Wednesday 16 November 2016

The accuracy of stereotypes


immigrants in Denmark


Are immigrants more likely to claim benefits, or is this a stereotype?

A stereotype is a preliminary insight. A stereotype can be true, the first step in noticing differences. For conceptual economy, stereotypes encapsulate the characteristics most people have noticed. Not all heuristics are false.

Here is a relevant paper from Denmark.

Emil O. W. Kirkegaard and Julius Daugbjerg Bjerrekær. Country of origin and use of social benefits: A large, preregistered study of stereotype accuracy in Denmark. Open Differential Psychology.


This study is interesting, in that it was pre-registered, so its absence would have been noticed.  It compares stereotypes against actual data to get a test of accuracy. I was particularly struck by how the authors studied the answers at each wave of data collection, and tracked down those who gave perplexing answers, then refining their survey questions to reduce misunderstandings.

The paper also points out an unremarked aspect of stereotypes: they may be too weak. Stereotypes have to show a correlation with the facts, and be good predictors. You have to get the slope right, and also the intercept. It is not enough to have a vague notion that immigrants are on benefits, you ought to be able to estimate how many are on benefits.  A stronger stereotype would be a more accurate perception of reality.

A nationally representative Danish sample was asked to estimate the percentage of persons aged 30-39 living in Denmark receiving social benefits for 70 countries of origin (N = 766). After extensive quality control procedures, a sample of 484 persons were available for analysis. Stereotypes were scored by accuracy by comparing the estimates values to values obtained from an official source. Individual stereotypes were found to be fairly accurate (median/mean correlation with criterion values = .48/.43), while the aggregate stereotype was found to be very accurate (r = .70). Both individual and aggregate-level stereotypes tended to underestimate the percentages of persons receiving social benefits and underestimate real group differences.
In bivariate analysis, stereotype correlational accuracy was found to be predicted by a variety of predictors at above chance levels, including conservatism (r = .13), nationalism (r = .11), some immigration critical beliefs/preferences, agreement with a few political parties, educational attainment (r = .20), being male (d = .19) and cognitive ability (r = .22). Agreement with most political parties, experience with ghettos, age, and policy positions on immigrant questions had little or no predictive validity.
In multivariate predictive analysis using LASSO regression, correlational accuracy was found to be predicted only by cognitive ability and educational attainment with even moderate level of reliability. In general, stereotype accuracy was not easy to predict, even using 24 predictors (k-fold cross-validated R2 = 4%).
We examined whether stereotype accuracy was related to the proportion of Muslims in the groups. Stereotypes were found to be less accurate for the groups with higher proportions of Muslims in that participants underestimated the percentages of persons receiving social benefits (mean estimation error for Muslim groups relative to overall elevation error = -8.09 %points).
The study was preregistered with most analyses being specified before data collection began



The observed correlation of .7 is big, and useful. A majority of immigrants from Syria, Somalia and Kuwait are on benefits, as are those from Iraq and Lebanon. Even more to the point, if the benchmark is 25% for Danish citizens, then there are 19 countries with higher benefit rates. More positively, there are countries with lower rates, presumably because they are younger and employed. The data plot does not give us any guide to numbers from each country. However, later in the paper it is shown that immigrant population size is not relevant in judging benefit rates accurately.

The best predictor of having accurate stereotypes was cognitive ability (81% of simulations), followed by educational attainment (74% of simulations). Respondents underestimate the number of Muslims on benefits.

This is a very good paper. Data handling is exceptional, and well explained. There are lots of Figures and Tables. The sample is large and representative. The results have been looked at carefully, to identify those who participated without paying much attention to the questions. The data are available for re-analysis.

The high accuracy of aggregate stereotypes is confirmed. If anything, the stereotypes held by Danish people about immigrants underestimates those immigrants’ reliance on Danish benefits.


  1. I wonder whether they meant literally immigrants (i.e. those born elsewhere) or "immigrants" as it is commonly used in Britain, to mean those born elsewhere plus those descended from immigrants who have arrived since the The War.

    Mind you, the discrepancy between those two uses might be pretty small in Denmark where (I'm guessing) mass Third World immigration might be a fairly recent phenomenon. Using people aged 30-39 might also help keep the meaning distinct.

    1. P.S. A sample of 484 persons seems rather small to me, but it's many a year since I did statistical sums.

    2. Immigrants include 1 and some 2. gen and no >=3 gen.

      Large scale immigration started in the 1960s, but accelerated in the 1980s (and now).

    3. Median sample size in psychology in mainstream journals is about 40. Probably university students.

  2. James, the benefits data concern only 30-39 year olds, so age cannot be a larger factor in differences. We specifically chose such an age bracket to avoid problems.

  3. Yes, saw that, but what I wondered was that if Nigerians had actually arrived only when in that age range, that could account for their better employment status.

  4. Re: DK immigrant numbers from some countries

    Recalculated data from IAB report.

    %Edu (DK immigrant qual. primary or less, 2010)
    %Edu NEdu SubTotal Origin
    44.35 161926 365095 Total
    48.69 764 1569 Syria
    55.84 3806 6816 Somalia
    45.06 2418 5366 Afghanistan
    44.78 6400 14291 Iraq
    44.77 381 851 Ethiopia
    43.35 561 1294 Ghana
    40.36 360 892 Nigeria
    40.61 372 916 Uganda
    32.43 3506 10810 Iran
    50.68 5402 10658 Lebanon
    67.62 19958 29514 Turkey
    53.07 5278 9945 Pakistan
    32.01 338 1056 Kenya
    29.63 232 783 Tanzania
    35.83 301 840 Algeria
    57.7 2343 4061 Romania
    33.26 441 1326 Egypt
    38.19 299 783 Tunisia
    53.94 513 951 Kuwait
    56.23 11519 20485 Poland
    58.03 477 822 Slovakia
    62.82 992 1579 Bulgaria

    %Edu (DK immigrant qual. degree or higher, 2010)
    %Edu NEdu SubTot Origin
    30.61 273 892 Nigeria
    29.28 8005 27338 Germany
    28.64 399 1393 Austria
    28.56 842 2948 Spain
    25.31 950 3754 Italy
    24.38 385 1579 Bulgaria
    24.19 457 1889 Hungary
    23.69 213 899 Portugal
    23.64 1085 4589 Ukraine
    22.24 903 4061 Romania
    22.04 3259 14784 BosniaHerzegovina
    19.79 857 4330 Lithuania
    19.67 4030 20485 Poland
    19.22 158 822 Slovakia
    14.93 343 2297 Macedonia
    14.71 392 2665 Latvia
    8.6 2537 29514 Turkey

    1. thanks. can you send me an excel file? It would make it easier to display. Direct Message me on Twitter?

    2. The IAB data is in

      Previously there were some questions if the Nigerian immigrants in UK were typical of those back in Nigeria. The Nigerians in UK are well up in the top 10 in terms of % degree holders and thus a very self selected group. If %Edu is a proxy for the expected relative IQ of the group, it might not be surprising if the UK Nigerians might be able to perform better than the Chinese
      or Indians there.

      %Edu (UK immigrant qual. degree or higher, 2010. NEdu>10000)
      %Edu NEdu SubTot Origin
      72.28 79647 110189 Nigeria
      63.86 33106 51840 China
      45.58 249502 547395 India

      %Edu (US immigrant qual. degree or higher, 2010. NEdu>10000)
      %Edu NEdu SubTot Origin
      82.41 136769 165963 Nigeria
      82.04 1195820 1457640 India
      53.97 644529 1194175 China

    3. Thanks. Elite performance is a good indicator of underlying average ability, so long as one has an accurate estimate of the size of the population. Nigeria is predicted to rise to 1 billion within a generation, so hard to get precise numbers of what the number of Nigerian are at the moment, I would think

    4. The brain drain data is a nice way of adjusting for emigration selection. The idea is that I do a few more of these large scale country of origin studies. Then do a big meta-analysis where I take into account the emigration selection data too, see how that affects results. I think there will be substantial effects for a few countries such as South Africa.

      But in general, it's hard to get the right data:

      Plenty of data for the US.

      An okay amount of data for the UK found so far.

      Only have crime data for Germany (study almost done).

      Sweden? France? Belgium? Italy? No data that I have found so far.