricans score so much higher than Spanish-speaking or bilingual mainlanders? It could be that learning Spanish is a proxy for acculturation to general Puerto Rican ways of thinking (whatever that may be) that are less beneficial for intellectual development. Of course, all else is not equal: English-use at home is also strongly linked to having a non-Hispanic parent.

The test performance of native born mainlanders puts limits on how much of the gap between whites and Puerto Ricans can be explained by first generation disadvantages. Similarly, the test performance of English-speaking mainlanders puts limits on how much of the gap can be explained by the retention of Spanish beyond the first generation.

__________________________________________
__________________________________________

                 SECTION IV: THE DEMOGRAPHY OF PUERTO RICAN ABILITY                  

In this final section I look at intelligence test differences among Puerto Rican sub-groups.

__________________________________________

■ IVa. Race

The first U.S. census on Puerto Rico in 1899 classified 32% of the population as mixed race, but today very few identify as such. According to the 2010 U.S. census, the island is 75.8% white, 12.4% black, 7.8% some other race, 3.3% two or more races, 0.5% Amerindian, 0.2% Asian, and <0.1% Pacific Islander.

Spain colonized Puerto Rico in the early 16th century—less than two decades after Columbus initiated European expansion into the New World. The colonists took native brides and formed an enduring mestizo base population, while the Indian people were trampled into demographic fragments. Black slaves and black freemen were another early element on the island, and another source of brides for the dominant caste. But the island population was relatively small until the 19th century when there was a boom in white (primarily Spanish) immigration and large numbers of Africans were imported as slaves.

Table XI is a list of modern genetic admixture studies for Puerto Rico. These studies tell us what percentage of Puerto Rican genes come from European, Sub-Saharan African, and Amerindian ancestors, respectively. Geneticists commonly determine this by looking at sets of Ancestry-Informative Markers (AIMs); these are alleles known to be at distinctively high frequencies in each of the source populations. While this method relies on dozens or hundreds of genetic variants, there are also more advanced genome-wide measurements of ancestry based on millions of variants.

Table XI. Racial Ancestry in Puerto Rico

European %	African %	Amerindian %	N	Method	Reference
63.7	21.2	15.2	642	93 AIMs	Via, 2011
65.5	16.2	18.3	181	44 AIMs	Salari, 2005
60.2	20.2	19.6	135	44 AIMs	Choudhry, 2006
62.7	22.8	14.6	223	104 AIMs	Risch, 2009
70	19	11	133	99 AIMs	Avena, 2012
67	20.6	12.4	803	genome-wide	Galanter, 2012
72.4	14.8	12.8	55	genome-wide	Gravel, 2013
	23.6		100	genome-wide	Bryc, 2010

__________________________________________

Averaging the results from 8 studies: Puerto Ricans are about 65.9% European, 19.8% African, and 14.8% Amerindian.

How these percentages overlap with ethnic self-descriptions in Puerto Rico is a more uncertain matter. Figure 3 is a graphic representation of the racial composition of 181 individual Puerto Ricans from Salari et al (2005). This sample exhibits a racial balance close to the island averages from Table XI (with a below average representation of the African element, and an above average representation of the Indian element). Although few Puerto Ricans see themselves as mixed, we can see that only about 15% of the population has more than 80% European ancestry. Similarly (according to this sample), most of the black genes are dispersed through the population, and aren’t highly concentrated in the 12.4% of the population who self-identify as black; the 22 individuals here with the most African ancestry are still only about half African. And, of course, very few Puerto Ricans identify as native Indian—almost all of that ancestry is dispersed through the population.

Figure 3. Racial Ancestry of 181 Individual Puerto Ricans

Table XII is a list of admixture studies for Puerto Rican Americans. Averaging the results of 4 studies: mainlanders are 58.3% European, 25.9% African, and 15.8% Amerindian.

Puerto Ricans in the U.S. have over 7 percentage points more nonwhite—mostly black—ancestry than islanders. While a plurality still identify as white, mainlanders are much less likely to identify themselves as either black or white, with many more embracing a multiracial or miscellaneous label (e.g. the 2000 census: 49% white, 8.2% black, and 42.8% “other race”).

__________________________________________

Table XII. Racial Ancestry of Mainland Puerto Ricans

European %	African %	Amerindian %	N	Method	Reference
53.3	29.1	17.6	64	35 AIMs	Bonilla, 2004
57.2	27.4	15.4	1129	100 AIMs	Lai, 2009
61.7	24.1	14.3	154	104 AIMs	Risch, 2009
61	23	16	92	171 AIMs	Peralta, 2010

__________________________________________

There may be an earlier study of race and IQ in Puerto Rico than “Racial Differences in Intelligence as Measured by Pictorial Group Tests” (Vincenty, 1930), but I’m sure it isn’t as clearly labeled. This unpublished doctoral dissertation unsurprisingly comes from Harvard University, an institution that is widely known to award sham PhDs for racialist pseudoscience.

In 1928 Nestor Vincenty examined first grade schoolchildren from seven different regions across the island on the Detroit First Grade Intelligence Test, the Pinter-Cunningham Primary Mental Test, and the Dearborn Group Tests of Intelligence.

The children were visually classified as either ‘white’ or ‘colored’ (i.e. black or mulatto) by their teachers. The average score from the three tests was 84.5 for 411 whites and 79.2 for 117 coloreds. This is a difference of 5.3 points—over 1/3 of a standard deviation.

Phenotypic data was also recorded for 1,127 people from the Puerto Rican standardization of the Wechsler Adult Intelligence Scale in 1965 (Green 1972). The test examiners were trained to rate each participant on a skin color scale from 1-5, although other racial characteristics were likely factoring into the subjective rankings; not least because the explicit purpose of the color rankings was to study race.

Each darker color ranking had progressively lower IQ scores, years of education, and occupational status. The IQ difference between the lightest color group and the darkest color group was 8 points (over ½ of a standard deviation). Green notes that people in categories 1-3 would more often be considered white in Puerto Rico, with category 4 being seen as mixed, and category 5 as negro. In Vincenty’s study, 78% of the children were classified as white, and 22% as colored. For the purpose of comparison we can merge skin tones 1-3 and 4-5, which is the closest approximation to Vincenty’s percentages that we can get with the categories in this study (86% white/14% colored). The IQ of 975 whites was 84.3 and the IQ of 152 coloreds was 80.2. This is a difference of 4.1 points, which accords with Vincenty’s results from nearly 40 years earlier.

Three different sources provide racial data for the United States mainland:

1. Puerto Rican respondents to the General Social Survey classified themselves as either ‘white’, ‘black’, or ‘other’ just as they did in the U.S. census, and the percentages choosing each were close to the census responses. The Wordsum IQ for 108 whites was 88; the IQ of 22 blacks was 85. The 79 mainlanders who self-identified as ‘other’ had an IQ of 88, just like the whites.
2. Puerto Ricans in the National Longitudinal Study of Adolescent Health also identified as ‘black’, ‘white’, or ‘other’, and their choices were also close to the census. The 42 whites had a PPVT IQ of 94.5 and the 13 blacks had an IQ of 93.4. 40 that identified as ‘other’ had an IQ of 91.7.
3. Section IIIb discussed the National Assessment of Educational Progress (NAEP); ethnic differences on these academic learning tests are largely isomorphic with ability test gaps. If the Achievement Quotient (AQ) for whites is 100 on the 2009 NAEP math and reading tests, then African Americans have an AQ of 86.8 (.88σ), and Puerto Ricans Americans have an AQ of 89.7 (.69σ). Puerto Rican scores can also be sub-divided by racial identity. The AQ of black Puerto Ricans is 85.4 (1.4 points below African-Americans). The AQ of white Puerto Ricans is 92.7. This gives us a black-white AQ gap of 7.3 points (.49σ) for mainlanders, similar to the 8 point IQ gap in Puerto Rico reported by Green (1972). The NAEP also allowed people to identify as ‘two or more races’, but this was a surprisingly unpopular identity label among the mainlanders; not even providing large enough sample sizes for the online Data Explorer to report the numbers on many of the tests. The tests that were reported gave scores that were very close to white Puerto Rican performance.

Five studies looking at race and test scores among Puerto Ricans are shown in Table XIII. Standardized mean differences are calculated for four comparison groups. ‘White-Black’ is the difference between those classified or self-identified as ‘black’ and ‘white’. ‘White-Nonwhite’ is the difference between those classified or self-identified as ‘white’ and everyone else (i.e. blacks and mulattos together, or blacks and “others” together). ‘Mulattos’ are those classified as something racially intermediate between ‘whites’ and ‘blacks’; in Green (1972) this is the intermediate color categories, and in Malloy (2014b) and Malloy (2014c) it is those who identify as neither white or black.

The average difference between white and black Puerto Ricans is 5.3 IQ points (.35σ). The average difference between whites and non-whites is 3.3 IQ points (.22σ). Those perceived as mixed race likewise perform almost exactly half way between those seen as black and those seen as white.

Table XIII. Standardized Test Score Differences between Differently Classified Racial Groups

Reference	Location	Test	White-Black	White-Nonwhite	White-Mulatto	Black-Mulatto
Vincenty, 1930	Island	Various		.35
Green, 1972	Island	WAIS	.53	.27	.17	.36
Malloy, 2014b	Mainland	PPVT	.07	.16	.20	-.09
Malloy, 2014c	Mainland	Wordsum	.20	.05	0	.20
Malloy, 2014e	Mainland	NAEP	.49
			(.35)	(.22)	(.17)	(.20)

How “should” Puerto Ricans perform given strict hereditarian assumptions about race and intelligence? Hard hereditarianism runs into obvious problems; for example, African-Americans score much higher than sub-Saharan Africans than we would predict given their relatively low amount of white admixture. On the other hand, racial reductionism seems to offer some surprisingly accurate predictions. For instance, in IQ & Global Inequality, Richard Lynn compared his previously estimated IQs for 22 Latin American and Caribbean nations with alternative IQ estimates based solely on the racial compositions of nations that were reported in various reference works. (For example, if European IQ is 100 and African IQ is 70, a nation that is 60% black and 40% white would have a predicted IQ of 82.) The correlation between the standard test-based IQ estimates and the racially predicted IQs was 0.81 (Lynn & Vanhanen, 2006, pp. 240-241).

This is perhaps more surprising because those almanac racial compositions (e.g. from the CIA World Factbook) aren’t very precise. The racial composition of Puerto Rico is given as “76% European, 24% African & Mulatto”, which is a shallow description of the island’s social, much less genetic, racial composition. Granted, Lynn’s estimated IQ and predicted IQ for Puerto Rico is one of his widest discrepancies (84 vs. 94). Perhaps the discrepancy would go away with more accurate data. We can improve Lynn’s race-based estimate by A) using the direct ancestry measurements from Table XI; B) using Rindermann’s (2013) updated African IQ estimate of 75; C) using Lynn’s (2012, p. 60) latest estimate of 96.6 for IQ in Spain; and D) using Lynn’s (2006, p. 158) estimate of 86 for Latin American Indians.

This gives us an expected IQ of 90.8 for Puerto Rico, which is still over 1/3 of a SD higher than the observed IQ. It can’t be ignored: Puerto Ricans are conspicuously underperforming relative to their European ancestry.

It is perhaps noteworthy that the race-based IQ estimate for Puerto Rican Americans (using ancestry data from Table XII) is 89.3, which is right in range. Although this doesn’t factor in the environmental effects of being raised in the United States (e.g. African Americans do not have an IQ of 75). Fortunately, this can be done, and with the added advantages of a common dataset: the General Social Survey provides IQ data for native-born African Americans, Spanish Americans, Native Americans, and Puerto Rican Americans. African- and Native-Americans are ~17% and ~40% white. Factoring out the theoretical score advantages from white admixture gives us an IQ of 87.8 for blacks, 96.1 for Spaniards, and 87.5 for Natives. The predicted GSS IQ from weighting these scores by the genetic ancestry of mainlanders is 92.6. The observed GSS IQ for mainlanders is 91.4!

__________________________________________

■ IVb. Region

Two of the larger studies have reported scores for different regions in Puerto Rico (Vincenty, 1930; Kahn et al, 1977). The territory is divided into 78 municipalities; these two studies have sampled students from 8 of the most populous regions. The scores are averaged and adjusted around the estimated IQ for Puerto Rico:

Arecibo— 88.4
Ponce— 87.4
Mayaguez— 86.8
Cabo Rojo— 86.2
San Juan— 85.8

(Puerto Rico— 84.6)

Caguas— 80.9
Humacao— 80.9
Vega Baja— 80.2

What do these regional IQs predict? To the extent there is an economic correlation there seems to be an association between IQ and having a larger budget deficit; with high public debt per capita in Arecibo and Ponce, and budget surpluses in Caguas and Humacao. There also seems to be some association with census racial identity; with higher scores in whiter regions (e.g. Arecibo: 84.5% white and 6.1% black) and lower scores in less white regions (e.g. Humacao: 66.1% white and 18.5% black).

This post is not for summarizing the association between race and economic performance in Puerto Rico—the focus here is intelligence research. However, recent genomic studies confirm that European ancestry is associated with socioeconomic status in Puerto Rico, as it is across Latin America. For a discussion of the links between geography, racial ancestry, and socioeconomic status in Puerto Rico, read Via et al (2011).

Rural/urban differences in Puerto Rico were also examined in two of the larger studies. In their 1965 standardization of the WAIS, Green & Martinez (1967) found an urban IQ of 88.1 and a rural IQ of 77.5, a difference of 10.6 points (.71σ).

In the 1977 standardization of the Raven Progressive Matrices, Kahn et al (1977) found an urban IQ of 86.8 and a rural IQ of 83.8—only 3 points (.20σ).

Is 12 years an implausible amount of time for a demographic gap to shrink by ½ of a standard deviation? The average age of the WAIS norm group was over two decades older than the RPM norm group, so we’re really looking at a 3 1/2 decade difference. A trend over the middle of the 20th century was a relatively rapid reduction in the rural/urban IQ gap. In the United States this meant a test gap of about 6.5 points circa WWII that dwindled to about 2 points by the 1970s (Loehlin, 2000, p. 181).

__________________________________________

■ IVc. Sex

In the 1970s Arthur Jensen investigated a possible sex difference in African-American IQ—a finding that might lend support to a non-genetic etiology for the black-white gap. Jensen pointed to a literature showing that males are more biologically sensitive to the effects of environment than women: their healthy development is more disrupted by shortages and stressors, and they also benefit more from advantages (Jensen, 1971). This is, of course, related to the wider-variability-in-male-phenotypes hatefact that set off the faculty revolt that got Larry Summers booted from the Harvard presidency in 2006.

Thomas Sowell, citing Jensen, argued that this signature of environmental insult was consistent with what we see among ethnic groups with relatively poor test performance: higher IQ scores were observed among female European immigrants in the early 20th century, higher IQ scores are observed among African-American females, and higher IQ scores are observed among Mexican- and Puerto Rican-American females (Table XIV, taken from Sowell 1979, p. 223).

Table XIV.

As late as 2002 Sowell has stated that black females have higher IQ scores than black males, but Arthur Jensen had already pored through massive amounts of data to reevaluate this possibility in The g Factor (1998, pp. 498-500, 527-528) and decided that the largest and best datasets showed no such sex difference, and that the “hypothesis must now be discarded” (p. 500).

As noted in Section IIc, Sowell’s published Urban Institute data isn’t very reliable. But that isn’t to say that Sowell is wrong about a Puerto Rican sex difference. We can evaluate this possibility with our own dataset: Table XV is a list of 17 studies that report IQ scores by sex. A negative difference indicates higher female scores.

Table XV. Puerto Ricans: Male & Female IQ Scores

Male IQ	Female IQ	Difference	Test	Male N	Female N	Location	Reference
85	83.7	1.3	Various			Island	Vincenty, 1930
78	72.3	5.7	Unique	120	123	Mainland	Smith, 1942
66	66	0	CCF	54	54	Mainland	Anastasi, 1953a
88.9	102.5	-13.6	DAM	25	25	Mainland	Anastasi, 1953b
83.7	85.8	-2.1	Otis	1469	1409	Mainland	Moriber, 1961
97.5	94.4	3.1	EIWA	53	71	Island	Green, 1967
99.7	100.5	-.08	Vane	105	120	Mainland	Walsh, 1971
85.2	85.1	.01	WISC	25	25	Mainland	Orsini, 1974
94.1	94.9	-.08	CIIS+SBIS	174	165	Mainland	Morris, 1977
84.2	86.4	-2.2	EIWN	101	109	Island	Fuertes, 1976
80.8	80.4	.04	CPM	1181	1179	Island	Kahn, 1977
82.6	83.3	-.07	SPM	1435	1448	Island	Kahn, 1977
110.1	111.7	-1.6	EIWN-R			Island	Vasquez, 1989
102.9	101.6	1.3	K-ABC(S)			Island	Vasquez, 1989
122.1	120.1	2	EIWA	15	15	Island	Andrue, 1991
102.8	94.9	7.9	WAIS-R	15	15	Island	Andrue, 1991
90	88.5	1.5	AFQT	137	131	Mainland	Malloy, 2014a
97.3	96.8	.05	PPVT	43	54	Mainland	Malloy, 2014b
86	88	-2	Wordsum	74	135	Mainland	Malloy, 2014c
88.1	90.4	-2.3	PPVT	57	52	Mainland	Malloy, 2014d

__________________________________________

Sowell’s Urban Institute data show a smallish gap of 1.7 points. The largest mainland study (Moriber, 1961) showed a similar .14 SD gap favoring females. The largest island samples were the two Raven standardizations, which showed a 0.4 point advantage for males on the Colored Matrices and a mostly counterbalancing 0.7 advantage for females on the Standard Matrices.

Looking at all the studies, there is a small median advantage for males in Puerto Rico (0.4 IQ points), and a small median disadvantage for males in the U.S. (-0.8 IQ points).

Nativity data from Moriber (1969) suggests a similar pattern. The female advantage among mainland born Puerto Ricans was twice the female advantage of island born Puerto Ricans (native .17σ/migrant 0.9σ). But the natives also had much higher IQ scores than the migrants: this pattern seemingly undermines the logic of the female advantage. Sowell had already pointed to the lower nonverbal IQ of Puerto Rican migrants as evidence against hereditarianism (Section IIIa). If the island environment—whatever that may entail—is inferior to the American environment, then shouldn’t the gap in favor of females be larger on the island instead of the opposite?

In his discussion, Sowell points to the masculine bias of Latin culture, so perhaps he might argue that the cultural disadvantages that specifically affect women on the island are large enough to offset the sex difference that would otherwise result from general disadvantages experienced equally by males and females.

Or the mainlander sex difference might fade away, just as the African American one did, given larger datasets; for example the SAT shows a Puerto Rican American male advantage of about 1/5 of a standard deviation.

Assuming there is a female advantage in the US, it might be due to the language deficits explored in Section IIIc. There is some support for this in the Add Health survey: Spanish-speaking males and females have an IQ gap of 6 points, but there is no gap between the English-speaking mainlanders. The sample sizes are low for the Spanish-speakers, so these results don’t inspire a great deal of confidence. More research is needed.

__________________________________________

■ IVd. Age

In May 2013 I summarized the conflicting opinions among social scientists about the onset and development of the black-white IQ gap. Some scientists argue that the gap is much smaller at the beginning of kindergarten and then progressively widens during the school years. Other scientists argue that the full gap is found in the early years of post-infancy, and remains stable after children enter school. I offered a large collection of evidence in support of the latter theory, showing that the black-white IQ gap is already 1 standard deviation by 3 years of age.

Researchers have not yet discussed or debated these important issues in the context of Hispanic Americans, who are now a larger minority population than blacks. The present dataset at least allows us to resolve unasked questions about the onset and development of the PRA (Puerto Rican American)-white gap.

Supplement 2 is a list of 50 samples providing IQ data by age. I have excluded disadvantaged samples like psychiatric patients. I also excluded the Hawaiian studies.

The Admin column is the year the test was administered. The Cohort column is the year the sample was born. The IQ column is the reported IQ score, while the FC IQ column is the IQ score adjusted for the Flynn Effect. The Method column describes the three different ways I picked a single age for each sample. ‘Reported’ indicates that either a single age was reported, or that a reported age range for the participants was not wider than 12 months (e.g. if subjects were all 58-63 months, this would be included as 5 years-old). ‘Estimated’ indicates that a school grade was reported instead of an age, so I assigned a typical age for that grade (i.e. 6 for kindergartners … 18 for high school seniors). ‘Averaged’ indicates that the study offered an average age for a sample with wider variability than 12 months. The typical age range was about 5 years. Several studies had grade ranges, and estimated ages were applied for each grade and then averaged together.

             Supplement 2. Age & Puerto Rican American IQ             

The Puerto Rican American IQ gap is fully formed before children enter school—as early as age 3. There is no strong indication here that the gap changes much between ages 3-18, or that it systematically improves or grows wider through the school years (Figure 4). The median IQ of the 50 samples is 86.9. The average for 8 samples of preschool children (ages 3-5) is 87.5. The average for 20 samples of primary school children (ages 6-11) is 85.1. The average for 18 samples of secondary school children (ages 12-18) is 87.5.

There are also four samples for adults. Three are individual samples, ages 19-21, from the NLSY. The fourth sample is all the adults tested for the General Social Survey. The average IQ for these four samples is 90.

Figure 4. Puerto Rican American IQ, Ages 3-18 years

How does IQ develop across the lifespan in Puerto Rico? Supplement 3 is a list of 49 age-stratified samples for the island territory.

                     Supplement 3. Age & Puerto Rican IQ                      

The term “cumulative deficit” was coined by psychologist Otto Klineberg, who believed that racial differences in intelligence should increase with age due to the compounding effect of bad environments on the disadvantaged group: “… it is as the children get older that differences in test performance appear. Surely this is to be expected on the basis of the cumulative effect of an inferior environment” (Klineberg, 1963, p. 200).

This remains an influential idea among social scientists, but African- and Puerto Rican-American children apparently show no progressive decrement in IQ scores with age. The IQ gaps are fully formed several years before school entry, and are stable throughout childhood and adolescence. This is not the case in Puerto Rico, where IQ appears to decrease during the school years (Figure 5). This pattern is in relation to U.S. norms, so if schooling genuinely has an effect on intelligence, then this might reflect the superiority of mainland schools (cf. the 2 ½ standard deviation shortfall on the NAEP mathematics tests: Section IIIb). In other words, it’s not so much that there is a disproportionate accumulation of harmful exposures across childhood in Puerto Rico, as there is a disproportionate accumulation of beneficial exposures across childhood in the United States.

The median IQ of the 49 samples is 84.6. The average for 6 samples of preschool children (ages 3-5) is 90.1. The average for 27 samples of primary school children (ages 6-11) is 84.6. The average for 14 samples of secondary school children (ages 12-17) is 83.6. The average of the two adult samples is 82.8.

Figure 5. Puerto Rican IQ, Ages 3-18 years

__________________________________________

■ IVe. Birth Cohort

Finally, how has Puerto Rican IQ changed across generations? The Flynn Effect is the secular inflation of IQ scores; in the United States average IQ scores have been rising by some 3 points a decade for about 100 years. The causes and nature of the Flynn Effect are still a mystery, and some people are quick to lump any changes in sub-group differences as part of an overarching phenomenon: e.g. if the black-white ability gap has narrowed, the Flynn Effect has simply operated more rapidly for blacks. This is probably a bad assumption (e.g. Wicherts et al, 2004), but to the extent that certain sub-group differences have changed over the period of score inflation (e.g. the rapid disappearance of the rural/urban gap), it will be difficult to theoretically disentangle them from the larger phenomenon without a better understanding of the rising scores. For the most part, sub-group differences have remained stable, suggesting that the Flynn Effect has comparably impacted most segments of society. However, there is no inherent reason that some groups can’t fall behind. Between 1924 and 1938, raw scores increased for Hawaiian Puerto Ricans on a battery of mental ability tests (Smith, 1942), but white scores increased even more over the same time period, so the Puerto Rican IQ dropped from 79 to 76 on majority norms.

To measure the absolute increase in Puerto Rican performance over time we would need 100 years of raw scores for the same test. We don’t have this. But we do have the 1965 Puerto Rican standardization of the WAIS, which gives us nearly a half-century of comparative data (Green, 1969). Figure 6 shows the WAIS performance (and average years education) of 11 different age groups from 15-64 years old. The oldest group was born between 1901-1905, and the youngest from 1940-1945.

Figure 6. WAIS Performance, Ages 15-64 years (Puerto Rico)

Studies that compare very young people and very old people typically find differences of 1-2 standard deviations. But plummeting IQ scores are not observed when we follow and re-test the same group of people as they age over time. What this shows is that aging is not associated with large losses in cognitive function; instead, older people have low scores on IQ tests because they always had low scores on IQ tests. Or rather what used to be average now looks low in comparison. This is how the Flynn Effect operates: scores aren’t increasing for everyone over time, performance is just always a little bit higher for those born later in time.

An exception to the stability of intelligence is the so-called “terminal decline”—IQ scores do not decline gradually over the later lifespan, but they do decline precipitously in the final years before death.

So Figure 6 is like a fossil record of IQ scores in Puerto Rico stretching back to the turn of the last century. There is a steep drop-off in scores for the oldest group that is consistent with a terminal decline. But otherwise we can see a cohort difference of about 12 points develop in the 40 years between 1903 and 1943. This is an average of 0.3 points per year, which is the standard rate of score inflation on Wechsler tests.

IQ scores are apparently increasing in Puerto Rico at about the same rate over time as they are on the mainland United States.

Another way to test this is to look at the IQ scores of different cohorts across time. If IQ is increasing over time in relation to mainland scores, then this would be consistent with a faster Flynn Effect. If IQ scores are declining over time, then perhaps the Flynn Effect is occurring at a slower rate in Puerto Rico.

Using the data from Supplement 3, there is an apparent decline in scores over time if we include the test scores from Monroe (1921) and Columbia University (1926). We can either assume these scores are not representative or that there was an abrupt decline in scores for those born after 1920. Excluding the 8 samples born before 1920, we are left with 41 samples that suggest a basic stability in IQ performance over the next 80 years (Figure 7). This supports the data from the WAIS suggesting that IQ scores are increasing at about the same rate in Puerto Rico as they are on the mainland United States.

Figure 7. Puerto Rican IQ & Birth Cohort

The dataset doesn’t contain anything similar to the WAIS standardization for mainland Puerto Ricans. The General Social Survey has Wordsum scores for Puerto Ricans born between 1917 and 1992, but the Wordsum doesn’t exhibit much of a Flynn Effect (There is a slightly negative correlation between birth year and Wordsum for whites). However, birth cohort trends in IQ scores are available in Supplement 2. Once again, a relative rise in scores could suggest a faster Flynn Effect (although independent causes would be just as possible), while a decline in scores might suggest a perpetually delayed exposure to whatever benign forces are causing the score inflations.

Figure 8 shows IQ data by birth year for 50 samples of mainland Puerto Ricans. The scatterplot suggests stability in mainland scores over 80 years with no obvious trends up or down on majority norms.

Figure 8. Puerto Rican American IQ & Birth Cohort

You might remember that I found higher post-1970s scores for mainlanders in Section IIe, in Section IIf, and in Section IIIc. This upward trend is observed when these same 50 data points are arranged by the study year instead of by birth date (Figure 9). The 50 samples have a median IQ of 86.9. The 24 samples tested between 1935 and 1976 have an average IQ of 85.1. The 26 samples tested between 1979 and 2004 have an IQ of 88.9.

Figure 9. Puerto Rican American IQ & Test Year

The discrepancy suggests that the sudden late 1970s jump in IQ scores could be a spurious pattern. On the other hand, SAT scores also showed something of a narrowing from the mid-1970s through the 1980s—albeit by 1 point (Section IIIb). So there is conflicting evidence for a small decrease in the Puerto Rican gap sometime in the 1970s or 1980s.

__________________________________________
__________________________________________

                 SECTION V: CONCLUSIONS                  

Now what have we learned?

IQ in Puerto Rico

Table I features 26 studies for Puerto Rico. The single best study for estimating intelligence in Puerto Rico is the large 1977 standardization of the Raven Progressive Matrices—a nonverbal test of reasoning ability that is well designed for cross-cultural comparisons. IQ on this test was 82.8. Three more well-known IQ tests were normed on large samples, but this involved adapting the test items to another language and culture, which requires additional assumptions about cross-cultural comparability. IQ on all four tests was 82.7. All together, 19 studies provide scores that might appropriately be compared with U.S. norms: from these samples I estimated an IQ of 84.6 for Puerto Rico (Section Ic).

IQ of Puerto Rican Americans:

Table II features 45 studies for Puerto Ricans living in the parts of the United States that aren’t Puerto Rico (Yes, it has been awkward trying to communicate this technicality throughout the post). Four normal samples from Hawaii in the 20s and 30s had an average IQ of 74.2. An additional 30 studies provide adequate scores for the rest of the mainland United States. My estimate for Puerto Rican Americans from these samples was 84.7, but other methods suggest that this is too low:

• The 14 studies administered since 1980 exhibit a higher average: 87.4 (Section IIe).
• The 13 studies that compare white and mainlander samples directly suggest a relative IQ of 89.4 (Section IIf).
• SAT Reasoning Test data was compiled for the 37 years 1976-2013. Some 500,000 Puerto Ricans had a standard score of 88.2 (Section IIIb).
• Scores from the 30 IQ studies can be sub-divided by age and analyzed by birth cohort, creating 50 different samples with a time-invariant average of 86.9 (Section IVe).

Taking an average of these five methods gives us a final IQ estimate of 87.3 for Puerto Rican Americans. This is close to Lloyd Dunn’s joint estimate of 88-90 for Puerto Ricans and Mexicans (Dunn, 1988).

Mainlander Characteristics:

Puerto Rican Americans develop a modest IQ advantage over their island cousins. Three additional test differences were observed between the island and the mainland:

• Ability Patterns: Mainlanders score 1/3 of standard deviation higher on nonverbal tests than they do on verbal tests. Islanders score equally well on verbal and nonverbal tests. This suggests language difficulties are hampering mainlander IQ performance (Section IIIc).
• Sex differences: There is a small sex difference in favor of females on the mainland. This IQ difference is not apparent on the island. In theory this suggests something of a negative environmental effect on mainland scores (perhaps the language handicap just described). The lack of a female superiority on the island (which has lower scores and more disadvantages than the mainland) might be explained by cultural exposures that uniquely disfavor females (Section IVc).
• Age differences: The entire score difference between whites and mainlanders is observed years before school entry, and this gap is stable throughout childhood and adolescence. The smallest differences are observed in adulthood, suggesting that the mainland environment does not cumulatively disfavor Puerto Ricans. In contrast, the difference between mainland whites and islanders is smallest before school entry. The comparative gap widens through childhood and adolescence, and is largest for adults. This pattern is consistent with the cumulative effects of differentially advantageous environments: the United States environment is apparently more beneficial for cognitive development (Section IVd).

The 2.7 IQ point difference between islanders and mainlanders is an underestimate of the beneficial effects of the mainland environment since migrants have lower initial test performance than islanders (Rodriguez et al, 1985; Section IIIa). Mainlanders also have more nonwhite ancestry, which is associated with lower IQ scores (Section IVa).

Speaking of which …

Heredity & Environment

Racial appearance and identity are highly associated with test performance on the island and on the mainland. The relationship between racial identity and IQ is not as strong among Puerto Ricans as it is among Americans, but the relationship between racial identity and racial ancestry is also significantly weaker among Puerto Ricans—an admixed population.

Lloyd Dunn estimated that about half of the gap between whites and Puerto Rican Americans could be explained by nonwhite racial ancestry. This is a pro forma hereditarian estimate for black and white differences (e.g. Rushton & Jensen, 2005). But I feel like this a better estimate for the difference between white Americans and ethnic groups in their source nations than whites and ethnic groups in the United States, where the population shares a high quality first world environment. I think its quite possible that ethnic gaps in the US are close to entirely genetic. Section IVa suggests that the average mainlander IQ is what you would predict just by knowing their racial ancestry.

Standard environmental theories don’t have much to offer in explaining ethnic gaps. For example, gaps are seen long before children enter school (Section IVd), but behavior genetics suggests that the shared family environment—which includes socioeconomic status—contributes almost nothing to long-term intellectual development. ‘Culture’ is a popular yet amorphous idea, that isn’t really quantified or empirically demonstrated. Language probably does contribute to the Puerto Rican American gap, but it is apparent that it can’t explain much. No matter what their language, Puerto Rican scores fall in a limited range: Spanish-speaking Puerto Ricans score comparatively low on translated tests, on tests that are adapted for use in multiple languages, and on tests that don’t use language at all. Bilingual Puerto Ricans score comparatively low when given English or Spanish language tests, and also on nonverbal tests. And English-speaking Puerto Ricans perform comparatively low on English language tests and on nonverbal tests (Section IIIc). The highest average scores were the post-1970s nonverbal IQ of mainlanders (90.4), and the SAT scores of monolingual English-speaking Puerto Ricans (91.6). This seems like an approximate limit to how much of the gap language differences can explain.

On the other hand, Puerto Rican islanders score over 1/3 of a standard deviation lower than would be predicted by their nonwhite ancestry. If we assume that moving to the mainland and fully assimilating the culture and the language could raise their IQ to, say, 92, then that would be exactly half of the gap between whites and Puerto Ricans explained by environmental differences.

__________________________________________
__________________________________________

__________________________________________
__________________________________________

Alvarez, V., Santos, J., & Lebrón, F. (1994). Efectos del programa de enriquesimiento instrumental de Feuerstein sobre las habilidades cognitivas de una muestra de estudiantes puertoriqueños. Revista Latinoamericana de Psicología, 26, 51-68.

Anastasi, A., & Cordova, F.A. (1953). Some effects of bilingualism upon the intelligence test performance of Puerto Rican children in New York City. Journal of Educational Psychology, 44, 1-19.

Anastasi, A., & deJesus, C. (1953). Language development and nonverbal IQ of Puerto Rican children in New York City. Journal of Abnormal & Social Psychology, 48, 357-366.

Anderson, R.T., & Souto, S.M. (2005). The use of articles by monolingual Puerto Rican Spanish-speaking children with specific language impairment. Applied Psycholinguistics, 26, 621-627.

Andreu, B.I., Alvarez, V., & Veray, B. (1991). Estudio comparativo de la Escala de Inteligencia Wechsler para Adultos y la Escala de Inteligencia Wechsler para Adultos Revisada en una poblacion universitaria puertorriquena. Revista Puertorriquena de Psicologia, 7, 1-11.

Armstrong, C.P., Achilles, E.M., & Sacks, M.J. (1935). A report of the Special Committee on Immigration and Naturalization of the Chamber of Commerce of the State of New York submitting a study on Reactions of Puerto Rican children in New York City to psychological tests. Special Committee on Immigration and Naturalization: NY, USA.

Avena, S., Via, M., Ziv, E., Pérez-Stable, E.J., Gignoux, C.R., Dejean, C., … & Fejerman, L. (2012). Heterogeneity in genetic admixture across different regions of Argentina. PloS one, 7, e34695.

Bonilla, C., Shriver, M.D., Parra, E.J., Jones, A., & Fernández, J. R. (2004). Ancestral proportions and their association with skin pigmentation and bone mineral density in Puerto Rican women from New York city. Human Genetics, 115, 57-68.

Boone, S.L., & Montare, A. (1976). Test of the language-aggression hypothesis. Psychological Reports, 39, 851-857.

Brown, F. (1960). Intelligence test patterns of Puerto Rican psychiatric patients. Journal of Social Psychology, 52, 225-230.

Bryc, K., Velez, C., Karafet, T., Moreno-Estrada, A., Reynolds, A., Auton, A., … & Ostrer, H. (2010). Genome-wide patterns of population structure and admixture among Hispanic/Latino populations. Proceedings of the National Academy of Sciences, 107(Supplement 2), 8954-8961.

Burton, N.W., Lewis, C., & Robertson, N. (1988). Sex differences in SAT scores. College Board Report No. 88-9. New York, NY: College Entrance Examination Board.

Busigo, S., Martino, L.C., Mercado, C.M., Perez, W.J., & Colon, L.J. (1998). Validacion de la prueba neuropsicologica Stroop con la Escala de Inteligencia Wechsler para ninos–Revisada de Puerto Rico. Revista Puertorriquena de Psicologia, 11, 149-156.

Matias Carrelo, L.E., & Zaidspiner, V. (1990). Comparación de los cocientes intelectuales obtenidos en el WISC y en el WISC-R por dos grupos de niños puertorriqueños. Revista Puertorriqueña de Psicología, 63-72.

Cebollero, P.A. (1936). Reactions of Puerto Rican children in New York City to psychological tests. Puerto Rico School Review: San Juan, Puerto Rico.

Choudhry, S., Burchard, E.G., Borrell, L.N., Tang, H., Gomez, I., Naqvi, M., … & Risch, N. J. (2006). Ancestry–environment interactions and asthma risk among Puerto Ricans. American Journal of Respiratory & Critical Care Medicine, 174, 1088-1093.

Coleman, J.S., Campbell, E.Q., Hobson, C.J., McPartland, J., Mood, A.M., Weinfeld, F.D., & York, R. (1966). Equality of educational opportunity, supplemental appendix, section 9.10, correlation tables. Washington, DC: US Office of Education.

Colón-Papazoglou, A. (1998). The relationship of bilingualism, Language of I.Q. test administration and maternal acculturation on the intellectual performance of bilingual Puerto Rican-American children. Unpublished doctoral dissertation, New York University, USA.

Columbia University. Teachers College. International Institute. (1926). A Survey of the Public Educational System of Porto Rico. NY, USA: Teachers College, Columbia University.

Darcy, N.T. (1952). The performance of bilingual Puerto Rican children on verbal and non-language tests of intelligence. Journal of Educational Research, 45, 499-506.

Dunn, L.M. (1988). Bilingual Hispanic Children on the US Mainland: A Review of Research on Their Cognitive, Linguistic, and Scholastic Development. Circle Pines, USA: American Guidance Service.

Dunn, L.M., Padilla, E.R., Lugo, D.E., & Dunn, L.M. (1986). Examiner’s manual for the Test de Vocabulario en Imagenes Peabody. Adaptacion Latinoamericana. Circle Pines, USA: American Guidance Service.

Duran, R.P., Enright, M.K., & Rock, D.A. (1985). Language Factors and Hispanic Freshmen’s Student Profile. College Board Report No. 85-3. New York, NY: College Entrance Examination Board.

Frey, M.C., & Detterman, D. (2004). Scholastic Assessment or g? The Relationship Between the Scholastic Assessment Test and General Cognitive Ability. Psychological Science, 15, 373-378.

Fuertes de la haba, A., Santiago, G., & Bangdiwala, I.S. (1976). Measured intelligence in offspring of oral and nonoral contraceptive users. American Journal of Obstetrics & Gynecology, 125, 980-983.

Galanter, J.M., Fernandez-Lopez, J.C., Gignoux, C.R., Barnholtz-Sloan, J., Fernandez-Rozadilla, C., Via, M., … & LACE Consortium. (2012). Development of a panel of genome-wide ancestry informative markers to study admixture throughout the Americas. PLoS genetics, 8, e1002554.

Glutting, J.J. (1986). Potthoff bias analyses of K-ABC MPC and Nonverbal Scale IQs among Anglo, Black, and Puerto Rican kindergarten children. Professional School Psychology, 1, 225-234.

Gravel, S., Zakharia, F., Moreno-Estrada, A., Byrnes, J.K., Muzzio, M., Rodriguez-Flores, J.L., … & Bustamante, C.D. (2013). Reconstructing Native American Migrations from Whole-Genome and Whole-Exome Data. PLoS genetics, 9, e1004023.

Green, R.F. (1969). Age-intelligence relationship between ages sixteen and sixty-four: A rising trend. Developmental Psychology, 1, 618-627.

Green, R.F. (1972). On the correlation between IQ and amount of “white” blood. Proceedings of the Annual Convention of the American Psychological Association, 7, 285-286.

Green, R.F., & Martínez, J.N. (1967). Standardization of a Spanish language adult intelligence scale. Final Report. Washington, DC, USA: US Department of Health, Education & Welfare.

Hakuta, K. (1987). Degree of bilingualism and cognitive ability in mainland Puerto Rican children. Child Development, 58, 1372-1388.

Hakuta, K. & Diaz, R.M. (1985). The relationship between degree of bilingualism and cognitive ability: A critical discussion and some new longitudinal data. In K.E. Nelson (Ed.), Children’s language, Vol. 5 (pp. 319–344). Hillsdale, USA: Lawrence Erlbaum Associates.

Han, W. (2006). Academic achievements of children in immigrant families. Educational Research & Review, 1, 286-318.

Hertzig, M.E., & Birch, H.G. (1971). Longitudinal course of measured intelligence in preschool children of different social and ethnic backgrounds. American Journal of Orthopsychiatry, 41, 416-426.

Hertzig, M.E., Birch, H.G., Thomas, A., & Mendez, O.A. (1968). Class and ethnic differences in the responsiveness of preschool children to cognitive demands. Monographs of the Society for Research in Child Development, 33, 1-69.

Jencks, C., & Phillips, M. (1998). The black-white test score gap. Washington, DC, USA: Brookings Institution Press.

Jensen, A. R. (1971). The race X sex X ability interaction. In R. Cancro (Ed.), Intelligence: Genetic and environmental influences, (pp. 107-161). New York: Grune & Stratton.

Jensen, A.R. (1998). The g factor: The science of mental ability. Westport, USA: Praeger.

Klineberg, O. (1963). Negro-white differences in intelligence test performance: A
new look at an old problem. American Psychologist, 18, 198-203.

Kobrin, J.L., Sathy, V., & Shaw, E.J. (2007). A historical view of subgroup performance differences on the SAT reasoning test. Research Report No. 2006-5. New York, NY: College Entrance Examination Board.

Lai, C.Q., Tucker, K.L., Choudhry, S., Parnell, L.D., Mattei, J., García-Bailo, B., … & Ordovás, J.M. (2009). Population admixture associated with disease prevalence in the Boston Puerto Rican health study. Human Genetics, 125, 199-209.

Lesser, G.S., Fifer, G., & Clark, D.H. (1965). Mental abilities of children from different social-class and cultural groups. Monographs of the Society for Research in Child Development, 30, 1-115.

Leventhal, T., Xue, Y., & Brooks-Gunn, J. (2006). Immigrant differences in school-age children’s verbal trajectories: a look at four racial/ethnic groups. Child Development, 77, 1359-1374.

Loehlin, J.C. (2000). Group differences in intelligence. In R.J. Sternberg (Ed.), Handbook of Intelligence (pp. 176-193). Cambridge, UK: Cambridge University Press.

López, S., & Romero, A. (1988). Assessing the intellectual functioning of Spanish-speaking adults: Comparison of the EIWA and the WAIS. Professional Psychology: Research and Practice, 19, 263-270.

Loyola, J.L., & McBride, D.F. (1991). Puerto Rican children’s codeswitching and their performance on receptive vocabulary instruments and a nonverbal cognitive measure. Diagnostique, 16, 239-253.

Lynn, R. (2006). Race differences in intelligence: An evolutionary analysis. Augusta, USA: Washington Summit Publishers.

Lynn, R., & Vanhanen, T. (2006). IQ & global inequality. Augusta, USA: Washington Summit Publishers.

Lynn, R., & Vanhanen, T. (2012). Intelligence: A Unifying Construct for the Social Sciences. London, UK: Ulster Institute for Social Research.

Maier, M.H. (1972). Effects of English language training on aptitude test performance of insular Puerto Ricans. Arlington, USA: US Army, Behavior and Systems Research Lab.

Manuel, H.T. (1966). Development of Inter-American test materials. Unpublished report, University of Texas, USA.

Marmorale, A.M. & Brown, F. (1975). Comparison of Bender-Gestalt and WISC correlations for Puerto Rican, white and Negro children. Journal of Clinical Psychology, 31, 465-468.

Marmorale, A.M., & Brown, F. (1981). Constancy of WISC IQs of Puerto Rican, White and Black children. In L. L. Adler (Ed.), Cross-cultural research at issue (pp. 137-143). New York, USA: Academic Press.

Melendez, F. (1994). The Spanish version of the WAIS: Some ethical considerations. Clinical Neuropsychologist, 8, 388-393.

Miccio, A., Tabors, P.O., Paez, M., Hammer, C.S., & Wagstaff, D. (2005). Vocabulary development in Spanish-speaking Head Start children of Puerto Rican descent. In J. Cohen, K.T. McAlister, K. Rolstad, & J. MacSwan (Eds.) Proceedings of the 4th International Symposium on Bilingualism (pp. 1614-1617). Somerville, USA: Cascadilla Press.

Monroe, W.S. (1921). Porto Rico school children and the Holley Picture Completion Test. School & Society, 14, 617-618.

Mora de Jesus, R.B. (1990). School-related stress in Puerto Rican children. Unpublished doctoral dissertation, University of Connecticut, USA.

Moriber, L. (1961). School functioning of pupils born in other areas and in New York City. Bureau of Educational Program Research and Statistics, Board of Education of the City of New York.

Morris, A.G. (1977). Effectiveness of educational intervention in a well baby clinic on the performance of low SES preschoolers. Unpublished dissertation, Fordham University, USA.

Morrison, J.C. (1958). The Puerto Rican study, 1953-1957; a report on the education and adjustment of Puerto Rican pupils in the public schools of the city of New York. NY, USA: Board of Education.

Murdoch, K. (1925). A study of differences found between races in intellect and in morality. School & Society, 22, 628-632, 659-664.

O’Neill, D.M., & Sepielli, P.J. (1985). Education in the United States, 1940-1983 (Vol. 81). US Department of Commerce, Bureau of the Census.

Oplesch, M., & Genshaft, J. (1981). Comparison of bilingual children on the WISC-R and the Escala de Inteligencia Wechsler Para Ninos. Psychology in the Schools, 18, 159-163.

Orsini, B. (1974). Intelligence testing and the culturally different: a comparison of Puerto Rican and black children. Unpublished paper, City College of New York, USA.

Ortiz Colón, R., Rosselló, J., Muñoz Vazquez, M., Maldonado Santiago, N., Suárez, N., & Donato Jímenez, L.J. (1993). Relación entre la exposición a sustancias tóxicas y el funcionamiento intelectual y afectivo de obreras textiles puertorriqueñas. Revista Puertorriqueña de Psicología, 9, 179-219.

Páez, M.M., Tabors, P.O., & López, L.M. (2007). Dual language and literacy development of Spanish-speaking preschool children. Journal of Applied Developmental Psychology, 28, 85-102.

Pennock-Roman, M. (1993). Differences among racial and ethnic groups in mean scores on the GRE and SAT: cross-sectional comparisons. ETS Research Report 91-13. Princeton, NJ: Educational Testing Service.

Peralta, C.A., Li, Y., Wassel, C., Choudhry, S., Palmas, W., Seldin, M.F., … & Shlipak, M.G. (2010). Differences in albuminuria between Hispanics and whites: An evaluation by genetic ancestry and country of origin The Multi-Ethnic Study of Atherosclerosis. Circulation: Cardiovascular Genetics, 3, 240-247.

Pons, J. I., Matías-Carrelo, L., Rodríguez, M., Rodríguez, J. M., Leticia Herrans, L., Jiménez, M. E., … & Yang, J. (2008). Estudios de validez de la Escala de Inteligencia Wechsler para Adultos Versión III, Puerto Rico (EIWA-III). Revista Puertorriqueña de Psicología, 19, 75-111.

Porteus, S.D. (1932). Human studies in Hawaii. In Pacific problems: the lectures and proceedings of the School of Pacific and Oriental Affairs, 1932 (pp. 82-111). Honolulu, USA, University of Hawaii.

Porteus, S.D., & Babcock, M.E. (1926). Temperament and Race. Boston, USA: Gorham Press.

Prewitt Diaz, J.O., & Munoz, G. (1980). A reliability study of the WISC with Puerto Rican children (grades K-9) in Puerto Rico. Unpublished paper, Pennsylvania State University, USA.

Prewitt Diaz, J.O., & Rivera, R. (1989). Correlations among scores on Woodcock- Johnson achievement subtest (Spanish), WISC-R (Spanish) and Columbia Mental Maturity Scale. Psychological Reports, 64, 987-990.

Rindermann, H. (2013). African cognitive ability: Research, results, divergences and recommendations. Personality and Individual Differences, 55, 229-233.

Risch, N., Choudhry, S., Via, M., Basu, A., Sebro, R., Eng, C., … & Burchard, E.G. (2009). Ancestry-related assortative mating in Latino populations. Genome Biology, 10, R132.

Roca, P. (1955). Problems of adapting intelligence scales from one culture to another. High School Journal, 38, 124-131.

Rodriguez, V.L., & Prewitt Diaz, J.O. (1990). Correlations among GPA and scores on the Spanish version of WISC-R and the Woodcock-Johnson Achievement subtests for 10- to 12-year-old Puerto Rican children. Psychological Reports, 66, 563-566.

Rodriguez, M.D., Rivera Ruiz, D., Prewitt Diaz, J.O. (1985). A study of the Spanish translation of the Wechsler Intelligence Scale for children-revised with Puerto Rican children and adolescents. Washington, D.C.: Department of Education.

Roselló, J., Hernández, J. J., Zayas, G., Ortiz, P., Matos, M., & Alvarez, L. (1988). Relación entre la contaminación con mercurio y el funcionamiento intelectual y afectivo en un grupo de niños de Humacao, Puerto Rico. Revista Puertorriqueña de Psicología, 5, 1-12.

Rushton, J.P., & Jensen, A.R. (2005). Thirty years of research on race differences in cognitive ability. Psychology, Public Policy, & Law, 11, 235-294.

Sahai, H. (1989). Relations of sociodemographic variables and cognitive ability: A comparative analysis of the cognitive scores of high school seniors. Perceptual & Motor Skills, 69, 1139-1157.

Salari, K., Choudhry, S., Tang, H., Naqvi, M., Lind, D., Avila, P.C., … & Ziv, E. (2005). Genetic admixture and asthma-related phenotypes in Mexican American and Puerto Rican asthmatics. Genetic Epidemiology, 29, 76-86.

San Miguel Montes, L.E., Allen, D.N., Puente, A.E., & Neblina, C. (2010). Validity of the WISC–IV Spanish for a clinically referred sample of Hispanic children. Psychological Assessment, 22, 465-469.

Sathy, V., Barbuti, S.M., & Mattern, K.D. (2006). The new SAT and trends in test performance. Statistical Report No. 2006-1. New York, NY: The College Board.

Shellenberger, S., & Lachterman, T. (1979). Cognitive and motor functioning on the McCarthy Scales by Spanish-speaking children. Perceptual and Motor Skills, 49, 863-866.

Shellenberger, S. (1982). Assessment of Puerto Rican children: a cross-cultural study with the Spanish McCarthy Scales of Children’s Abilities. Bilingual Review, 9, 109-119.

Shor, M.D. (1961). Psychological traits of a selected sample of Puerto Rican born homeless men : a comparative study. Unpublished doctoral dissertation, Yeshiva University, USA.

Shuey, A.M. (1966). The testing of Negro intelligence. NY, USA: Social Science Press.

Smith, S. (1942). Language and non-verbal test performance of racial groups in Honolulu before and after a fourteen-year interval. Journal of General Psychology, 26, 51-93.

Sowell, T. (1978). Race and I.Q. Reconsidered. In T. Sowell (Ed.), Essays and Data on American Ethnic Groups (pp. 203-238). Washington, DC, USA: Urban Institute.

Special cable (1936, Feb 11) Mentality report angers San Juan. New York Times, p 5.

Talerico, M., & Brown, F. (1963). Intelligence test patterns of Puerto Rican children seen in child psychiatry. The Journal of social psychology, 61, 57-66.

Tamayo, J.M. (1990). A validated translation into Spanish of the WISC-R vocabulary subtest words. Educational & Psychological Measurement, 50, 915-921.

Thomas, A., Hertzig, M.E., Dryman, I., & Fernandez, P. (1971). Examiner effect in IQ testing of Puerto Rican working-class children. American Journal of Orthopsychiatry, 41, 809-821.

Thomas, L. (2010). Puerto Rican Citizen: history and political identity in twentieth-century New York City. University of Chicago Press.

Torres Diaz, P., Romero Garcia, I., Rodriguez, Irizarry, W., Rodriguez Fernandez, C. (2009). Validacion concurrente de la Wechsler Intelligence Scale for Children-Spanish con la Escala de Inteligencia Wechsler para Ninos-Revisada de Puerto Rico. Revista Puertorriquena de Psicoloia, 20, 83-102.

United States Employment Service (1965). Technical report on standardization of the General Aptitude Test Battery: general working population norm study for Puerto Rico. U.S. Dept. of Labor, Washington, D.C., USA.

Vazquez, C.D. (1989). A comparison of Puerto Rican children’s performance on the Spanish version of the Wechsler Intelligence Scale for Children-Revised (WISC-R) and the Kaufman Assessment Battery for Children (K-ABC). Unpublished doctoral dissertation, University of Massachusetts, USA.

Via, M., Gignoux, C.R., Roth, L.A., Fejerman, L., Galanter, J., Choudhry, S., … & Martínez-Cruzado, J.C. (2011). History shaped the geographic distribution of genomic admixture on the island of Puerto Rico. PLoS One, 6, e16513.

Vincenty, N.I. (1930). Racial differences in intelligence as measured by pictorial group tests with special reference to Puerto Rico and the United States. Unpublished doctoral dissertation, Harvard University, USA.

Walsh, J.F., & D’Angelo, R. (1971). IQs of Puerto Rican head start children on the Vane Kindergarten Test. Journal of School Psychology, 9, 173-176.

Wicherts, J.M., Dolan, C.V., Hessen, D.J., Oosterveld, P., Van Baal, G.C.M., Boomsma, D.I., & Span, M.M. (2004). Are intelligence tests measurement invariant over time? Investigating the nature of the Flynn effect. Intelligence, 32, 509-537.