Tuesday, 25 May 2010

Daniel Nettle: "Dying Young and Living Fast: Variation in Life History across English Neighbourhoods

Not a talk this time, but a recent article by Daniel Nettle in Behavioral Ecology, discussed by Robin Dunbar in the weekly Journal Club at the Institute of Cognitive & Evolutionary Anthropology:

Nettle, D. (2010). Dying young and living fast: Variation in life history across english neighbourhoods. Behavioral Ecology, 21(2), 387-395.

Where the expected reproductive life span is short, theory predicts that individuals should follow a "fast" life-history strategy of early reproduction, reduced investment in each offspring, and high reproductive rate. I apply this prediction to different neighborhood environments in contemporary England. There are substantial differences in the expectation of healthy life between the most deprived and most affluent neighborhoods. Using data from the Millennium Cohort Study (n= 8660 families), I show that in deprived neighborhoods compared with affluent ones, age at first birth is younger, birthweights are lower, and breastfeeding duration is shorter. There is also indirect evidence that reproductive rates are higher. Coresidence of a father figure is less common, and contact with maternal grandmothers is less frequent, though grandmaternal contact shows a curvilinear relationship with neighborhood quality. Children from deprived neighborhoods perform less well on a verbal cognitive assessment at age 5 years, and this deficit is partly mediated by parental age and investment variables. I suggest that fast life history is a comprehensible response, produced through phenotypic plasticity, to the ecological context of poverty, but one that entails specific costs to children.

My reactions:
I am greatly in sympathy with Nettle's aim of showing that much of working-class behaviour is not pathological, but an adaptive response. There is compelling evidence for this from other sources - for example, the age of menarche is lower for girls from deprived backgrounds, which is suggestive of a need to procreate earlier to make up for a shorter healthy adult lifespan. Similarly, the greater propensity of young working-class men to engage in risky behaviour (drinking, fighting, drug-taking, gambling, etc) can perhaps be explained, at an ultimate level, by their shortened future horizons and lack of access to mates, compared to their middle-class peers. But I'm not sure that Nettle's dependent variables can be similarly explained, at least directly (some of them may be an indirect consequence of risk-taking behaviour, a possibility which he doesn't mention in the article).

His dependent variables are set out most clearly in Figure 2 of the article. I'll go through them one by one.

  1. Maternal age: This seems the clearest candidate for an adaptive response (as already alluded to in the discussion of menarche above). But what is the proximate mechanism here? It is likely to be a complex mixture of age of menarche, greater risk-taking behaviour, cultural imitation (if all one's friends and relatives have children early, one is likely to do the same), and lack of feasible alternatives (we should avoid the trap - as Nettle doesn't, quite - of assuming that middle-class behaviour is natural, and working-class behaviour needs to be explained as a deviation from this norm; from a wider cultural perspective, it's really middle-class behaviour in developed countries that is weird, and we need to explain why so many women sacrifice the possibility of having children at a young age for the sake of "having a career").
  2. Mean birthweight: There doesn't seem to be much of a relationship with neighbourhood quality here, if one excludes the lowest-quality areas. And the low birthweight in these areas can presumably be explained by poor nutrition and other lifestyle factors (notably smoking and/or drinking during pregnancy). Again, this is caused by cultural variation, and it would seem to be pathological rather than adaptive: I'm not sure why Nettle included it as a measure of life-history strategy at all.
  3. Breastfeeding duration: There is a nice strong relationship here, which I would guess results from a cultural difference in the normativity of breastfeeding. Again, I would question the relevance of breastfeeding duration to life-history strategy. Presumably the argument would be that the instinct is not to breastfeed for long, as this inhibits conception, and poor people need to churn babies out quickly. This might work if the mean for the poorest neighbourhoods were 6 months, and the mean for the richest neighbourhoods 18 months; but in fact the mean for the former is 1.5 months, and the mean for the latter is 4 months. Therefore this would not make much difference to overall life history (and anyway how many mothers, even in the most deprived areas, conceive 2 months after a birth?!). It might have been more interesting to look at age of weaning, or perhaps even age of first tooth!
  4. Maternal activities: As Nettle points out, there is no relationship with neighbourhood quality. In fact, this is perhaps the most interesting result in the article, and one that he could have made more of. The range of the neighbourhood means is only about the same as the 95% confidence interval for the mean of each group, indicating that individual differences are far more important than inter-group differences, at least for this sample of white Englishwomen. (I wonder whether the inter-group variation would be similarly small if one looked at different ethnic groups?) This really speaks against the demonisation of working-class mums, and particularly single mums, that tends to take place in the middle-class media.
  5. Paternal involvement: This effect is driven by the large proportion of single mums in poor-quality neighbourhoods: as Nettle points out, if the 0s are removed, the relationship is not significant. So again there is an interesting point to make here about working-class fathers spending more time with their kids than middle-class people might think. Instead of highlighting this, though, Nettle tells a convoluted story about how fathers in high-quality areas are older and therefore more likely to settle down with their children's mother (which doesn't really fit with the longer future horizons that these men are supposed to have). I would think that an equally important factor might be the degree of engagement in risky behaviour leading to casual, unprotected sex. Furthermore, having lived in a very low-quality area myself I would point out that any data on single-parent percentages from such areas are not to be trusted: women there are often "officially" single mums for benefit reasons, while in fact having a partner (or ex-partner) who spends a lot of time with their kids.
  6. Maternal grandmother contact: This relationship is curvilinear: grannies in both low-quality and high-quality areas do not spend that much time with their grandkids (presumably in the latter case, as Nettle points out, due to geographical separation). The difference between low-quality and medium-quality areas seems to work against his theory about women having kids earlier because of their inability otherwise to spend much time with their grandkids: if this really worked, one would expect it to smooth out the curve in Figure 2(f).
So in all, a very interesting study with a laudable aim, and a great use of the Millennium Cohort data, but I'm just not convinced by the central thesis. Nettle needs to explain how proximate mechanisms (such as cultural imitation) come together to result in adaptive responses, and what the evidence for this is. As Celia Heyes pointed out at Journal Club, he also needs to show that people in poor areas who have kids early do better than those who have them later. Otherwise, most of these variables just look like the effects of class differences: how can we assess whether they are adaptive or not. I think he might be better sticking with more physiological variables (such as age of menarche) as it is hard to draw the link from adaptive value to complex psychology, in such detailed contexts.

No comments:

Post a Comment