When knowing something makes it true

February 7, 2010 at 2:03 pm (science) (, , , , , )

I recently posted a discussion on whether races exist.  I argued that races might exist, but that it wasn’t useful to use the word.

A comment by JL made me rethink my argument.  I haven’t changed my conclusions (notice how rarely this happens?  Its almost like we we use logic to justify our conclusions rather than to deduce our conclusions… but that is a different post entirely). But I have realised that I missed an extremely important point, one which changes the whole concept of scientific hypothesis.

Whether we believe a-priori that IQ differences between races exist can affect whether it is true.

Consider this.  Imagine that scientists say “differences in IQ between races might exist”.  We all see differences in IQ in the real world.  People say, “yes, this could be true”, and act accordingly.  Perhaps, all other things being equal, schools invest in children from the perceived higher IQ race (lets call them race 1).  Perhaps people give jobs to those people from race 1 preferentially – all other things being equal.  Of course, when someone from race 2 is better for the job, they get it.

This generation of children grow up; they are educated in the same way as their parents; they get jobs the same way as their parents.  They have children, and so it goes on.

Now imagine that IQ is determined by both race and upbringing.  People from race 2 have, on average, worse jobs.  They can’t afford high quality education.  So they do, in fact have lower IQ.  The scientists can measure this; the hypothesis is confirmed.  Breaking news!

Does this all sound familiar?  That’s because it already happened.  It is of course trivial; both scientists and non-scientists alike have seen this in action.  But the ramifications for social science are immense.  Normally science works by starting with a “null hypothesis” (how we believe the world might work) and comparing it to a “hypothesis” (something we want to test).  But in measuring IQ differences, our choice of “null hypothesis” can affect the truth of the hypothesis!  If we say, as above, that differences might exist in our null hypothesis, then they do.  If we instead choose the null hypothesis that all races are equal – and insist on this to the world at large – then, and only then, might we be able to measure that IQ differences do not exist.

In other words, the whole world is an experiment: by banning racism we have started a test of whether racial differences really exist.  Only time will tell if this is true or false, whether IQ differences exist are equalising.  But this is only possible since we chose to treat the world as if racial differences do not exist.

This conclusion could be reached on any social science problem where our measures are imperfect.  The problem lies in measures of IQ being biased to an unknown degree by upbringing; but finding a perfect measure is a hopeless task.  It means that science has to work intimately with policy; to measure people we have a scientific and moral obligation to treat all people as equal, because without doing so we can never know if they are.

Whether people really are equal is in some sense irrelevant.  They can only be equal if we assume that they are.



  1. JL said,

    From your earlier reply: I don’t find it clear that what we call “race” has any special value compared to the other levels that we could consider differences at; race, nationality, cultural identity, language etc. all tell you similar information about genes – differences exist, in a weak statistical sense, between all of them.

    The question is whether it is possible that the distributions of complex traits like intelligence differ between races (as categorized by self-assessment) due to genetic factors. We know that human genetic variation is geographically continuous, with all local varieties merging with their neighbors. Does this mean that there cannot be meaningful differences between two far-flung slices of that continuum, such as white Europeans and East Asians? Of course not. Europeans and Asians have been separated from each other in their evolution to the extent that they form two genetic clusters that are distinct from each other.

    We also know that traits such as intelligence are polygenic, and therefore may differ between individuals who have different sets of alleles. This means that our knowledge of human genetic variation is compatible with the possibility of inter-racial differences in, say, IQ.

    I’m not quite sure what to make of your specific argument about IQ. It lacks references to support your rather strong conclusions, which are contradicted by tons of research.

    Firstly, it is not true that upbringing has a large effect on IQ. Consider identical twins separated at birth and raised in different families. According to many studies (e.g. Bouchard’s Minnesota study) the IQs of such twins are within a few points of each other. Furthermore, biologically unrelated children raised in the same family have IQs that are no more similar to each other than those of randomly chosen people in the same population. Still more, the IQs of adopted children are more like those of their biological parents rather than their adoptive parents.

    Due to results like these the prevailing view in psychology is that heritability accounts for most of the variation in IQ, at least within populations. Parents influence the IQs of their children mainly through their genes, not their parenting styles. Some of the variation in IQ is explained by non-genetic factors, but parenting is not important among them.

    As to racial IQ differences, the debate is of course entirely about the black-white mean IQ gap (no one is interested in, say, the fact that East Asians routinely outscore Europeans). The problem for those who insist that the IQ gap is entirely environmental is that the predictions based on this hypothesis have again and again failed, whereas those of the hereditarian camp have been confirmed.

    For example, huge intervention programs like the Head Start and No Child Left Behind have failed to appreciably decrease racial achievement gaps. Another example is that the children of upper middle class blacks do not have as high standardized test scores as their white peers or even lower class white kids. What’s more, black children adopted by white upper middle class parents have IQs similar to black kids raised by black parents. All of these results match the predictions of the hereditarians, and contradict those of the “nurturists”. So on what grounds do you insist on privileging the environmental hypothesis?

    Lastly, I’ll quote from the A.W.F. Edwards article I referred to in my earlier reply: “[I]t is a dangerous mistake to premise the moral equality of human beings on biological similarity because dissimilarity, once revealed, then becomes an argument for moral inequality.”

    • thinkingdan said,

      My apologies for the extremely long post – I’m trying to be thorough (for a change…)

      I agree entirely that differences in all sorts of traits can and do exist between populations (which means I concede your first two points). My concern is whether it benefits us to explicitly include differences in our “prior beliefs” or not. (Not whether we should study them or not – I think we should).

      I must confess to being fairly ignorant of much of the IQ literature, which is why I included no references (that wasn’t my original purpose of discussion, but a side line to it). I was operating on “self-evident truths” from personal observation and popular science news, which are of course uncontrollably biased. An example: people from poor schools do dramatically worse on standard exams than those from wealthy schools; I’m sure this corresponds to IQ to some degree. Yet we don’t really believe that there is a huge difference in natural ability between those from poor backgrounds and those from wealthy ones (though again, there could well be a small real difference).

      To rectify this, I’ve done a bit of searching in the literature to find the state of IQ thoughts in science. I think it can be reasonably described as “controversial”. Nature recently featured a two-sided discussion on whether science should even study IQ in the context of race at all (http://www.nature.com/nature/journal/v457/n7231/full/457786a.html). As you point out, twin studies show remarkable correlation between genetics and IQ. This is no surprise – we all believe that intelligence is inherited to some degree within families. Additionally, there is a consistent IQ difference of 4-5 points between the sexes (Lynn, Personality and Individual Differences
      February 1998, 24:289-290; Blinkhorn, Nature 438:31-32 2005). There are huge continental differences of tens of points (nicely illustrated by http://alfin2100.blogspot.com/2009/04/iq-by-nation-iq-by-race-us-iq-inherited.html), and large (10 point) geographical differences between areas of the single country of Italy (Lynn 2010, Intelligence 38:93-100). A summary of such results is given by (Rushton and Jensen 2005, Psychology, Public Policy, and Law 11:235–294).

      These researchers look for differences and find them. The problem is the circular nature of intelligence: low IQ in parents leads to poor nutrition, low childhood support, poor education, and hence low IQ in children (this is called the “self-fulfilment hypothesis”). Again, the evidence that such a cycle exists is uncontroversial – whether it explains the whole distribution of Iq’s is strongly disputed. Much apparently “genetic” variation in IQ is explained by conditions in the womb (Devlin et al. . 1997, Nature 31;388(6641):468-71). Perceptions of IQ change behaviour and hence lifelong learning potential (The Confounding of Perception of I.Q. on a Measure of Adaptive
      Behavior, Bobner, Ronald F et al – sorry this is a conference proceeding; also Sutherland and Goldschmid 1974, Child Development, 45:852-856). Early parenting factors are important for long term academic achievement (Englund et al. Journal of Educational Psychology, 2004), which means that family size and social class are going to be important too.

      The consensus in the literature is that self-fulfilment does occur but nobody has modelled it in such a way that it accounts for all observed IQ differences (Jussim and Harber 2005, Personality and Social Psychology Review, 9:131-155; also http://psychology.uwo.ca/faculty/rushtonpdfs/PPPL1.pdf). I got the feeling that IQ experts take this as evidence for the existence of real IQ differences. I don’t know whether they are wrong; I’m not an expert in intelligence or sociology. However I know about statistics and I’m not at all surprised that they can’t explain the data. On such a circular problem, traditional statistical methods are hopelessly inadequate. A linear regression can only account for linear effects of a confounding variable, but because low IQ of a parent results in low IQ of a child through many environmental factors, there is a multi-generational correlation between IQ and any factor, i.e. a highly non-linear relation. Only a full model-based method could hope to unravel this, and I’m sure nobody has done this (probably because the data – and inference techniques – is unavailable). It might not be possible with the currently observable properties of people and social constructs.

      These factors can’t explain the observation that twins brought up apart are very similar in IQ whereas adopted children differ – that is uncontroversial – but differences between populations are much harder to show, because populations are geographically, socially and historically segregated leading to the confounding factors above. Even adoption of white children into black families and vice versa can’t tell us the truth because society places different pressures on each, and pre-birth effects might be significant (so called “epigenetics” – a currently trendy and upcoming field). The simple fact of being black creates a different environment that we can’t handle in our models.

      So we are back to square one – and my point in this second post – that unless we perform the “global experiment” of treating everyone as equal, they will remain different and our best scientific methodology cannot detect the truth. There may well be important, significant differences between populations in terms of IQ but scientifically we have to proceed on the basis that there isn’t. (Or, we need to convince society not to care about such things, but I’m not hopeful about that…)

      We don’t expect the “no child left behind” and other such programs to work completely, at least not within several hundred years, because society, people and culture change only slowly; everything is contingent (e.g. the stigma against the Jews has persisted throughout recorded history). Of course, we would hope for a consistent decrease in the gap, measurable on a generational scale, which I don’t believe has been disproven.

      I’m not sure I agree with “positive discrimination”, despite all these arguments – there are many people of different races that are in the same sort of rut. But certainly an “equivalent” achievement for a person from a poor background in “more” of an achievement than for someone who had a lot of social support, which leads to a mild (and I believe unpractised) form of positive discrimination being acceptable.

      The environmental hypothesis is privileged not because we believe it more likely to be true, but because it is scientifically the “correct” one (meaning it will give the least bias in the long run), and morally also the “correct” one (because we can’t justify treating people unequally unless we are sure).

      I really like the Edwards quote, and it does lead to some deep questions relevant to the future; some lessons we might need to learn from the problems of race and racism. For example, if we can identify desirable genes such as for intelligence, we would know a persons “value” a-priori and not just judge on the basis of actions. We could even create “better” people (a common theme in literature). How should society treat people in this case? Of course my argument breaks down.

  2. thinkingdan said,

    In case it wasn’t clear from above: JL’s view is the dominant one in the literature. People find differences despite accounting for other factors as well as they can. I draw a different conclusion than in the majority of the literature for 2 reasons: firstly, the statistical methods used are inadequate; secondly, the feedback loop changes the correct societal position (in my own, and at least some experts opinions).

  3. Social feedback and IQ « Thoughts about thinking said,

    […] made me realise that the main reason that we should ignore these differences is because there is a feedback between how we see the world, and how it is. This probably applies to lots of social phenomena, but I’m thinking about […]

  4. S Blumenthal said,

    ***The simple fact of being black creates a different environment that we can’t handle in our models***

    The idea of “expectancy effects,” is contradicted by the findings of the Minnessota Transracial Adoption study (also note the research by David Rowe on ‘X factor effects’). A special analysis of the Scarr study compared parents who believed that they had adopted a Black baby but, really, had adopted a Mixed-Race (Black-White) child. The average IQ for these Mixed-Race children was just about the same as for other Mixed-Race children and above that for adopted Black children. This was true even though the parents who adopted these Mixed-Race children thought their babies really had two Black parents.

  5. S Blumenthal said,

    There’s also a good summary on the Gene Expression site.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: