Abstract
In the UK, immigrant groups frequently have lower mean socioeconomic status (SES) than do White British, which is a source of concern for the British government. Group-level SES tends to show positive relationships with cognitive ability scores. Thus, the authors estimate the mean cognitive and SES scores of various ethnic groups and test empirically if they correlate. They compute SES and cognitive ability scores using high-quality representative samples of adults. They then computed correlations between the two measures. General SES and group-cognitive ability correlated strongly at r = .59 to r = .79 (N = 18 groups). Finally, the authors computed cognitive scores predicted by the nation or region-of-origin of the ethnic groups and calculated correlations between these expected scores and the measured scores. The predicted and measured scores correlated strongly at r = .93 (N = 16 groups). The authors conclude that ethnic differences in SES are partly linked to differences in cognitive ability.
1 Introduction
1.1 Socioeconomic Status and Ethnicity in the UK
Socioeconomic status (SES) is a measure of the social standing of an individual or group in society. SES is an important indicator of human functioning, well-being, health, and development (APA, 2007; Kezer & Cemalcilar, 2020). Moreover, researchers often conceptualize SES as an indicator of an individual’s (or group’s) access to social, cultural, financial, and human capital resources (APA, 2007). Similarly, researchers typically measure SES in terms of income, occupation, education, or neighborhood quality.
Following the British Nationality Act of 1948, the United Kingdom (UK) experienced a substantial influx of non-European migrants. By 1981, the UK population was 6% non-European (Owen, 1995). Today, this value is 13%, based on the most recent census data, namely those from 2011, and will likely increase to 26% by 2051 (Rees, Wohland, Norman, Lomax, & Clark, 2016). These migrants to the UK come from various localities: roughly 38% are South Asian, 16% are Chinese and “other Asian” (excluding South Asians), 25% are African/Black, and 22% are mixed/other. These minority groups are geographically concentrated, primarily located in Greater London, the West Midlands, Greater Manchester, West Yorkshire, or the Leicester/Nottingham region.
UK government reports show moderate to large mean differences in SES by race and ethnicity. Examples of these reports include the Racial Disparity Audit of 2017 (Cabinet Office, 2017), The McGregor-Smith Review (McGregor-Smith, 2017), and The Lammy Review (Lammy, 2017). In these reports, and across many measures, Chinese, Whites, and Indians experience relatively better social outcomes; whereas, Blacks, Pakistanis, Bangladeshis, plus those of mixed ethnic backgrounds, tend to experience worse outcomes.
1.2 Explaining Group Differences in SES: Discrimination and Racism
Researchers often attribute ethnic differences in SES to the direct effects of discrimination and racism (McGregor-Smith, 2017; Ashe & Nazroo, 2016). For example, according to the McGregor-Smith’s review on disparities in the workplace:
In the UK today, there is a structural, historical bias that favours certain individuals. This does not just stand in the way of ethnic minorities, but women, those with disabilities and others… Conscious or unconscious, the end result of bias is racial discrimination, which we cannot and should not accept… There is discrimination and bias at every stage of an individual’s career, and even before it begins.
MCGREGOR-SMITH, 2017, p. 2–3
Employment audit studies seemingly bolster the hypothesis that racial discrimination directly causes outcome differences in SES. For example, these studies show that certain non-White UK groups receive fewer interview callbacks, despite having similar qualifications with matched White applicants (Department for Work and Pension, 2009; Di Stasio & Heath, 2019). However, the limitations of using audit studies to infer discrimination have been previously discussed (Heckman, 1998). Importantly, individuals of different ethnic groups cannot be matched on all variables valued by employers, such as communication and social skills (Mobius & Rosenblat, 2006). Moreover, while these studies typically match on education, there are often unmatched cognitive or personality differences between ethnic groups of similar educational levels (Heckman, 1998).
1.3 Explaining Group Differences in SES: Cognitive Ability
An alternative account for SES differences across UK ethnicities focuses on cognitive competency as defined by Rindermann (2018, p. 43): “[T]he ability to think (intelligence), knowledge (the store of true and relevant knowledge) and the intelligent use of knowledge.” By this definition, cognitive competency includes general intelligence, broad abilities, and/or the specific mental abilities measured by cognitive tasks.
In various studies, group differences in average cognitive ability correlate positively with group-level measures of SES (Pesta, McDaniel, & Bertsch, 2010). As it is, the finding that cognitive ability varies across ethnic groups in certain countries, such as the USA, is one of the most replicated effects in psychology (Baron, Martin, Proud, Weston, & Elshaw, 2003). But, whether current ethnic groups in the UK, specifically, differ in cognitive competency and whether such differences predict SES differences between ethnic groups has not been thoroughly studied.
Cognitive models, of course, are not incompatible with discrimination models. For example, cognitive ability differences could lead to adverse impact, or indirect discrimination, when cognitive selection tests are used for making decisions in employment and education. Cognitive models simply propose that cognitive ability differences are antecedent to social outcome differences and thus that addressing SES disparities requires considering cognitive differences.
1.4 Group Differences in Cognitive Ability in UK
Much research focuses on monitoring the magnitudes of ethnic differences in the USA (Roth, Bevier, Bobko, Switzer, & Tyler, 2001; Roth et al., 2017). However, differences may not generalize from one country to the next since ethnic diasporas often have radically different histories, and they are often not representative of their region-of-origin populations (e.g., South Asians in the USA, the UK, Guyana, Kenya, and Trinidad & Tobago).
Regarding the UK, earlier research has shown that Black and South Asian children generally have lower mean scores on cognitive ability tests relative to White British children (Lynn, 2008; Taylor & Hegerty, 1985; Tomlinson, 1980; Tomlinson, 1983). Additionally, literature reviews from the early 20th century have noted ethnic differences on occupational and military selection tests (Baron et al., 2003; Evers, te Nijenhuis, & van der Flier, 2005). However, this literature is based primarily on convenience samples.
The literature is also dated and relies either on adolescent samples from the 1960s to the 1990s or adult samples from the 1990s and the early 2000s. Given possible secular and/or age-related changes in the magnitude of ethnic gaps (Dickens & Flynn, 2006), and also continual compositional changes due to ongoing immigration (e.g., African British now comprise a majority of “Black British,” displacing the Caribbean as the major source of Afro-descent migrants), both the magnitude and the direction of differences may not be generalizable across time or geography. Thus, the literature continually needs to be updated (Roth et al., 2001).
1.5 Research Question
Many British studies have suggested that discrimination causes group differences in SES, but no research exists on how well group differences in cognitive ability might also predict the SES differences. We thus estimated the mean cognitive scores of various ethnic groups by using only high-quality, nationally representative samples. We also empirically tested how well cognitive ability predicts ethnic differences in SES. That is, we first estimated global SES scores by ethnicity, and then we estimated the mean cognitive scores for various ethnic groups. After that, we tested how well-derived cognitive scores predict ethnic group differences in SES. Finally, we examined to what extent ethnic group differences in cognitive ability can be accounted for by country-of-origin differences (i.e., via the World Bank’s harmonized national cognitive scores).
2 Methods
2.1 Samples, Variables, and Methods
We carried out a large number of computations on our data and reporting them all would have led to an overly long article. While many details are reported in the Appendix, others were excluded due to the length of the manuscript. Researchers who are interested in more detailed outcomes of analyses can contact the authors.
2.2 Ethnic Groups
In certain governmental reports, groups distinguishable by hereditary traits are referred to as “racial groups”. However, the terminology used to describe European and non-European groups has changed. Currently, the UK government uses the term “ethnic group” (UK Government, 2019a). As such, we follow contemporary terminology and refer to the groups herein as “ethnic groups.” The UK Government recommends the use of 18 narrow ethnic groups (UK Government, 2019a). We sorted these as follows:
1) White (English, Welsh, Scottish, Northern Irish or British; Irish; any other White),
2) Mixed or Multiple (White and Black Caribbean; White and Black African; White and Asian; any other Mixed or Multiple),
3) Chinese and Other Asian (Chinese; Any other Asian, which includes all other geographically-defined Asians not otherwise mentioned),
4) South Asian (Indian; Pakistani; Bangladeshi),
5) Black (Caribbean or Black British; African; any other Black),
6) Other ethnic groups (Any other ethnic group; Arab; Gypsy or Irish Traveller).
2.3 Socioeconomic Status
2.3.1 Socioeconomic Status Variables
We searched for reports on adult socioeconomic differences, focusing on those providing data for the 18 narrow UK ethnic groups (presuming the data could be converted into effect sizes). We identified the following variables: (1) arrest rate, (2) the percentage living in a deprived neighborhood, (3) the percentage economically inactive, (4) net income, and (5) the percentage achieving at least three A-levels in any subject.
The latter four variables fit with APA’s (APA, 2007) expanded conception of SES (which includes neighborhood quality). Here, percent unemployment indexes occupational/employment status, while percentage achieving at least three A-levels indexes educational quality. We further include the social outcome of arrest rate because 1) this variable is a specific concern of the government, 2) it positively covaries with the other indexes and loads well on the general factor of socioeconomic status, and 3), conceptually, it is a proxy for involvement with the criminal justice system (which, due to common stigmas, can limit social, cultural, and financial capital).
We next describe these variables in greater detail.
(1) Arrest rate: The UK Government (2020) published arrest rates out of one thousand by ethnic group between April 2018 and March 2019. However, data were not provided for Arabs or Gypsies but rather for the total “Other” ethnic group.
(2) Deprived neighborhoods: Jivraj and Khan (2015) reported the percentage of people living in deprived neighborhoods in 2011 by ethnic group. Deprived neighborhoods are areas with a high proportion of adults with low socio-economic status and are characterized by indicators such as low-paying jobs, low education, low income, and high levels of unemployment. The linked data file includes percentages where deprived is defined based on income, employment, health, education, barriers to housing, crime, and living arrangement. We averaged these seven percentages.
(3) Economically inactive: Kapadia, Nazroo, and Clark (2015) report the percent of people who were economically active in 2011 by ethnic group. For this analysis, we used the “Percent economically active, 25–49-year-old men” as an indicator since there are cultural variations in expectations for women to work.
(4) Net income: We computed net income using the UK Household Longitudinal Study. This study is a large panel survey covering England, Scotland, Wales, and Northern Ireland. We used data from wave three of the survey, the same year, the cognitive data discussed below were available.
(5) A-levels: As a measure of educational quality, we use the percent of people achieving three or more A-levels. A-levels are non-compulsory, subject- based, class performance assessments. These are taken between ages 16 and 18 for entrance both into higher education and higher apprenticeships. Employers also value this distinction since it signifies a quality education. As an index of social access, A-levels are, in many ways, preferable to years of education since many people in the UK go directly to apprenticeships. The UK Government (UK Government, 2019b) published the percent of people achieving at least three A grades at A-level in any subject. We averaged percentages, weighted by the number of students in each cohort, from the 2010 to 2017 school years.
2.3.2 Effect Size Computations for Social Outcomes
We computed effects sizes for the five socioeconomic outcomes. For net income, we calculated means and standard deviations by ethnic group. Using these data, we computed Cohen’s d using the standard deviations pooled across all ethnic groups. For the other variables, we had neither means nor standard deviations. As such, we applied an inverse cumulative transformation to the percentages to compute the standardized normal deviate with respect to the White group. Ho and Reardon (2012) denote this value as dtpac, and note that dtpac is only equal to Cohen’s d on condition of normality and equal variances. For summary socioeconomic scores, we computed SES average scores from the five d and dtpac values; additionally, we computed general factor scores based on the ethnic group’s SES scores (using principal factor analysis and replacing missing variables with means). For the latter, the factor loadings were as follows: arrest rate (.735), deprived neighborhood (.726), economically inactive (.360), net income (.669), and three or more A-levels (.925). There was a one-factor solution, which explained 58% of the total variance.
2.4 Cognitive Ability Samples
We conceptualize cognitive ability following Rindermann (2018, p. 43): “[T]he ability to think (intelligence), knowledge (the store of true and relevant knowledge) and the intelligent use of knowledge.” Understood here, “cognitive ability”, is not limited to stratum-III of the three-stratum intelligence model. That is, our focus is on measured cognitive ability differences, not the latent source of these. This procedure is comparable to how meta-analyses on ethnic differences in the US are conducted (Roth et al., 2001) in that the focus is on measured scores.
For the present article, we focus only on national samples. These samples have the advantage over many samples used for selection tests, for example, by being representative. Since our concern is with the predictors of contemporaneous SES differences, we investigate only cognitive ability differences in adults (i.e., individuals approximately 18 years of age or older). We limit consideration to surveys conducted this century since it is not clear that differences based on the second half of the 20th century reflect current differences among adults.
We searched the UK Data Service for samples that met five criteria. These included that: (a) the sample was comprised of UK adults, (b) the sample had a nontrivial number of specific non-White groups (e.g., Indian; n > 50), (c) the survey contained a reasonable measure of cognitive ability, (d) the data were publicly available, and (e) the survey was published this century.
We identified five datasets that met these criteria: The Adult Psychiatric Morbidity Survey 2000; Skills for Life, 2003/11; The Adult Psychiatric Morbidity Survey 2007; UK Household Longitudinal Study, 2009; and The Millennium Cohort Study, 2015. We further conducted a Google Scholar search for papers on ethnic differences in cognitive ability. Through this search, we identified an additional study, the Programme for the International Assessment of Adult Competencies (OECD, 2013). We obtained this dataset via a Freedom of Information Act request. Results from all six samples are discussed and analyzed in detail below. When possible, we decompose results by region of birth (UK-born or otherwise), language proficiency (English as a first language or otherwise), and age cohort.
Non-British samples may be small, but they are representative. By aggregating several smaller samples of non-British people, we arrive at more substantial sample sizes. We used the powerful principles of meta-analysis here: small samples can be combined over different studies.
2.4.1 Specific Samples
APMS (2000) and APMS (2007): Since 1993, The Adult Psychiatric Morbidity Survey has been conducted every seven years. This nationally representative survey provides data on the prevalence of specific psychiatric disorders. In 2000, the participants were from England, Scotland, and Wales and ranged in age from 16 to 74 years (Bebbington et al., 2000). In 2007, participants were from England only, and all were 16-years-old or older (McManus et al., 2009). For English-speaking participants only, the APMS includes the administration of the National Adult Reading Test (NART). The NART is a vocabulary exam that requires participants to correctly pronounce irregularly-spelled words (e.g., “NAÏVE,” “EPITOME”). The exam was developed to predict premorbid cognitive ability in patients with neuropsychological conditions. It has been found to strongly correlate with Wechsler Test of Adult Reading (WTAR) Scores measured at the same age (r = .89) and also with cognitive ability measured in childhood (r = .68) (Dykiert & Deary, 2013).
For the 2000 wave, scores were available for the following ethnic groups: White; Oriental and other Asian; South Asian; Black; and Other. For the 2007 wave, scores were available for: White British; White non-British; South Asian; Black; Mixed, and Other. We requested data from the 2014 wave but were unable to gain access due to data use restrictions.
PIAAC and Skills for Life: The Programme for the International Assessment of Adult Competencies (PIAAC) is a worldwide study of cognitive skills (literacy, numeracy, and problem solving) coordinated by The Organization for Economic Co-operation and Development (OECD, 2013). The first assessment was conducted in 2012 and involved individuals ranging in age from 16 to 65 years.
UKHLS: The UK Household Longitudinal Study (UKHLS) is a large panel survey covering England, Scotland, Wales, and Northern Ireland. Data collection began in 2009. In wave three of the survey, conducted between 2011 and 2013, six measures of cognitive ability were given to those aged 16 years or older. There were six elements to the cognitive module.
Prior to assessment, Gray, D’Ardenne, Balarjan, and Uhrig (2011) conducted a qualitative assessment of potential language biases with the measures. Their recommendations were implemented. To further minimize bias, the test questions were translated into nine languages – Arabic, Bengali, Cantonese, Gujarati, Punjabi (in either Gurmukhi or Urdu script), Somali, Urdu, and Welsh – for those with English language difficulties.
MCS: The Millennium Cohort Study (MCS) is a survey conducted by the Centre for Longitudinal Studies. It follows a sample of over 18,000 individuals born in the UK between 2000 to 2001. Results for children in this sample have been previously reported (Lynn & Cheng, 2013; Zilanawala, Kelly, & Sacker, 2016; Skopek & Passaretta, 2018, Hoffmann, 2018). Parents were interviewed extensively. In Wave 6, collected from 2015 to 2016, parental verbal ability was assessed with a 20-question (multiple-choice) vocabulary test, wherein participants selected synonyms for presented words (e.g., MOAN; WAIL). Ethnic categories included: White, Mixed, South Asian (Indian, Pakistani, Bangladeshi), Other Asian (Chinese, Other Asian), Black (Caribbean, African, Other Black), and Other Ethnic. Finally, scores were split by region of birth (UK or foreign).
2.4.2 Effect Size Computations for Cognitive Ability
We created two summary variables to compare with socioeconomic status. For the first, we use scores based on the full samples, irrespective of region of birth and language proficiency, since the socioeconomic status scores are based likewise. For the second, we used only individuals born in the UK or who spoke English as their first language. We did this because previous research has suggested that the relation between socioeconomic status and ethnicity is mediated by English proficiency (Aoki & Santiago, 2018). For the summary cognitive ability scores, we n-weight-averaged the scores across the samples.
2.4.3 Predicted Differences Based on Countries of Origin
To determine if country of origin differences could explain UK ethnic differences, we mapped the World Bank’s Human Capital Index Harmonized Test Score values (World Bank, 2020) onto the ethnic groups. To do this, we used the Office for National Statistics (ONS, 2015) 2011 census data and the Racial Disparity Unit (2020) report to identify the major nations of origins of the ethnic groups. We then weighted the Harmonized Test Scores for the nations by the number of non-UK-born individuals in the UK from these nations to get expected cognitive differences based solely on region-of-origin. For mixed ethnic groups (e.g., African-White), we averaged the scores for the parental groups. We then correlated these countries of origin predicted scores with the summary cognitive scores.
3 Results
3.1 Results for Social Outcomes
Table 1 shows the d-equivalent values for arrest rates, percentages living in deprived neighborhoods, percentages economically inactive, net income, and percentages achieving three or more A-levels in any subject. Following convention (Roth et al., 2001), we coded the variables so that a positive effect size represents worse socioeconomic performance relative to the White group. For net income, the d-values were small owing to the large variance of income within the White British group.
Across ethnic groups, the variables formed a positive manifold. A positive manifold refers to the almost lawlike phenomenon that scores on a wide collection of very different IQ tests correlate strongly to very strongly and positively with each other. So, verbal tests correlate strongly with non-verbal tests and spatial tests correlate strongly with memory tests. This indicates there is a common latent dimension which is traditionally known as general intelligence or g, for short. This concept of a positive manifold can be applied to other variables, such as SES. This positive manifold justified the combination of these variables into summary scores. For these, we computed both the average d-equivalent scores and general socioeconomic factor scores based on the 18 narrow ethnic group scores mentioned above. These two scores are shown in the last column of Table 1. The average and the general factor scores correlate at r = .94 (N=18) (rho = .90; N = 18).
Effect sizes for the socioeconomic variables along with summary scores by ethnic group
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
3.2 Results for Specific Cognitive Samples
Detailed results for the cognitive samples are provided in the Appendix. Therein scores are also provided by age cohort and nationality. Table 2 gives the mean scores (SD = 15) by ethnic group for the six samples (with the two APMS samples combined). Table 3 repeats Table 2, but just for UK-born (PIAAC, UKHLS, and MCS) or English-as-a first-language (SfL) individuals. Data from APMS were excluded from Table 3 since scores were not available by either birth region or rearing language.
Scores (in IQ metrics) by UK ethnic groups for the six cognitive samples based on all individuals
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
Scores (in IQ metrics) by UK ethnic groups for the six cognitive samples based on UK born (PIAAC, UKHLS, and MCS) or English as a first language (SfL) individuals
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
3.3 General Discussion of Cognitive Samples
Summary results from the six national studies for all individuals are provided in the first two columns of Table 4. For these, we n-weighted the scores for the six samples from Table 2. The only group excluded was the “Other & Mixed” sample from APMS 2000/2007. This exclusion was because this combined category from AMP was a heterogeneous group that did not align with our ethnic categories. We also computed n-weighted summary scores for just UK-born and English-as-first-language individuals. These scores were based on the PIAAC, SfL, UKHLS, and MCS data shown in Table 3. These scores are presented in the last two columns of Table 4.
Summary results (in IQ metrics) from national studies of ethnic differences in cognitive abilities among UK adults
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
The two sets of cognitive scores correlated at r = .88 (N = 18 groups) (rho = .92; N = 18 groups) with one another. Weighted by the square root of the harmonic means of the two samples to take into account unreliability due to small sample sizes, the correlation was r = .93 (N = 18 groups). Thus, ethnic means both for all individuals, and for UK-born/English-as-a-first-language individuals, are quite similar.
It should be noted that these summary results exclude notable heterogeneity across birthplace, generation, and nation of origin. More details are provided in the Appendix. The reason for the birthplace and generational differences is not clear. Acculturation is likely a substantial part of the explanation; however, changes in the emigrant source populations may also play a role. Additionally, for some ethnic groups, large discrepancies also existed between verbal and non-verbal based tests. For example, the Chinese scored 100.41 on the numerically loaded UKLS general factor but only 89.66 on the MCS vocabulary test.
3.4 Correlation between Socioeconomic Status and Cognitive Ability
Given the concern about social inequality between ethnic groups in the UK (Cabinet Office, 2017; McGregor-Smith, 2017; Lammy, 2017), surprisingly little attention has been paid to variability in measured cognitive ability in the UK. To address the gap in the literature, we next examine the relation between SES and cognitive scores. The correlation matrices for the summary cognitive measures and the socioeconomic measures are shown below in Table 5.
Correlation for cognitive and socioeconomic status variables
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
The original Cognitive x SES associations were inverted, because for SES we used conventional d-values (from Table 1), in which a positive score indicates lower scores; while for cognitive ability we used IQ-metric scores, in which a positive score indicates higher scores. For ease of interpretation, we reverse the signs of the correlations so that a positive correlation indicates both an increase in cognitive ability and an increase in better SES outcomes.
The cognitive ability scores, based on only UK-born or English-as-a-first-language individuals, correlated with average SES at r = .59 and the general factor of SES at r = .75. The respective Spearman’s correlations were rho =.56 and rho = .77. The cognitive ability scores based on all individuals correlated with average SES at r = .70 and with the general factor of SES at r = .79; this relation is depicted in Figure 1. The respective Spearman’s correlations were rho = .66 and rho = .81. These are strong associations. As the outcomes are comparable for a correlation based on an approximately normal distribution and a correlation based on a non-normal distribution, we reported only the value of the Pearson coefficient in Table 5.
Regression plot for cognitive ability scores and the general factor of SES by ethnic group using all individuals (N=18)
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
There is quite a bit of variability in the size of the correlations between cognitive variables and socioeconomic status variables. For Cognitive UK born, the correlations range from a low of r = .145 for Economically Inactive to a high of r = .829 for Deprived Neighborhood. For Cognitive all, the correlations range from a low of r = .298 for Economically Inactive to a high of r = .845 for Deprived Neighborhood. Figure 2 depicts the regression plots for Cognitive all and each of the five SES component d-values.
Regression plots for cognitive ability scores and SES component scores by ethnic group
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
An additional analysis shows that for Cognitive UK born, the correlations between cognitive variables and socioeconomic status variables on the one hand and factor loadings of the SES variables on the general SES factor show a value of r = .70; the corresponding value for Cognitive all is r = .68. This shows that the relationship between cognitive variables and socioeconomic status variables is very strongly determined by how well the specific SES variables measure the construct of SES: the weakest measures of SES show the weakest link with cognitive variables and the strongest measures of SES show the strongest link with cognitive variables.
As seen in Table A3 and Table A6 in the Appendix, migrant groups perform differently depending on nationality. To see if country of origin effects can partially explain UK ethnic group differences, we computed expected differences. To do this, we mapped the World Bank’s national Harmonized Test Score values (World Bank, 2020) onto the corresponding UK ethnic groups. We were not able to identify the source countries for the Other Mixed and the Black Other groups. For Gypsy/ Irish Traveller, we used the score of Romania for Romanichal Travellers and Ireland for Irish Travellers. Based on reports, this group is approximately 75% Romanichal Travellers, so we weighted the scores accordingly.
The predicted scores correlated at r = .93 (rho = .92) with the measured scores based on all individuals; this relation is depicted in Figure 3. They also correlated at r = .77 (rho = .77) with the measured scores based only on UK-born or English-as-a-first-language individuals. Moreover, the predicted scores also correlated with the general socioeconomic factor scores at r = .68 (rho = .68) and the average socioeconomic scores at r = .56 (rho = .50). Thus, UK cognitive ability differences seem to track source-country differences.
Regression plot for cognitive ability scores and international test scores by ethnic group using all individuals (N=16)
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
4 Conclusion
4.1 Discussion
Many British studies have suggested that discrimination causes UK group differences in SES, but there is little research on how well group differences in cognitive ability predict these SES differences. So, in this study, we estimated the mean cognitive ability scores of various ethnic groups, using only high-quality representative samples, and we tested empirically how well they predicted ethnic differences in SES. Chinese, Whites, Indians, and White-Asians ranked highest in terms of SES, followed by Other Asians, White-Africans, Any Other, Arabs, and Other Mixed. Finally, Pakistani, Bangladeshi, Blacks, and Gypsy/Irish Travellers ranked lowest. There were clear differences in mean cognitive ability scores. Whites and Chinese scored highest, followed by Mixed and Gypsy/Irish Travellers. South Asians, Blacks, and Others performed least well. Moreover, group cognitive ability predicted general SES strongly, albeit clearly not perfectly. While ethnic group differences in cognitive ability themselves could be attributed to the effects of ongoing racism and discrimination within the UK, we find that these cognitive differences between ethnic groups are strongly related to those between the ethnic group’s source countries and the corresponding social, cultural, and educational differences between these. It should not be entirely surprising that first and second generation immigrants from countries with less developed educational systems, as indexed by international test scores, tend to perform poorly on cognitive measures in the UK. This finding makes explanations for the origin of the cognitive differences based on racism and discrimination less plausible. Discrimination, however, could play a role in perpetuating differences.
Moreover, cognitive differences could lead to indirect discrimination. This is because, in the UK, cognitive selection tests are frequently used. For example, a survey by the Chartered Institute of Personnel and Development (CIPD, 2017) found that 41%, 53%, and 38% of organizations in the UK rely on tests of general ability, specific skills, and literacy/numeracy, respectively. While concerns about ethnic discrimination led to the passage of the Equality Act 2010, which outlaws indirect discrimination (in the absence of an objective justification for selection criteria), the legal environment is relatively relaxed. For example, while employers must show that selection tests are justified, and that they took reasonable measures to prevent ethnic discrimination, they are not required to conduct validity studies showing that their measures are statistically unbiased (Shen et al., 2017). Given the prevalent use of cognitive selection tests and the relatively relaxed legal environment, along with the cognitive differences among ethnic groups, indirect discrimination from the use of cognitive tests could lead to or perpetuate disparities in education and employment. Addressing this concern will require the close monitoring of cognitive differences (Roth et al., 2001).
It is interesting to look at different possible interpretations of the results. A reviewer mentioned that the group-level correlation between cognitive ability and SES might just reflect a correlation at the individual level. Indeed, it has been established for decades that there is a clear positive correlation between IQ scores and SES at the individual level (Jensen, 1998). Findings at the individual level do not necessarily translate into findings at the group level, but in our study, the relationship is also shown at the group level. We would argue that the findings at both the individual level and the group level strengthen and not weaken each other. Correlations at the individual level and at the group level, also known as ecological correlations, do not necessarily have to be of the same size. In the present study, the ecological correlations are clearly larger than the correlations at the individual level; a classical explanation is that there is less noise at the ecological level.
A reviewer mentioned that another different possible interpretation of the results is that there is some degree of “statistical discrimination” – that group membership is taken as revealing something about individual qualities. According to the theory of statistical discrimination, group inequality could take place when assessors have imperfect information about the individuals they interact with. Indeed, although our five SES variables are reported at the group level, they are an aggregation of individual-level assessment data, which makes it difficult to rule out some degree of statistical discrimination. However, cognitive assessments tend to be standardized and do not require evaluations by assessors, so there is no place for statistical discrimination.
4.2 Limitations
While mostly representative, the samples analyzed here have relatively low sample sizes and high standard errors for many ethnic groups. While this concern is minimized in the case of the larger ethnic groups (e.g., Asians, Blacks) for which it was possible to meta-analyze data, the estimates for the smaller ethnic subdivisions (e.g., Chinese, Other Asian) are less certain. Therefore, it would be worthwhile to complement this study by reviewing scores on employment selection tests (Roth, Huffcutt, & Bobko, 2003). These latter measures have considerably better psychometric properties, albeit the samples are not representative of the general population.
Additionally, we focused on measured cognitive ability and did not explore the issue of psychometric bias for general and broad latent ability, for example, by testing measurement invariance. While such an exploration is necessary to understand the psychometric nature (and ultimately the cause) of the observed differences, it is not necessary for the quantification of the magnitude of the measured differences. These differences – for example, in the ability to solve every day mathematical problems (e.g., PIAAC, UKHLS), represent real and practically important cognitive differences as defined here (Rindermann, 2018). In the generally accepted three-stratum model of intelligence, they represent stratum-I cognitive-ability differences (Carroll, 1997). Such differences are relevant to everyday functioning in relation to capital accumulation and socioeconomic attainment (Lane & Conlon, 2016).
Since narrow cognitive differences can lead to social inequalities irrespective of their correspondence to gaps in latent general ability, the issue of measured differences is of interest, regardless of their psychometric cause (Baron et al., 2003; Evers et al., 2005; Roth et al., 2001; Roth et al., 2003; Roth et al., 2017). For this reason, it is important to track such cognitive differences in the population. This tracking is done for many countries, including the UK (Baron et al., 2003; Evers et al., 2005) however, as discussed, the literature is dated for the UK.
A reviewer raised the possibility that reporting measured cognitive differences could lead to stereotypes, which could then perpetuate social inequalities. For this reason, we were clear to emphasize that we were focused on measured cognitive differences, not differences in general intelligence. These observed differences could, for instance, arise owing to education opportunities in the countries of origin. It is important to recognize these differences and their relations to socioeconomic inequality to take steps to mitigate their effects.
As noted above, a major concern is adverse impact from the pervasive use of cognitive tests in the UK (CIPD, 2017). Research into group differences can inform policy on the use of these tests (Baron et al., 2003; Evers et al., 2005). Such research can also guide employers when it comes to the appropriate use of tests. For example, we find a large numerical/verbal discrepancy for Chinese and Asians in general. Similar results have been reported for the United States (Roth et al., 2017). This discrepancy could be due to linguistic and cultural effects. Regardless, the results suggest that verbal tests will have more adverse impact on British of Asian heritage than quantitative ones.
We are fully aware that discrimination can have profound effects on life outcomes. However, the analyses presented above make direct discrimination less likely as the cause for UK differences in cognitive test scores, as the differences can largely be accounted for by region of origin. Also, the cognitive differences we observed among contemporary adults in the UK are roughly consistent with what has been reported among children and adolescents in the second half of the 20th century. This comparability might indicate a certain degree of temporal stability for the decades under study but does not necessarily imply that these cognitive differences will be reproduced in subsequent generations born in the 21st century. Indeed, educational data from the current century suggests minimal academic achievement gaps as measured by General Certificate of Secondary Education scores (Lynn & Fuerst, 2021; Strand, 2014). Determining if cognitive differences reproduce is important but will require separate analyses.
As discussed above, cognitive ability has been found to correlate strongly with SES. Thus, research should attempt to determine if the cognitive differences present among the current adult generation account for a portion of the social inequalities of concern (Cabinet Office, 2017; McGregor-Smith, 2017; Lammy, 2017). Such a finding could help policymakers more effectively address SES differences across ethnicities.
5 Authors’ Contributions
Fuerst and Te Nijenhuis conceived of the idea. Fuerst performed the computations. Pesta and Shibaev verified the analytical methods. Te Nijenhuis, Pesta, and Shibaev edited and revised the manuscript. All authors discussed the results and contributed to the final manuscript.
6 Conflict of Interest
The authors declare no conflict of interest.
7 Ethical Approval
Approval from an institutional review board was not required because these analyses involved only analyses of de-identified secondary data.
8 References
American Psychological Association (APA) (2007). Task Force on Socioeconomic Status. Report of the APA Task Force on Socioeconomic Status. Washington, DC: Author.
Aoki, Y., & Santiago, L. (2018). Deprivation, Segregation, and Socioeconomic Class of UK Immigrants: Does English Proficiency Matter? IZA DP No. 11368. www.doi.org/10.2139/ssrn.3137499
Ashe, S. D., & Nazroo, J. (2016). Equality, Diversity and Racism in the Workplace: A Qualitative Analysis of the 2015 Race at Work Survey. Commissioned BITC in partnership with CODE, The University of Manchester and YouGov.
Baron, H., Martin, T., Proud, A., Weston, K., & Elshaw, C. (2003). Ethnic group differences and measuring cognitive ability. International Review of Industrial and Organizational Psychology, 18, 191–238. www.doi.org/10.1002/0470013346.ch6
Bebbington, P., Brugha, T., Meltzer, H., Farrell, M., Ceresa, C., Jenkins, R., & Lewis, G. (2000). Psychiatric disorder and dysfunction in the UK National Survey of Psychiatric Morbidity. Social Psychiatry and Psychiatric Epidemiology, 35(5), 191–197. www.doi.org/10.1007/s001270050227
Cabinet Office (2017). Race Disparity Audit: Summary Findings from the Ethnicity Facts and Figures Website. Retrieved from: www.ethnicity-facts-figures.service.gov.uk/static/race-disparity-audit-summary-findings.pdf
Carroll, J. B. (1997). The three-stratum theory of cognitive abilities. In D. P. Flanagan, & P. L. Harrison (Eds.), Contemporary intellectual assessment: Theories, tests and issues. (pp. 122–130). New York: Guilford.
Chartered Institute of Personnel and Development (CIPD) (2017). Resourcing and talent planning survey 2017. Retrieved from www.cipd.com
Department for Work and Pension (2009). A test for racial discrimination in recruitment practice in British cities. Carried out by NatCen. Retrieved from: www.natcen.ac.uk/media/20541/test-for-racial-discrimination.pdf Access: March 2019.
Dickens, W. T., & Flynn, J. R. (2006). Black Americans reduce the racial IQ gap: Evidence from standardization samples. Psychological Science, 17, 913–920. www.doi.org/10.1111/j.1467–9280.2006.01802.
Di Stasio, V., & Heath, A. (2019). Are employers in Britain discriminating against ethnic minorities? Centre for Social Investigation, 1–10.
Dykiert, D., & Deary, I. J. (2013). Retrospective validation of WTAR and NART scores as estimators of prior cognitive ability using the Lothian Birth Cohort 1936. Psychological Assessment, 25(4), 1361. www.doi.org/10.1037/a0033623
Evers, A., Te Nijenhuis, J., & Van der Flier, H. (2005). Ethnic bias and fairness in personnel selection: Evidence and consequences. In A. Evers, N. Anderson, & O. Smit-Voskuijl (Eds.), The Blackwell Handbook of Personnel Selection (pp. 306–328). Oxford, England: Blackwell. www.doi.org/10.1111/b.9781405117029.2005.00018.x
Gray, M., D’Ardenne, J., Balarjan, M., & Uhrig, N. (2011). Cognitive testing of wave 3 Understanding Society questions. Understanding Society working paper. Retrieved from www.understandingsociety.ac.uk/research/publications/workingpaper/understanding-society/2011–03.
Heckman, J. J. (1998). Detecting Discrimination. Journal of Economic Perspectives, 12, 101–116. www.doi.org/10.1257/jep.12.2.101
Ho, A. D., & Reardon, S. F. (2012). Estimating achievement gaps from test scores reported in ordinal “proficiency” categories. Journal of Educational and Behavioral Statistics, 37(4), 489–517. www.doi.org/10.3102/1076998611411918
Hoffmann, N. I. (2018). Cognitive achievement of children of immigrants: Evidence from the Millennium Cohort Study and the 1970 British Cohort Study. British Educational Research Journal, 44(6), 1005–1028. www.doi.org/10.1002/berj.3476
Jivraj, S. & Khan, O. (2015) How likely are ethnic minorities to live in deprived neighbourhoods? In: Jivraj S and Simpson L (eds), Ethnic Identity and Inequalities in Britain: The Dynamics of Diversity. Bristol: Policy Press, pp. 199–214. Retrieved from: www.ethnicity.ac.uk/research/publications/ethnic-identity-and-inequalities-in-britain/; www.doi.org/10.2307/j.ctt1t89504.18
Kapadia, D. Nazroo, J., & Clark, K. (2015). Have ethnic inequalities in the labour market persisted? In: S. Jivraj and L. Simpson (eds), Ethnic Identity and Inequalities in Britain: The Dynamics of Diversity, Policy Press, Bristol. Link to data: www.ethnicity.ac.uk/research/publications/ethnic-identity-and-inequalities-in-britain/
Kezer, M., & Cemalcilar, Z. (2020). A comprehensive investigation of associations of objective and subjective socioeconomic status with perceived health and subjective well-being. International Review of Social Psychology, 33(1). www.doi.org/10.5334/irsp.364
Lammy, D. (2017). The Lammy review: An independent review into the treatment of, and outcomes for, Black, Asian and Minority Ethnic individuals in the Criminal Justice System. HM Government.
Lane, M., & Conlon, G. (2016). The impact of literacy, numeracy and computer skills on earnings and employment outcomes. (OECD Education Working Papers No. 129). Paris, France: OECD Publishing.
Lynn, R. (2008). The Global Bell Curve: Race, IQ, and Inequality Worldwide. Washington Summit Publishers.
Lynn, R., & Cheng, H. (2013). Recent data for majority and racial minority differences in intelligence of 5-year olds in the United Kingdom. Intelligence, 41(5), 452–455. www.doi.org/10.1016/j.intell.2013.06.019
Lynn, R., & Fuerst, J. G. (2021). Recent Studies of Ethnic Differences in the Cognitive Ability of Adolescents in the United Kingdom. Mankind Quarterly, 61(4). www.doi.org/10.46469/mq.2021.61.4.11
Nelson HE, & Wilson J (1991) National Adult Reading Test (NART). NFER-Nelson, Windsor, UK.
McGregor-Smith, R. (2017). Race in the workplace: The McGregor-Smith review. Gov. uk. Retrieved from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/594336/race-inworkplace-mcgregor-smith-review.pdf.
McGurn, B., Starr, J. M., Topfer, J. A., Pattie, A., Whiteman, M. C., Lemmon, H. A., … & Deary, I. J. (2004). Pronunciation of irregular words is preserved in dementia, validating premorbid IQ estimation. Neurology, 62(7), 1184–1186. www.doi.org/10.1212/01.WNL.0000103169.80910.8B
McManus, S., Meltzer, H., Brugha, T., Bebbington, P. E., & Jenkins, R. (2009). Adult psychiatric morbidity in England: results of a household survey. Health and Social Care Information Centre.
Mobius, M. M., & Rosenblat, T. S. (2006). Why beauty matters. American Economic Review, 96(1), 222–235. www.doi.org/10.1257/000282806776157515
Pesta, B. J., McDaniel, M. A., & Bertsch, S. (2010). Toward an index of well-being for the fifty US states. Intelligence, 38(1), 160–168. www.doi.org/10.1016/j.intell.2009.09.006
Office for National Statistics (ONS) (2015). 2011 Census analysis: Ethnicity and religion of the non-UK born population in England and Wales: 2011: Appendix: The number and percentage of the non-UK born population by world region and the top ten countries within each world region. Retrieved from: www.ons.gov.uk/peoplepopulationandcommunity/culturalidentity/ethnicity/articles/2011censusanalysisethnicityandreligionofthenonukbornpopulationinenglandandwales/2015-06-18
Organisation for Economic Co-operation and Development (OECD) (2012). PISA 2009 Technical Report. PISA, OECD Publishing, Paris.
Organisation for Economic Co-operation and Development (OECD) (2013). Technical Report of the Survey of Adult Skills (PIAAC). OECD Publishing, Paris.
Owen, D. (1995). Ethnic Minorities in Great Britain: Patterns of Population Change, 1981–91 1991: 1991 Census Statistical Paper No. 10.
Plomin, R., & Spinath, F. M. (2004). Intelligence: genetics, genes, and genomics. Journal of Personality and Social Psychology, 86(1), 112. www.doi.org/10.1037/0022-3514.86.1.112
Race Disparity Unit (2020). Research and analysis: Ethnicity data: how similar or different are aggregated ethnic groups? Retrieved from: www.gov.uk/government/publications/ethnicity-data-how-similar-or-different-are-aggregated-ethnic-groups/ethnicity-data-how-similar-or-different-are-aggregated-ethnic-groups
Reardon, S. F., & Ho, A. D. (2015). Practical issues in estimating achievement gaps from coarsened data. Journal of Educational and Behavioral Statistics, 40(2), 158–189. www.doi.org/10.3102/1076998615570944
Rees, P., Wohland, P., Norman, P., Lomax, N., & Clark, S. (2016). Population projections by ethnicity: Challenges and a solution for the United Kingdom. In D. Swanson (Ed.), The frontiers of applied demography (pp. 383–408). Amsterdam, The Netherlands: Springer. www.doi.org/10.1007/978-3-319-43329-5_18
Roth, P. L., Bevier, C. A., Bobko, P., Switzer, & Tyler, P. (2001). Ethnic group differences in cognitive ability in employment and educational settings: A meta‐analysis. Personnel Psychology, 54(2), 297–330. www.doi.org/10.1111/j.1744-6570.2001.tb00094.x
Roth, P. L., Huffcutt, A. I., & Bobko, P. (2003). Ethnic group differences in measures of job performance: A new meta-analysis. Journal of Applied Psychology, 88(4), 694. www.doi.org/10.1037/0021-9010.88.4.694
Roth, P. L., Van Iddekinge, C. H., DeOrtentiis, P. S., Hackney, K. J., Zhang, L., & Buster, M. A. (2017). Hispanic and Asian performance on selection procedures: A narrative and meta-analytic review of 12 common predictors. Journal of Applied Psychology, 102(8), 1178–1202. www.doi.org/10.1037/apl0000195
Rindermann, H. (2018). Cognitive Capitalism: Human Capital and the Wellbeing of Nations. Cambridge University Press. www.doi.org/10.1017/9781107279339
Shen, W., Sackett, P. R., Lievens, F., Schollaert, E., Van Hoye, G., Steiner, D. D. … Cook, M. (2017). Updated perspectives on the international legal environment for selection. In Farr, J. L. and Tippins, N. T. (Eds.), Handbook of Employee Selection (2 ed., pp. 659–677). New York, NY: Routledge. www.doi.org/10.4324/9781315690193–29
Skopek, J., & Passaretta, G. (2018). The Evolution of Social and Ethnic Inequalities in Cognitive Achievement from Preschool to Secondary Schooling. In: Passaretta, Giampiero
Skopek, Jan (Ed.): Roots and Development of Achievement Gaps. A Longitudinal Assessment in Selected European Countries, pp. 95–144, ISOTIS Report for European Commission, Trinity College Dublin, 2018.
Strand, S. (2014). Ethnicity, gender, social class and achievement gaps at age 16: Intersectionality and ‘Getting it’ for the white working class. Research Papers in Education, 29(2), 131–171. www.doi.org/10.1080/02671522.2013.767370
Taylor, M., & Hegerty, S. (1985). The Best of Both Worlds? A Review of Research Into the Education of Pupils of South Asian Origin. Delmar Pub.
Tomlinson, S. (1980). The educational performance of ethnic minority children. New Community, 8(3), 213–234. www.doi.org/10.1080/1369183X.1980.9975630
Tomlinson, S. (1983). The educational needs of ethnic minority children. Journal of Biosocial Science, 15(S8), 157–166. www.doi.org/10.1017/S0021932000024962
UK Government (2019). Ethnicity Facts and Figures: List of ethnic groups. Retrieved from: www.ethnicity-facts-figures.service.gov.uk/ethnic-groups
UK Government (2019b). Ethnicity Facts and Figures: Education, skills and training: Students aged 16 to 18 achieving at least 3 A grades at A level. Retrieved from: www.ethnicity-facts-figures.service.gov.uk/education-skills-and-training/a-levels-apprenticeships-further-education/students-aged-16-to-18-achieving-3-a-grades-or-better-at-a-level/latest
UK Government (2020). Ethnicity facts and figures: Crime, justice and the law: Arrests. Accessed at: www.ethnicity-facts-figures.service.gov.uk/crime-justice-and-the-law/policing/number-of-arrests/latest
The World Bank (2020). Human Capital Project. Retrieved from: www.worldbank.org/en/publication/human-capital#Index; www.doi.org/10.1596/33671
Zilanawala, A., Kelly, Y., & Sacker, A. (2016). Ethnic differences in longitudinal latent verbal profiles in the millennium cohort study. The European Journal of Public Health, 26(6), 1011–1016. www.doi.org/10.1093/eurpub/ckw184.
9 Appendix
9.1 Detailed Results for Specific Cognitive Samples
9.1.1 APM (2000) and APM (2007)
For comparisons across waves, we derived the category “White” by weighting the White British and White non-British means by the respective Weighted Ns. For both the 2000 and 2007 waves, the effects of age and sex (dummy coded with female = 1, male = 0) were regressed out via OLS regression. Scores were then weighted to be nationally representative. The actual N and weighted N are both reported. The results for both AMP (2000) and AMP (2007) appear in Table A1. Consistent with previous reports, the N-weighted average across the two survey waves shows that Jews score the highest (105.11), followed by combined East and other Asians (100.61), Whites (100), South Asians (95.59), and Blacks (92.74). Interpretation of these results is limited by the relatively small samples sizes of the non-White groups, and the incomplete validity of NART as an index of mental ability. However, the general trend is concordant with that found for other assessments discussed below.
Mean scores by ethnicity from the AMP (2000; 2007) surveys
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
9.1.2 PIAAC (2012) and Skills for Life (2003)
The results for PIAAC (2012) appear in Table A2. Among South Asians and Blacks, a 5 to 10 point difference exists between UK and foreign-born individuals. This is consistent with the international results discussed by Batalova and Fix (2016), who found a generational convergence for migrants in Canada, France, Germany, the United Kingdom, and the United States. This likely reflects the effects of enculturation, particularly language acquisition and improved living conditions. Depending on the group, it may also reflect the effect of secular changes in migrant composition and selectivity.
Mean scores by ethnicity from the PIAAC (2012) survey
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
National origin was reported for foreign-born individuals, as shown in Table A3. For several countries (e.g., South Africa) many of the immigrants were not from the country’s majority ethnic/racial group. However, owing to the small sample sizes, we did not try to decompose results further by ethnicity. For the African countries (specifically, Kenya, Nigeria, and South Africa) the scores were notably higher than the national means reported by Lynn and Becker (2019). These relatively high scores are somewhat surprising, as one would expect that the tests would be more biased against foreign-born migrants. The relatively good performance here could be due to a national origin by ethnicity compositional effect, for countries such as South Africa, or, alternatively, to a migrant selection effect, in the case of more ethnically homogenous countries, such as Nigeria (Model, 2008; Easterly & Nyarko, 2008).
Mean scores by nationality from the PIAAC (2012) survey for foreign born individuals
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
Results for SfL (2003) appear in Table A4. These generally concord with the PIAAC (2012) ones. Chinese and Indians are exceptions in that they do substantially better in the SfL (2003) survey than in PIAAC (2012). This difference, however, may result from the transformation used to convert the discretized values back to continuous ones. Note, descriptively similar results have been found for the 2011 SfL survey (Department for Business Innovation & Skills, 2012). However, this later survey is not included in the review owing to the overlap with the PIAAC (2012) survey.
Mean scores by ethnicity from the SfL (2003) survey
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
9.1.3 UKHLS (2012)
Results for UKHLS appear in Table A5. Note, as done previously, the total White group, not the British White group, was set as the reference group. This is to allow summary across analyses, since for some surveys (e.g., AMP, 2000), British and non-British Whites were not disaggregated. On this measure, the difference between UK and foreign-born individuals is smaller at around five to ten points. It is notable that Chinese participants did better here than they did on the previously discussed, verbally loaded tests.
Mean scores by ethnicity from the UKHLS (2011–13) survey by region of birth
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
The dataset also included national origins for foreign-born participants. However, for several countries (e.g., South Africa), many emigrants were not from their country’s majority ethnic group. Owing to the larger sample sizes here than with the PIAAC (2012) survey, we were able to decompose results by self-reported ethnicity. We did this for National x Ethnic Groups with N > 6. Following the UKHLS classifications, we grouped “White British” and “White Other” separately. For the purposes of these classifications, “White Other” includes “Irish” and “Any Other White Background.” Typically, this refers to members of the dominant ethnic group (e.g., ethnic Germans in Germany). “South Asian” refers to people who identify as Bangladeshi, Pakistani, Indian, and Sri Lankan. For “Black,” Black Africans and Black Caribbean were grouped together regardless of whether the country was African or Caribbean. Results appear in Table A6. Note, owing to the legacy of the British Empire, the number of British Whites born outside of the UK was non-trivial.
Mean scores by ethnicity from the UKHLS (2011–13) survey by nation of birth and self-reported ethnicity for foreign born individuals
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
A point of interest is that the scores of foreign-born Black African and Caribbean immigrants were substantially higher, with a median advantage of 17 points, than one would expect based on Lynn and Becker’s (2019) quality N-weighted (QNW) National IQs (Ghana: 89 vs. 62; Jamaica: 84 vs. 75; Kenya: 92 vs. 75; Nigeria: 88 vs. 68; South Africa: 96 vs. 80; Uganda: 86 vs. 76). It is not clear why this is the case, since one would expect tests taken in a foreign culture (UK) to be more biased than ones taken in the home country. Migrant selection is a possible explanation, since it has been reported that both Black African and Caribbean emigrants are highly selected in educational attainment (Model, 2008; Easterly & Nyarko, 2008).
A final point of interest is that ethnic group differences show up between foreign born individuals from certain countries (Ghana, India, Kenya, Nigeria, and South Africa) but not others (Hong Kong, Jamaica, and Uganda). The reason for this is not clear; however, the sample sizes are quite small, so not much should be made of these results.
It was also possible to analyze the data by age group. We report the scores for South Asians and Blacks for five age groups (16–24, 25–34, 35–44, 45–54, and 55–65). For the other ethnic groups, sample sizes were too small for reliable analyses. We calculated the scores using the pooled standard deviations for Whites, South Asians, and Blacks of all ages and birth regions. We also report results for UK born individuals too, because of a possible Age x Generation interaction. Results appear in Table A7. For both South Asians and Blacks, there was a cohort effect, such that the differences were smaller in younger age groups (though this effect was less clear when limiting analysis to UK born individuals).
Asian-White and Black-White standard differences (d) by five age groups and region of birth.
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
9.1.4 MCS (2015)
The MCS results appear in Table A8. As seen, for most broad ethnic groups, there is approximately a ten-point difference between UK and foreign-born individuals. This is most likely because the test featured vocabulary items which could exhibit substantial linguistic bias against non-native English speakers. That said, the issue of psychometric bias is best evaluated using multigroup confirmatory factor analysis, the performance of which is outside the scope of the current article. Despite this, all second-generation non-White groups (except for the mixed one) perform substantially worse than the White ethnic group.
Mean scores by ethnicity from the MCS (2015) survey by region of birth
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
To control for possible linguistic bias, an alternative approach is to restrict scores to individuals who report speaking only English at home. These results, again split by region of birth, appear in Table A9. Except for Indians (and South Asians in general), these results were substantially the same as those in Table A8 above. Regarding Indians, it is not clear if the relative advantage is due to reduced linguistic bias or if, instead, there was selection for a cognitively advantaged subgroup. A compositional effect is nonetheless suggested, given that foreign-born Indians who were reared to speak English scored about ten points above all foreign-born Indians.
Mean scores by ethnicity from the MCS (2015) survey by region of birth for those who only speak English at home
Citation: Comparative Sociology 22, 6 (2023) ; 10.1163/15691330-bja10094
9.2 Data Availability
All of the data used for the computation of results are available at, or through, the following sites/linked resources:
9.2.1 Cognitive Data
The Adult Psychiatric Morbidity Survey 2000: Office for National Statistics. (2003). Psychiatric Morbidity among Adults Living in Private Households, 2000. [data collection]. UK Data Service. SN: 4653, DOI: 10.5255/UKDA-SN-4653–1 Accessed at: https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=4653
The Adult Psychiatric Morbidity Survey 2007: National Centre for Social Research, University of Leicester. (2017). Adult Psychiatric Morbidity Survey, 2007. [data collection]. 4th Edition. UK Data Service. SN: 6379, DOI: 10.5255/UKDA-SN-6379–2 Accessed at: https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=6379
Skills for Life, 2003: BMRB International. (2013). Skills for Life Survey, 2003. [data collection]. UK Data Service. SN: 7239, DOI: 10.5255/UKDA-SN-7239-1 Accessed at: https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=7239
Skills for Life, 2011: TNS BMRB. (2013). Skills for Life Survey, 2011. [data collection]. UK Data Service. SN: 7240, DOI: 10.5255/UKDA-SN-7240-1. Accessed at: https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=7240
UK Household Longitudinal Study, 2009: University of Essex, Institute for Social and Economic Research. (2022). Understanding Society: Waves 1–12, 2009–2021 and Harmonised BHPS: Waves 1–18, 1991–2009: Secure Access. [data collection]. 15th Edition. UK Data Service. SN: 6676, DOI: 10.5255/UKDA-SN-6676-15. Accessed at: https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=6676
The Millennium Cohort Study, 2015: University of London, Institute of Education, Centre for Longitudinal Studies. (2022). Millennium Cohort Study: Age 14, Sweep 6, 2015. [data collection]. 7th Edition. UK Data Service. SN: 8156, DOI: 10.5255 /UKDA-SN-8156-7. Accessed at: https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=8156
PIAAC, 2012: Department of Education. (2019). 2012 International Survey of Adult Skills Survey dataset for the UK: UK survey SPSS file. Via FOIA request at: www .whatdotheyknow.com/request/adult_skills_international_surve_2?nocache =incoming-1343050#incoming-1343050
9.2.2 Socioeconomic Status Data
Arrest rates: UK Government (2020). Ethnicity facts and figures: Crime, justice and the law: Arrests. Accessed at: www.ethnicity-facts-figures.service.gov.uk/crime-justice -and-the-law/policing/number-of-arrests/latest
Deprived neighborhood status: Jivraj, S. & Khan, O. (2015). How likely are ethnic minorities to live in deprived neighbourhoods? In: Jivraj S and Simpson L (eds), Ethnic Identity and Inequalities in Britain: The Dynamics of Diversity. Bristol: Policy Press, pp. 199–214. Retrieved from: www.ethnicity.ac.uk/research/publications /ethnic-identity-and-inequalities-in-britain; www.doi.org/10.2307/j.ctt1t89504.18 (linked data)
Economically inactivity: Kapadia, D. Nazroo, J., & Clark, K. (2015). Have ethnic inequalities in the labour market persisted? In: S. Jivraj and L. Simpson (eds), Ethnic Identity and Inequalities in Britain: The Dynamics of Diversity, Policy Press, Bristol. Link to data: www.ethnicity.ac.uk/research/publications/ethnic-identity-and-inequalities -in-britain (linked data)
Net income: University of Essex, Institute for Social and Economic Research. (2022). Understanding Society: Waves 1–12, 2009–2021 and Harmonised BHPS: Waves 1–18, 1991–2009: Secure Access. [data collection]. 15th Edition. UK Data Service. SN: 6676, DOI: 10.5255/UKDA-SN-6676–15 Accessed at: www.beta.ukdataservice.ac.uk/data catalogue/studies/study?id=6676
A-levels: UK Government (2019b). Ethnicity Facts and Figures: Education, skills and training: Students aged 16 to 18 achieving at least 3 A grades at A level. Retrieved from: www.ethnicity-facts-figures.service.gov.uk/education-skills-and-training/a -levels-apprenticeships-further-education/students-aged-16-to-18-achieving-3-a -grades-or-better-at-a-level/latest
9.2.3 International Test Score Data
The World Bank (2020). Human Capital Project. Retrieved from: www.worldbank.org/en/publication/human-capital#Index.
9.3 References
Batalova, J. and Fix, M. (2016). Literacy and Numeracy Skills of Second-Generation Young Adults: A Comparative Study of Canada, France, Germany, the United Kingdom, and the United States. Retrieved 19 Nov 2019, from https://static1.squarespace.com/static/51bb74b8e4b0139570ddf020/t/578d1e95579fb360acc952bf/1468866198724/Batalova_Fix_PIAAC.pdf.%20Washington%20DC. Washington DC.
Department for Business Innovation & Skills (2012). The 2011 Skills for Life Survey: A Survey of Literacy, Numeracy and ICT Levels in England. Bis Research Paper No. 81.
Easterly, W., & Nyarko, Y. (2008). Is the brain drain good for Africa? Brookings Global Economy and Development Working Paper, (19).
Lynn, R., & Becker, D. (2019). The intelligence of Nations. London: Ulster Institute for Social Research.
Model, S. (2008). The Secret of West Indian Success. Society, 45(6), 544–548.