A motte-and-bailey is a fallacious argument in which an easy to defend position (motte) is conflated with a difficult to defend position (a bailey) – the latter is claimed but only the former is actually defended. Burt's article comprises such an argument. The bailey is that genetics in general is not valuable to social science. From the abstract: “Here, I challenge arguments about the value of genetics for social science,” and later: “My explicit aim is to challenge the claim that genomics has much to offer social science, so much so that social science sans genetics is fatally flawed, scientifically indefensible, and possibly even morally suspect.” But her actual arguments are about the limitations of one particular genetics tool, the polygenic score (PGS), and against “the claimed necessity of incorporating PGSs into social science models as measures of genetic influences.” So Burt's motte is that PGSs are imperfect and should be used cautiously in social science – a position of which we know no opponents. By demonstrating this truism, she implies she has defended a much less tenable position against the value to social science of genetics in general.

First, let's establish why critics argue, and we agree, “that social science research that neglects genetics is, at best, partial and potentially flawed and misleading” (target article, sect. 1, para. 4). The fact that human behavioural traits are ubiquitously heritable (Polderman et al., Reference Polderman, Benyamin, de Leeuw, Sullivan, van Bochoven, Visscher and Posthuma2015) – which Burt does not dispute – creates an enormous problem for social science research that ignores that fact. It means that a substantial source of individual differences remains unobserved, potentially leading to biased estimations and wrong conclusions. Any associations among different behaviours, or associations between the behaviour of parents and their children, or associations between children's behaviour and any variable influenced by parental behaviour, are likely confounded by genetic effects. Ignoring this confounding, which much social science does, renders inferences about causes of these associations invalid. For example, we might interpret the observation that children growing up with a home library have more intellectual skills as adults as a causal effect of the presence of books (Sikora, Evans, & Kelley, Reference Sikora, Evans and Kelley2019). Or we might interpret an association between the warmth of the parent–offspring relationship during adolescence and the quality of the offspring's romantic attachments 60 years later as evidence of “the far-reaching influence of childhood environment on well-being in adulthood” (Waldinger & Schulz, Reference Waldinger and Schulz2016). The unacknowledged genetic confounds do not rule out the hypothesised causal effects, but they invalidate the evidence proffered for these effects (Sherlock & Zietsch, Reference Sherlock and Zietsch2018).

It can therefore be vital to account for genetic confounds. PGSs are one avenue for integrating genetics into social science, but we agree with many of Burt's concerns about the usefulness and misinterpretation of PGSs, several points of which derive from our own work. We are especially concerned about the use of PGSs for “getting genetics out of the way” (target article, sect. 4.1) – that is, including a PGS in an attempt to control for genetic confounding. Isungset et al. (Reference Isungset, Conley, Zachrisson, Ystrom, Havdahl, Njølstad and Lyngstad2022) did this in claiming to demonstrate a causal effect of parents' education on their children's school performance. They concluded that “parental educational advantage is attenuated only to a small degree when accounting for genetics.” But they accounted for a PGS for educational attainment, which captures only a minority of the total genetic variance in school test scores – therefore, it is inevitable that this will only attenuate the parent–child correlation a small amount. It is invalid to infer, as the authors do, that the remaining parent–child correlation is because of a social–environmental effect of parents' education.

But this inappropriate use and interpretation of PGSs does not support Burt's argument against the value of genetics for social science. Ignoring genetics would only worsen the issue. There are various possibilities for integrating genetics into social science so as to identify, minimise, or account for genetic confounds, for example, by testing hypotheses using twin/pedigree datasets, large genetically informed (biobank) datasets, or summary-level genome-wide genetic data. Another possibility would be to adjust for the weakness of the PGS – for example, in the aforementioned Isungset et al. (Reference Isungset, Conley, Zachrisson, Ystrom, Havdahl, Njølstad and Lyngstad2022) study the educational attainment PGS accounted for 6.3% of the variance in school test scores, whereas twin studies estimate that genetic variance accounts for ~55% of variance (Bartels, Rietveld, Van Baal, & Boomsma, Reference Bartels, Rietveld, Van Baal and Boomsma2012). Given that even accounting for this weak PGS already reduces the parent–child correlation by 14–18%, this could be consistent with complete genetic confounding of the parent–child correlation.

It might seem that there is a symmetry in Burt's arguments and ours: Burt is concerned about environmental confounding of genetic effects, whereas we are concerned about genetic confounding of environmental effects. But this leaves out important asymmetries that make Burt's overall argument unreasonable and untenable. First, while Burt argues against the value of genetics for social science, we argue it is important to account for both genetics and environmental effects, and to disentangle them where possible. Second, Burt acknowledges the great efforts that are made in genetics research to minimise the kind of environmental confounding she warns of; but on Burt's side of the debate, without taking into account genetics social science cannot minimise or even recognise genetic confounding. Third, the fixed nature of genes and well-understood process of inheritance provide natural experiments (e.g., identical and non-identical twins, Mendelian randomisation of alleles) that form the bedrock of genetics research and enable detection of genetic (and environmental) variance in traits using different analytic methodologies with different assumptions, as well as allowing cautious causal inferences. In contrast, observational/correlational research in non-genetic social science has no such avenues for establishing causality, leaving associations hopelessly confounded and making it difficult to make inferences about environmental effects.

In conclusion, Burt's argument against the value of genetics for social science is fallacious and counterproductive. The goal of understanding humans and society is best served by making the most of all available methods; accordingly, efforts should be made to integrate genetics into empirical approaches. Articles like Burt's only impede such integration.