Intelligence of 5 year olds in the UK

 

Birth samples are usually the best, because they define a population exactly. There should be no biases in the sample. However, unless one is willing to pursue them to the ends of the earth, as the children grow up they tend to get lost to researchers. Families move, break up, forget to answer letters, and generally drift off in their own very human way. This is particularly the case for those who don’t have much of an interest in scientific surveys, which after all pay them nothing, and rarely have anything interesting to report for many years. Taking part in a longitudinal survey is an almost perfect measure of delayed gratification, and appeals to higher intelligence, conscientious families.

So it is a pleasure to report on a study which has done its very best with the available data on the intelligence of 5 year olds. Although quite a lot of the probably less able subjects dropped out of the study, the authors have made the best of what data they have available, despite selective fall-out. The best is the best, and better than none. We must be Bayesians in the real world, rather than dreaming of data perfection.

http://www.sciencedirect.com.libproxy.ucl.ac.uk/science/article/pii/S0160289613000925#

http://dx.doi.org.libproxy.ucl.ac.uk/10.1016/j.intell.2013.06.019,

Recent data for majority and racial minority differences in intelligence of 5 year olds in the United Kingdom

Richard Lynn and Helen Cheng looked at the UK Millenium Cohort Study of 14,379 5 year old British children. The study began with 19,000 babies born in September and August 2001 and over-sampled ethnic minorities. 79% made it to the 5 year stage, when they were tested with the British Ability Scale. Parental education, roughly measured, correlated .31 with children’s IQs, and .38 for maternal education and child IQ for the non-whites.

The sampling is somewhat complicated, because in searching for poor, disadvantaged children they over-sampled poor wards, and so the figures do not directly match population statistics. There are advantaged wards, disadvantaged wards and ethnic wards. The eyes glaze. Laudable aims, but confusing for the researcher looking for general trends. I thought they had got 72% of the advantaged subjects, 66% of the disadvantaged and 61% of the ethnic subjects, but the next table suggested that calculation was wrong. You try reading the technical paper on sampling and see if you can pin down exactly how they went about their business.    http://www.cls.ioe.ac.uk/shared/get-file.ashx?id=108&itemtype=document

Advantage or disadvantage makes no difference to household size at 3.9 persons, but ethnic households at 4.8 persons seem to have an extra child.

The results are interesting, but the numbers in the ethnic minorities do not correspond to the proportions of ethnic numbers in the population in 2001. This suggests that the selective fall-out has been considerable, and has varied between different ethnic groups. Unless the actual fall-out rate is known, little reliance can be placed on the figures, other than to say that participating children are probably more educated than average.

At a glance, even allowing for selective fall-out, the results are roughly in line with other school and population surveys, but the differences are often smaller, probably because of sampling.  

IQs of 5 year olds in the United Kingdom in the Millennium Cohort Study

Group	N	Mean	SD	d	ANOVA
	p
White	12,417	100.87	14.61	–		–
Mixed	424	99.73	15.25	− .08		.988
Indian	377	96.87	14.80	− .27		.033
Pakistani	664	85.62	14.90	− 1.04		.000
Bangladeshi	275	86.00	16.28	− 1.02		.000
Black Caribbean	170	96.68	15.28	− .29		.426
Black African	294	90.02	14.49	− .74		.000
Other Black	34	91.95	15.58	− .61		.736
Other Asian	104	94.11	15.75	− .46		.148
Chinese	20	102.06	14.88	.08		1.000
Other ethnic	81	93.16	16.83	− .53		.058
Total	14,860	100.00	15.00	–		.000