One of the talking points in the discussion of average group differences in IQ is Irish IQ. People on both the hereditarian and environmentalist side of the debate have seized on this tidbit of information to advance their arguments.
Lynn and Vanhanen (2006) estimated that the average IQ of the Irish population to be 92, which is noticeably lower than the rest of northern and western Europe. Lynn (2015, pp. 38-39) attributed this lower value to the dysgenic effect of Roman Catholicism (where more educated people become celibate nuns and priests) and selective migration of brighter individuals to other countries.
On the environmentalist side, some have noticed that the estimated mean IQ Ireland has increased noticeably in the late 20th century. Unz (2012), for example, used Lynn’s data to identify a 13-point increase from 1972 (IQ = 87) to 2000 (IQ = 100). Lynn’s international IQs are standardized so that the British mean is 100. Therefore, these numbers would indicate that the Irish-British IQ gap decreased from 13 points to . . . zero in just a generation. My colleague, Wilfred Reilly, also sees a massive increase in Irish IQ.
Environmentalists are heartened by this massive rise in Irish IQ. In that perspective, if one group difference can close so quickly, then the most controversial group mean IQ difference — between White Americans and African American — can also close. It is a plausible hypothesis, and it is worth investigating the Irish IQ data in order to see whether it provides grounds for hope regarding other group differences in IQ.
Origin of the Belief in Low Irish IQ
In surveying the history of Irish IQ data, I found nothing in the first 50 years of Irish IQ studies about this group having especially low IQs compared to other White populations. If Irish IQ were typically in the mid-80s during the 20th century, it would have caught some attention.
The first example I could see of any scientist isolating Irish IQs as being particularly low was Hans Eysenck. In 1971, he proposed that selective immigration over the centuries had lowered the average IQ of the population remaining in Ireland. His basis for this view was a study by MacNamara (1966) which “. . . found the Irish to have IQs which were not very different from those observed in American negroes, and far below comparable English samples” (Eysenck, 1971, p. 127).
This single quote from a book aimed at the general public not only created the belief in low Irish IQ, but also linked the topic to the interracial IQ gap in the United States.
Eysenck (1981) later revisited the idea of lower Irish IQ, though not saying that Irish IQ scores were similar to African Americans’ scores. This is probably because Lynn (1979) had since published an article estimating that the average IQ in the Republic of Ireland was 96. Eysenck (1981) reproduced a figure from Lynn’s article showing this information in his book.
MacNamara was not pleased to be dragged into this debate, and he wrote an article (MacNamara, 1988) disavowing Eysenck’s interpretation of his data. Benson (1987) also addressed the possibility of lower Irish IQ in Eysenck’s work and criticized the data behind claims of low Irish IQ, much of which comes from Richard Lynn.
Still, a scholarly argument over a few articles is not the best way to settle the debate about Irish IQ. A better strategy is to identify all the Irish IQ data that one can find in order to understand the entire literature on the topic and determine a typical average IQ for this population.
Irish IQ Data
I searched Google Scholar for the terms “irish IQ” (without quotes) and identified as many reported samples of Irish IQ data as I could. (Note that Google automatically returns search results for “Ireland IQ” with this search term.) I also searched references of literature review articles on the topic and the references in Lynn’s various international IQ data estimates. When possible, I tracked down the original source for Lynn’s data, though I used his IQ estimates if the original source was not available to me.
To be included in this analysis, data must (1) be reported as a mean or median IQ or easily convertible to a mean IQ, (2) include the sample size, (3) report data for an ethnic Irish sample separately from any other populations, (4) be derived from a test normed in the United States or the United Kingdom, (5) not be selected for high or low IQ, and (6) not be a clinical sample or a sample referred to a physical or mental health clinic.
The first record of someone with Irish ancestry being given an intelligence test comes from Terman (1917), who reported a study of intelligence of civil service candidates in San Jose, California, which occurred on October 31, 1916. Two of these people were Americans of Irish descent. One of these Irish American individuals was a streetcar conductor who scored 83. The other Irish person’s IQ was not reported.
In total, I found 55 samples that reported data from 28,751 examinees. The sample sizes ranged from 1 to 4,215 (mean = 522.6, median = 170, SD = 887.3), and were collected between 1916 and 2015. The unweighted mean IQ was 98.3 (median = 97.4, SD = 9.0). All the data were collected in the Republic of Ireland, Northern Ireland, or the United States. Here is a scatterplot showing the samples I identified:
The scatterplot shows that the early samples were collected in the United States. These examinees were Irish immigrants or the descendants (usually children) of immigrants. It is not until the 1960s and 1970s when IQ data collected in Ireland dominates the literature.
The weighted mean of all these scores is 97.2. At the study level, there is a slight linear trend (r = .272), where later studies report higher means — a trend that is barely statistically significant (p = .045). But when the data are weighted by sample size, this correlation disappears (r = -.087). It is clear that there has not been a massive increase in Irish IQ during the 20th or 21st centuries.
Differences among Irish examinees
However, the country of data collection matters. The weighted mean IQ for data collected in Ireland was 94.3. In the USA, it was 95.4, and in Northern Ireland, it was 100.2. However, these might be distorted by some outliers.
The two lowest scoring American samples were both from data collected during World War I and reported by Yerkes (1921): 205 illiterate, foreign-born Irish draftees taking the Army Beta (avg IQ = 80.9) and 25 low-functioning foreign-born Irish draftees taking the 1916 Stanford-Binet (avg IQ = 77.4). Both of these samples are clearly not representative of the general Irish population. Removing these individuals increases the American Irish weighted mean to 97.8.
Another group of outlier means comes from Sutherland (1961), who reported four samples of high school students in Northern Ireland who attended grammar schools. Most students in these samples obtained admission by scoring well on the 11+ examination. Dropping these samples lowers the Northern Ireland mean to 97.8. The IQ data from the Republic of Ireland is still lower than both of these adjusted means, but the differences among groups are much smaller. The weighted mean of all non-outlier data is 96.1. This information is summarized in the table below.
|Country||Unweighted mean||Weighted mean||Weighted mean (no outlier samples)|
The preferred mean is a subjective choice. In my opinion, the weighted means without the outlier samples are most trustworthy.
It is important to be aware of a methodological difference between some of the Irish data and the scores from elsewhere. Unlike the data from the U.S. or Northern Ireland, some of the data for the Republic of Ireland were taken from Lynn’s reports if I could not track down the original source. This occurred for the three of the largest samples from Ireland, collected by Byrt and Gill (reported by Lynn, 1979, p. 4), Raven (reported in Lynn & Vanhanen, 2002, Appendix 1), O’Connor et al., and Jeffers and Fitzgerald (both reported in Lynn, 2015). Lynn corrected for the Flynn effect in his averages, which lowers these sample means. That is a legitimate methodological choice, but because I did not do so when collecting data from the U.S. and Northern Ireland, the data from the latter countries is systematically higher than the Irish IQ. Without these four samples, the weighted mean IQ for the Republic of Ireland is 97.5 and the weighted mean IQ for all remaining samples is 97.7.
Regardless of the methodological choices, it is clear that typical scores for ethnic Irish individuals are in the mid-90s. Eysenck was, therefore, wrong in declaring that Irish examinees typically score at a level that is typical of African Americans at the time (which would have been in the mid-80s).
Indeed, with the exception of some outlier samples in the United States and Northern Ireland, Irish IQs are rather consistent, and it is clear that there has not been a massive increase in Irish IQ during the 20th or 21st centuries.
[Update: I have posted the data, R syntax, and markdown files for this analysis on GitHub.]
Analysis of a False Belief
How could Eysenck, Unz, and others have been wrong? The answer is that they did not consider the entire literature. Isolated samples in my collection do obtain mean scores in the 80s. But almost all of them are atypical in some way.
For Eysenck specifically, there seems to have been an eagerness to subscribe to his preferred interpretation of the data. Eysenck’s (1971, p. 127) claim that MacNamara (1966) showed that Irish IQs were “. . . not very different from those observed in American negroes, and far below comparable English samples” simply is not true. The relevant data in MacNamara’s book shows IQs from a non-verbal test and a verbal test, the Moray House Test (MacNamara, 1966, p. 101, Table 11.1). This table divides the sample into six groups, in ascending order of the degree of the use of the Irish language in instruction in school. The IQs range from 91.9 to 102.9, with a weighted mean of 100.3 — which is almost exactly equal to the British mean on these tests (not the African American mean on similar tests). Eysenck ignored this data in MacNamara’s study.
Even this value of 100.3 may be an underestimate. MacNamara (1966) oversampled Irish-language schools (which had students scoring lower than the other groups) in his sampling plan so that the groups would have approximately the same number of students. Reweighting the data so that schools are represented proportionately, causes to average IQ to increase by approximately 1.5 points. (An exact adjusted estimated IQ is not possible, because MacNamara did not report the necessary information to calculate a sampling weight for students — instead of schools.)
MacNamara (1966, 1988) also saw these scores as underestimates. He believed that his Irish-speaking examinees scored lower on the tests because a lack of test familiarity. According to him, none of the examinees had ever taken a standardized test before, which depressed their IQs. There is evidence that he was correct. In a different study of Irish schoolchildren (MacNamara, 1964), examinees took 6 different versions of the Moray House Test over the course of three weeks. The mean for the first examination was 90.1 (the IQ I use in this blog post). The average IQ increased with each test administration until the mean for the sixth test was 96.13. These Irish students gained an average of 1.208 IQ points every time they sat down to take a test. That’s quite a practice effect! Eysenck should (must?) have known about this 1964 study, because MacNamara discusses it in his 1966 book that Eysenck cited.
Additionally, Eysenck (1981) was aware of a 1979 article by Lynn in which he reported a mean IQ of 96.0 for the Republic of Ireland — a much higher value than was typical of African Americans at the time. But because the average Irish IQ in Lynn’s (1979) study was 6.1 points below Lynn’s (1979) average IQ for London and southeastern England, Eysenck still emphasized the Irish deficit — even though it was less than half of the magnitude he had previously believed.
Eysenck believed that the Irish-British IQ difference was genetic in origin, even though these two populations are genetically and culturally very similar. His argument was that selective migration caused smarter Irish individuals to emigrate to other countries. This would have a dysgenic effect on IQ in the remaining population and its descendants.
I do not discount the possibility that selective migration could have impacted the mean IQ of Ireland today. In the 21st century, emigration is positively associated with IQ (Belsky et al., 2016; Dutton et al., 2018). Maybe the same was true in 19th and 20th century Ireland. According to Wikipedia, 40%(!) of people born in Ireland were living abroad in 1890. If the propensity to emigrate were positively correlated with genes for higher intelligence in historic Ireland, then it could noticeably lower the IQ of the remaining population (and their descendants). But we simply do not know whether this actually occurred. Eysenck treated his supposition as fact.
Eysenck was disposed to believe that many (all?) group differences in IQ had a genetic explanation, and he did not let inconvenient data deter him from that belief. Ironically, some environmentalists have (probably unknowningly) built on Eysenck’s incorrect interpretation of the early data to argue that non-genetic explanations must be the cause of IQ gaps between groups.
As for Unz (2012), his conclusion of a massive rise in IQ is based on Lynn’s compilation of IQs (though I cannot determine which version). The lower IQs tend to be from older datasets (including MacNamara’s 1964 study). Unz found a correlation of r = .86 between IQ and year of study publication, apparently unaware that calculating correlations of trend data with small samples can produce misleading results. I’m willing to cut Unz more slack than Eysenck; at least Unz did not ignore data that undermined his beliefs.
I think there are important lessons to learn from this controversy. First, it is always best to go back to the original data. If people had checked the underlying data that Eysenck used to make claims about Irish IQ scores being similar to African American IQs, then they would have seen that Eysenck’s statements were not correct.
Second, a comprehensive search of the literature is always more informative than a single study. There are isolated studies of Irish IQ where the mean is below 90, including the World War I samples of illiterate men (Yerkes, 1921) and a study of children from deprived areas of Dublin (Kellaghan, 1977). Lynn and Vanhanen (2002, Appendix 1) also claim that a 1981 study in Ireland produced an IQ of 87. But most samples of Irish people produced means in the 90s or 100s. Those are far more likely to represent the population average IQ in Ireland.
Third, I believe that this study is an example of how national IQs can be improved. The latest version of Lynn’s dataset of national IQs (Lynn & Becker, 2019) is better than earlier ones (Warne, in press). However, the documentation of how Lynn identified datasets is lacking, and there is no discussion of why samples were included in the calculations (or not). Why has Lynn never included data from MacNamara’s larger (1966) study, despite its excellent sampling procedure? Why are some datasets dropped from the calculation of Irish IQs in later versions? It is not clear.
Finally, this analysis shows that the idea of a meteoric rise in Irish IQ during the 20th century is incorrect. The average Irish IQ has likely been close or equal to the mean British IQ for decades. Irish IQ is an interesting topic in its own right, but it probably has no relevance to discussion of large group differences in IQ.
Belsky, D. W., Moffitt, T. E., Corcoran, D. L., Domingue, B., Harrington, H., Hogan, S., Houts, R., Ramrakha, S., Sugden, K., Williams, B. S., Poulton, R., & Caspi, A. (2016). The genetics of success: How single-nucleotide polymorphisms associated with educational attainment relate to life-course development. Psychological Science, 27(7), 957-972. https://doi.org/10.1177/0956797616643070
Benson, C. (1987). Ireland’s low IQ: A critique of the myth. The Irish Journal of Psychology, 8(1), 61-70. https://doi.org/10.1080/03033910.1987.10557693
Dutton, E., Essa, Y. A. S., Bakhiet, S. F., Ali, H. A. A., Alqafari, S. M., Alfaleh, A. S. H., & Becker, D. (2018). Brain drain in Syria’s ancient capital: No Flynn Effect in Damascus, 2004–2013/14. Personality and Individual Differences, 125, 10-13. https://doi.org/10.1016/j.paid.2017.12.025
Eysenck, H. J. (1971). Race, intelligence, and education. Temple Smith.
Eysenck, H. J. (1981). Racial and cultural factors. In H. J. Eysenck & L. Kamin (Eds.), Intelligence: The battle for the mind (pp. 74-83). Pan Psychology.
Lynn, R. (1979). The social ecology of intelligence in the British Isles. British Journal of Social and Clinical Psychology, 18(1), 1-12. https://doi.org/10.1111/j.2044-8260.1979.tb00297.x
Lynn, R. (2015). Race differences in intelligence: An evolutionary analysis (2nd ed.). Washingtom Summit Publishers.
Lynn, R., & Becker, D. (2019, June 21). The NIQ-Dataset version 1.3.3. https://viewoniq.org/
Lynn, R., & Vanhanen, T. (2002). IQ and the wealth of nations. Praeger.
Lynn, R., & Vanhanen, T. (2006). IQ and global inequality. Washington Summit Publishers.
MacNamara, J. (1964). Zero error and practice effects in Moray House English quotients. British Journal of Educational Psychology, 34(3), 315-320. https://doi.org/10.1111/j.2044-8279.1964.tb00642.x
MacNamara, J. (1966). Bilingualism and primary education: A study of Irish experience. University of Edinburgh Press.
MacNamara, J. (1988). Race and intelligence. The Irish Review, (3), 55-60. https://doi.org/10.2307/29735312
Sutherland, M. B. (1961). Co-education and school attainment. British Journal of Educational Psychology, 31(P2), 158-159. https://doi.org/10.1111/j.2044-8279.1961.tb02928.x
Terman, L. M. (1917). A trial of mental and pedagogical tests in a civil service examination for policemen and fireman. Journal of Applied Psychology, 1(1), 17-29. https://doi.org/10.1037/h0073841
Unz, R. (2012, August 14). Race/IQ: Irish IQ & Chinese IQ. American Conservative. https://www.theamericanconservative.com/raceiq-irish-iq-chinese-iq/
Warne, R. T. (in press). National mean IQ estimates. Validity, data quality, and recommendations. Evolutionary Psychological Science. https://doi.org/10.1007/s40806-022-00351-y
Yerkes, R. M. (1921). Psychological examining in the United States army. Government Printing Office.