Content uploaded by Emil O. W. Kirkegaard
Author content
All content in this area was uploaded by Emil O. W. Kirkegaard on Dec 09, 2017
Content may be subject to copyright.
MANKIND QUARTERLY 2017 58:2 312-323
312
Employment Rates for 11 Country of Origin Groups in
Scandinavia
Emil O. W. Kirkegaard*
Ulster Institute for Social Research, London, UK
*Email: emil@emilkirkegaard.dk
Employment rates for 11 country of origin groups living in the
three Scandinavian countries are presented. Analysis of variance
showed that differences in employment rates are highly predictable
(adjusted multiple R = .93). This predictability was mostly due to
origin countries (eta = .89), not sex (eta = .25) and host country
(eta = .20). Furthermore, national IQs of the origin countries
predicted employment rates well across all host countries (r’s = 0.74
[95%CI: 0.30, 0.92], 0.75 [0.30, 0.92], 0.66 [0.14, 0.89] for Denmark,
Norway and Sweden, respectively), and so did Muslim % of the
origin countries (r’s = -0.80 [-0.94, -0.43], -0.78 [-0.94, -0.37], -0.58 [-
0.87, -0.01]).
Key Words: Immigration, Employment, Scandinavia, Denmark,
Norway, Sweden, Country of origin, National IQ, Islam, Muslim
Numerous studies have reported very large differences in social status and
related outcomes for different country of origin groups in Western countries
(Fuerst, 2012; Fuerst & Kirkegaard, 2014; Jones & Schneider, 2010; Kirkegaard,
2014a, 2015; Kirkegaard & Becker, 2017; Kirkegaard & Fuerst, 2014). Detailed
data has previously been reported for Denmark and Norway (Kirkegaard, 2014a;
Kirkegaard & Fuerst, 2014). However, no data has yet been published (in English)
for Sweden to the author’s knowledge. Recently, a new edition of the annual
publication Indvandrere i Danmark (Immigrants in Denmark) (Danmarks Statistik,
2016) was published. Each edition of the series has a particular focus, and the
focus of the 2016 report is comparative analysis of immigrant outcomes in the
Scandinavian countries. The purpose of this paper is to present the main data in
English as well as some analyses of them.
KIRKEGAARD, E.O.W. EMPLOYMENT RATES IN SCANDINAVIA
313
Data and analyses
Danmarks Statistik (the Danish statistics bureau) collaborated with their
Swedish and Norwegian analogues (Statistiska centralbyrån and Statistisk
sentralbyrå), to compile matching data for employment rates for 2014. The data
concern only foreign born (1st generation) immigrants aged 20-64, and cover a
diverse set of 11 countries which they deemed to have sufficient numbers for
comparison purposes.1 The data only cover persons with residence and thus
work permits, not illegal immigrants. The employment rates are shown in Figure
1.2
Figure 1. Employment rates by country of origin and host country. The thick
blocks for the Scandinavian countries are for natives in their own countries.
Employment rates were not available for Scandinavians living in other
Scandinavian countries.
1 I asked DST if they could provide results for additional countries, which they
declined.
2 The numbers were given by sex, but they have been averaged here for
simplicity of presentation. The full data are given in the supplementary
materials.
MANKIND QUARTERLY 2017 58:2
314
The employment rates are similar across host countries with a mean
correlation of .86. The rate for Somalis in Denmark is lower than those reported
in Kirkegaard and Fuerst (2014). This probably has to do with the fact that the
present numbers only concern foreign born persons, while the previously reported
numbers concerned all persons from Somalia, including later generations who
have higher employment rates.
Analysis 1 ― Analysis of variance
Two regressions were run to see how host, origin and sex predicted the
employment rate. The first model (n = 36) did not include sex and had an adjusted
(multiple) R of .93 (R2 = 87%).3 Analysis of variance showed that origin was the
most important predictor with an eta of .88 while host had an eta of .05. In the
second model (n = 72), sex-disaggregated data were used and sex was added
as a predictor. This model had an adjusted R of .93 (R2 = 86%). Analysis of
variance etas were: origin .89, host .20 and sex .25. Thus, origin was still by far
the best predictor, and the more complex model with disaggregated data was not
superior to the simpler model. However, because sex itself was non-redundant
and because it was based on more data points, the betas for the complex model
are reported in Table 1. The output from the first model can be found in the
supplementary materials.
The beta for each origin country can be regarded as an estimate of that
country’s human capital taken as composite variable consisting of cognitive,
personality, interest, temperament, knowledge and skills (Jones & Potrafke,
2014). Such estimates will only be precise under strict assumptions, which are
that migrants are equally representative of their country of origin with respect to
human capital (no differences in migrant selectivity), and lack of host x origin
interactions (equal opportunity for each immigrant group).
The host country predictor can be interpreted in multiple ways. One might
regard it as a measure of integrational success. This could be, but then it should
not affect the natives. Figure 1 shows that native Danes have lower employment
rates than native Swedes and Norwegians. The employment rate of Danes in
Denmark cannot easily be ascribed to integration efforts. For that reason, the host
effects are more parsimoniously explained by system-wide effects such as slightly
different ways of counting employment across countries or generosity of
unemployment benefits.
3 The design was nearly but not fully crossed because there was no cross-
Scandinavian data (e.g. no data for Norwegians in Denmark).
KIRKEGAARD, E.O.W. EMPLOYMENT RATES IN SCANDINAVIA
315
Table 1. Regression standardized betas. Outcome: employment rate. N = 72.
Predictor
Beta
CI lower
CI upper
Origin: Afghanistan
0.00
Origin: Bosnia & Herzogovina
1.14
0.71
1.57
Origin: China
0.73
0.29
1.16
Origin: Denmark
1.99
1.37
2.62
Origin: Germany
1.40
0.97
1.83
Origin: Iran
0.69
0.26
1.12
Origin: Iraq
-0.05
-0.49
0.38
Origin: Norway
1.76
1.14
2.38
Origin: Poland
1.48
1.05
1.91
Origin: Somalia
-0.74
-1.17
-0.31
Origin: Sweden
2.10
1.47
2.72
Origin: Syria
-1.11
-1.54
-0.68
Origin: Thailand
1.12
0.68
1.55
Origin: Turkey
0.63
0.20
1.07
Host: Denmark
0.00
Host: Norway
0.50
0.28
0.73
Host: Sweden
0.25
0.03
0.48
Sex: men
0.00
Sex: women
-0.50
-0.68
-0.32
An alternative analytic approach here is to recode employment rates as
fractions of the native employment rate in that country. This is based on the
assumption that human capital of the natives is equal, which seems a fair
assumption for the Scandinavian countries. Table 2 shows the modeling results
from the recoded data.
The model fit using relative rates had similar fit to the original data: adjusted
R = .92 vs. original coding .93. Using relative rates coding, Norway still did seem
to have better immigrant outcomes, which could be interpreted as better
integrational success or stronger immigrant selection across all origin countries.
Still, the effect size was quite small, etas: origin = .91, host = .14, and sex = .20.
MANKIND QUARTERLY 2017 58:2
316
Table 2. Regression standardized betas with relative rate of natives coding.
Outcome: employment rate. N = 72.
Predictor Beta SE CI lower CI upper
Origin: Afghanistan 0
Origin: Bosnia & Herzogovina 1.16 0.22 0.71 1.61
Origin: China 0.75 0.22 0.30 1.20
Origin: Denmark 2.11 0.33 1.46 2.76
Origin: Germany 1.43 0.22 0.98 1.88
Origin: Iran 0.70 0.22 0.25 1.15
Origin: Iraq -0.06 0.22 -0.51 0.39
Origin: Norway 1.78 0.33 1.13 2.43
Origin: Poland 1.51 0.22 1.06 1.96
Origin: Somalia -0.76 0.22 -1.21 -0.30
Origin: Sweden 2.08 0.33 1.43 2.73
Origin: Syria -1.13 0.22 -1.58 -0.68
Origin: Thailand 1.15 0.22 0.70 1.61
Origin: Turkey 064 0.22 0.19 1.09
Host: Denmark 0
Host: Norway 0.33 0.12 0.10 0.57
Host: Sweden 0.03 0.12 -0.20 0.27
Sex: men 0
Sex: women -0.39 0.09 0.58 -0.21
Analysis 2 ― Analysis of variance with PISA data
Dronkers et al. (2014) also analyzed origin effects on immigrant outcomes.
They did not use analysis of variance so their results are not directly comparable
to those reported in this study. However, they give the PISA 2006 sciences scores
in a table for each host x origin cell with available data (immigrants only). To
expand the dataset and make it more comparable to the present study’s, the
scores for the native populations were added. Unfortunately, the PISA 2006
reports do not contain scores for natives and immigrants separately as the later
KIRKEGAARD, E.O.W. EMPLOYMENT RATES IN SCANDINAVIA
317
reports do, so the total country scores were used as a proxy. For European
countries, this results in a slight underestimate of the scores in most cases
because immigrants on average perform worse than natives in most European
countries (Rindermann & Thompson, 2016). These data were then analyzed
using the same method as before, with the PISA scores as the outcome and the
origin country as the predictor (n = 147, 72 origin countries, 58 host countries).
The adjusted R was .89 (R2 = 78%). Analysis of variance showed that the most
important predictor was origin country, eta = .64, while host country also had
sizable validity, eta = .33. Interpretation of these results is complicated by the fact
that there were many missing cells: 4,176 host x origin combinations, but only
147 were present, or 3.5%. These cells are unlikely to be randomly missing and
will thus likely cause misestimation of betas, especially because immigrant
selection effects will be partially included in the host term. In contrast, for the
employment data, there were 42 host x origin combinations of which 36, or 86%,
were present. The results nevertheless illustrate that country of origin is more
important than host country for two measures of human capital: employment
rates, and PISA scores.
Analysis 3 ― Country of origin characteristics regression
As in previous studies (cited in the introduction), the outcome (employment
rates) was correlated with national IQ (Lynn & Vanhanen, 2012) and Muslim%
(Pew Research Center, 2011) of the origin countries. National IQs reflect the
populations’ average levels of cognitive ability. Cognitive ability has been found
to positively predict almost all socially valued outcomes, including employment
rates, at the individual-differences level (Fergusson, John Horwood & Ridder,
2005; Gottfredson, 1997; Herrnstein & Murray, 1994; Lynn, Hampson & Magee,
1984). This is also true at the aggregate level, in that groups with higher cognitive
ability have better outcomes (Herrnstein & Murray, 1994; Kirkegaard & Fuerst,
2016), and countries or areas within a country with lower average cognitive ability
have worse outcomes (Fuerst & Kirkegaard, 2016; Jones, 2016; Jones &
Potrafke, 2014; Kirkegaard, 2014b; Kirkegaard & Fuerst, 2017; Lynn, Cheng &
Grigoriev, 2017; Lynn & Vanhanen, 2012). Less work has been done on Islamic
beliefs and Muslims at the individual level (but see Dronkers, van der Velden &
Dunne, 2011), however, a number of previous studies showed that this is a strong
negative predictor of immigrant group-level outcomes (see for review Kirkegaard,
2017), and probably has incremental validity over national IQ (Kirkegaard &
Fuerst, 2014), though this has not yet been thoroughly examined.
MANKIND QUARTERLY 2017 58:2
318
Figure 2. Scatterplot of national IQs in the home country and employment rates
in Scandinavian countries.
Figure 3. Scatterplot of Muslim% in the home country and employment rates in
Scandinavian countries.
KIRKEGAARD, E.O.W. EMPLOYMENT RATES IN SCANDINAVIA
319
Figures 2 and 3 show the correlations across all three countries and for both
predictors. Both plots show strong relationships between the country-of-origin
predictors and the outcome that are similar across host countries. Table 3 gives
the numerical relationships. Correlations were similar in size, but somewhat
weaker for Sweden and for women. Not much can be made of this because of
the small sample sizes.
Table 3. Correlations between predictors and employment rates by sex.
Numbers in brackets are 95% analytic confidence intervals.
Country
Employment rate
IQ
Muslim%
Denmark
Men
0.71 [0.24 – 0.91]
-0.73 [-0.92 – -0.28]
Women
0.77 [0.35 – 0.93]
-0.86 [-0.96 – -0.55]
Total
0.74 [0.30 – 0.92]
-0.80 [-0.94 – -0.43]
Norway
Men
0.67 [0.15 – 0.90]
-0.69 [-0.90 – -0.18]
Women
0.82 [0.46 – 0.95]
-0.85 [-0.96 – -0.53]
Total
0.75 [0.30 – 0.92]
-0.78 [-0.94 – -0.37]
Sweden
Men
0.59 [0.03 – 0.87]
-0.50 [-0.83 – 0.10]
Women
0.72 [0.24 – 0.91]
-0.67 [-0.90 – -0.15]
Total
0.66 [0.14 – 0.89]
-0.58 [-0.87 – -0.01]
Finally, the country of origin human capital estimates from the analysis of
variance (Analysis 1 and 2) were correlated with the predictors. For the
employment data, national IQ and Muslim% correlations were .78 [.43 to .93, n =
14]4 and -.81 [-.94 to -.48, n = 14], and for the PISA data, they were .62 [.45 to
.74, n = 71] and -.35 [-.59 to -.05, n = 71]. The PISA results were weaker. This is
probably in part because of oddities in the data presumably related to sampling
error. For instance, Denmark’s beta was estimated at .57, which is lower than
expected. If one traces the origin of the result, it is found to be based on only two
datapoints. Danes in Denmark who obtained a score of 494 and Danes in Norway
who obtained the implausibly low PISA score of 411 (compare: Romania 418,
Mexico 410, Indonesia 393). It seems likely that if a comprehensive PISA dataset
were constructed and analyzed in the same way, many of these irregularities
would vanish (for a comprehensive, though different, analysis, see Rindermann
& Thompson, 2016).
4 The increase to n = 14 from n = 11 is due to the addition of the values from the
three Scandinavian countries.
MANKIND QUARTERLY 2017 58:2
320
Analysis 4 – Sex interaction
During review, it was noted that the sex difference in employment rates is
higher in lower IQ and higher Muslim% countries, while the effect of sex was
modeled only as a constant (main) effect. To investigate this more formally, the
data were recoded to be relative rates of men’s employment rate from that
country.5 After this, the data were analyzed as before using analysis of variance
except with the addition of a sex x origin term. This model had an adjusted R of
.94 vs. .92 for the same model without the interaction term but with relative rates.
Thus, the effect size was quite small, even if it was perhaps not a fluke (p = .038).6
Discussion and conclusion
The present study found that a socially important outcome, employment
rates, is predicted quite well by average IQ and Muslim% in the country of origin.
This finding is congruent with nearly all previous studies on immigrant groups
(Jones & Schneider, 2010; Kirkegaard, 2014a, 2015; Kirkegaard & Becker, 2017;
Kirkegaard & Fuerst, 2014). In terms of comparing host and origin country
predictive validity, origin was found to dominate effects from the host countries
quite strongly. This conclusion is similar to those reached by other researchers
who have examined e.g. PISA data (Dronkers et al., 2014; Rindermann &
Thompson, 2016). When data from PISA 2006 were examined using the same
methods as those used elsewhere in this study, they were fairly similar but with
somewhat stronger host country effects. Interpretation of this is not easy due to
very large amounts of probably non-randomly missing data in the PISA dataset
and well as substantial sampling error.
In general, the relative importance of the predictors in analysis of variance
as quantified by etas should be interpreted with caution. Etas are non-negative
metrics7 and thus sampling error necessitates validities above zero for any
5 For example, if the male employment rate for a given origin country was 70%
and women’s was 60%, the relative rate would be .86.
6 Using the original data instead of the relativized data gave near identical results:
adjusted R .94 vs. .93, for model with and without interaction term, respectively
(p = .041).
7 Eta is based on eta2, which are variance-type metrics (0-1 range). Because
variances are based on squared deviations from the mean, they cannot be
negative, and the derived eta values likewise cannot. I report eta instead of the
KIRKEGAARD, E.O.W. EMPLOYMENT RATES IN SCANDINAVIA
321
predictor with zero true validity. This is especially important for small sample
models such as those reported in this study.
Supplementary material
Supplementary materials including code, high quality figures and data can
be found at https://osf.io/4c62n/files/.
References
Danmarks Statistik (2016). Indvandrere i Danmark 2016. Retrieved from
http://www.dst.dk/da/Statistik/Publikationer/VisPub?cid=20704
Dronkers, J., van der Velden, R. & Dunne, A. (2011). The effects of educational systems,
school-composition, track-level, parental background and immigrants’ origin on the
achievement of 15-years-old native and immigrant students: A reanalysis of PISA 2006.
ROA Research Memorandum.
Dronkers, J., Levels, M. & de Heus, M. (2014). Migrant pupils’ scientific performance: The
influence of educational system features of origin and destination countries. Large-Scale
Assessments in Education 2(1): 3. https://doi.org/10.1186/2196-0739-2-3
Fergusson, D.M., John Horwood, L. & Ridder, E.M. (2005). Show me the child at seven
II: Childhood intelligence and later outcomes in adolescence and young adulthood.
Journal of Child Psychology and Psychiatry 46(8): 850-858.
https://doi.org/10.1111/j.1469-7610.2005.01472.x
Fuerst, J. (2012). L&V’s (2012) National IQs predict 2011-2012 GRE scores for 114
citizenship groups, 2010 + 2012 TOEFL scores for 157 citizenship groups, PISA scores
of migrants from 62 nations of origin across 17 destination nations, .... Retrieved January
3, 2016, from http://humanvarieties.org/2014/02/11/lvs-2012-national-iqs-predict-2011-
2012-gre-scores-for-114-citizenship-groups-2010-2012-toefl-scores-for-157-citizenship-
groups-pisa-scores-of-migrants-from-62-nations-of-origin-across-1/
Fuerst, J. & Kirkegaard, E.O.W. (2014). Do national IQs predict U.S. immigrant cognitive
ability and outcomes? An analysis of the National Longitudinal Survey of Freshmen. Open
Differential Psychology. Retrieved from https://openpsych.net/paper/20
Fuerst, J. & Kirkegaard, E.O.W. (2016). Admixture in the Americas: Regional and national
differences. Mankind Quarterly 56: 256-374.
more common eta² for the reasons discussed by, among others, Hunter and
Schmidt (2004, p. 189ff).
MANKIND QUARTERLY 2017 58:2
322
Gottfredson, L.S. (1997). Why g matters: The complexity of everyday life. Intelligence 24:
79-132. https://doi.org/10.1016/S0160-2896(97)90014-3
Herrnstein, R.J. & Murray, C. (1994). The Bell Curve: Intelligence and Class Structure in
American Life. New York: Free Press.
Hunter, J.E. & Schmidt, F.L. (2004). Methods of Meta-Analysis. Correcting Error and Bias
in Research Findings. Thousand Oaks, CA: Sage. Retrieved from
http://site.ebrary.com/id/10387875
Jones, G. (2016). Hive Mind: How Your Nation’s IQ Matters so Much More than Your
Own. Stanford, California: Stanford University Press.
Jones, G. & Potrafke, N. (2014). Human capital and national institutional quality: Are
TIMSS, PISA, and national average IQ robust predictors? Intelligence 46: 148-155.
https://doi.org/10.1016/j.intell.2014.05.011
Jones, G. & Schneider, W.J. (2010). IQ in the production function: Evidence from
immigrant earnings. Economic Inquiry 48: 743-755. https://doi.org/10.1111/j.1465-
7295.2008.00206.x
Kirkegaard, E.O.W. (2014a). Crime, income, educational attainment and employment
among immigrant groups in Norway and Finland. Open Differential Psychology. Retrieved
from http://openpsych.net/ODP/2014/10/crime-income-educational-attainment-and-
employment-among-immigrant-groups-in-norway-and-finland/
Kirkegaard, E.O.W. (2014b). The international general socioeconomic factor: Factor
analyzing international rankings. Open Differential Psychology. Retrieved from
http://openpsych.net/ODP/2014/09/the-international-general-socioeconomic-factor-
factor-analyzing-international-rankings/
Kirkegaard, E.O.W. (2015). Crime among Dutch immigrant groups is predictable from
country-level variables. Open Differential Psychology.
Kirkegaard, E.O.W. (2017, May). Differential immigrant group performance: A matter of
intelligence? Presented at the London Conference on Intelligence 2017. Retrieved from
https://www.youtube.com/watch?v=jMSRZQ1usDc
Kirkegaard, E.O.W. & Becker, D. (2017). Immigrant crime in Germany 2012-2015. Open
Quantitative Sociology & Political Science 1(1). Retrieved from https://openpsych.net/
paper/50
Kirkegaard, E.O.W. & Fuerst, J. (2014). Educational attainment, income, use of social
benefits, crime rate and the general socioeconomic factor among 70 immigrant groups in
Denmark. Open Differential Psychology. Retrieved from https://openpsych.net/paper/21
Kirkegaard, E.O.W. & Fuerst, J. (2016). Inequality in the US: Ethnicity, racial admixture
and environmental causes. Mankind Quarterly 56: 580-606.
KIRKEGAARD, E.O.W. EMPLOYMENT RATES IN SCANDINAVIA
323
Kirkegaard, E.O.W. & Fuerst, J. (2017). Admixture in Argentina. Mankind Quarterly 57:
542-580. Retrieved from http://mankindquarterly.org/archive/issue/57-4/4
Lynn, R., Cheng, H. & Grigoriev, A. (2017). Differences in the intelligence of 15 year olds
in 42 provinces and cities of the Russian Federation and their economic, social and
geographical correlates. Mankind Quarterly 57: 659-668. Retrieved from
http://mankindquarterly.org/archive/issue/57-4/9
Lynn, R., Hampson, S. & Magee, M. (1984). Home background, intelligence, personality
and education as predictors of unemployment in young people. Personality and Individual
Differences 5: 549-557. https://doi.org/10.1016/0191-8869(84)90029-1
Lynn, R. & Vanhanen, T. (2012). Intelligence: A Unifying Construct for the Social
Sciences. London: Ulster Institute for Social Research.
Pew Research Center (2011, January 27). Table: Muslim Population by Country.
Retrieved from http://www.pewforum.org/2011/01/27/table-muslim-population-by-
country/
Rindermann, H. & Thompson, J. (2016). The cognitive competences of immigrant and
native students across the world: An analysis of gaps, possible causes and impact.
Journal of Biosocial Science 48: 66-93. https://doi.org/10.1017/S0021932014000480