The analysis of ‘moderation’, ‘interaction’, ‘mediation’ and ‘longitudinal growth’ is widespread in the human sciences, yet subject to confusion. To clarify these concepts, it is essential to state causal estimands, which requires the specification of counterfactual contrasts for a target population on an appropriate scale. Once causal estimands are defined, we must consider their identification. I employ causal directed acyclic graphs and single world intervention graphs to elucidate identification workflows. I show that when multiple treatments exist, common methods for statistical inference, such as multi-level regressions and statistical structural equation models, cannot typically recover the causal quantities we seek. By properly framing and addressing causal questions of interaction, mediation, and time-varying treatments, we can expose the limitations of popular methods and guide researchers to a clearer understanding of the causal questions that animate our interests.

The human sciences should seek generalisations wherever possible. For ethical and scientific reasons, it is desirable to sample more broadly than ‘Western, educated, industrialised, rich, and democratic’ (WEIRD) societies. However, restricting the target population is sometimes necessary; for example, young children should not be recruited for studies on elderly care. Under which conditions is unrestricted sampling desirable or undesirable? Here, we use causal diagrams to clarify the structural features of measurement error bias and target population restriction bias (or ‘selection restriction’), focusing on threats to valid causal inference that arise in comparative cultural research. We define any study exhibiting such biases, or confounding biases, as weird (wrongly estimated inferences owing to inappropriate restriction and distortion). We explain why statistical tests such as configural, metric and scalar invariance cannot address the structural biases of weird studies. Overall, we examine how the workflows for causal inference provide the necessary preflight checklists for ambitious, effective and safe comparative cultural research.

Confounding bias arises when a treatment and outcome share a common cause. In randomised controlled experiments (trials), treatment assignment is random, ostensibly eliminating confounding bias. Here, we use causal directed acyclic graphs to unveil eight structural sources of bias that nevertheless persist in these trials. This analysis highlights the crucial role of causal inference methods in the design and analysis of experiments, ensuring the validity of conclusions drawn from experimental data.

The transition to sedentary agricultural societies in northern China fuelled considerable demographic growth from 5000 to 2000 BC. In this article, we draw together archaeobotanical, zooarchaeological and bioarchaeological data and explore the relationship between several aspects of this transition, with an emphasis on the millet-farming productivity during the Yangshao period and how it facilitated changes in animal husbandry and consolidation of sedentism. We place the period of domestication (the evolution of non-shattering, initial grain size increase and panicle development) between 8300 and 4300 BC. The domestication and post-domestication of foxtail (Setaria italica) and broomcorn (Panicum miliaceum) millet increased their productivity substantially, with much greater rate of change than for rice (Oryza sativa). However, millets are significantly less productive per hectare than wet rice farming, a point reflected in the greater geographical expanse of northern Neolithic millet cultures (5000–3000 BC) in comparison with their Yangtze rice-growing counterparts. The domestication of pigs in the Yellow River region is evidenced by changes in their morphology after 6000 BC, and a transition to a millet-based diet c. 4500–3500 BC. Genetic data and isotopic data from dogs indicate a similar dietary transition from 6000 to 4000 BC, leading to new starch-consuming dog breeds. Significant population increase associated with agricultural transitions arose predominately from the improvement of these crops and animals following domestication, leading to the formation of the first proto-urban centres and the demic-diffusion of millet agriculture beyond central northern China between 4300–2000 BC.

Hunter–gatherer populations underwent a mass extinction in the Neolithic, and in present times face challenges such as explicit sedentarisation policies. An exception is in Nepal, where the nomadic Raute people receive monthly governmental individual payments. One consequence of the money transfers has been a significant increase in alcohol consumption, with nearly all individuals drinking industrially produced alcohol. Here we investigate the Raute demography based on a full census of 144 individuals. We show that the Raute exhibit the short life expectancies typical of other hunter–gatherer populations from Africa, Asia and America. Bayesian survival trajectory analysis demonstrated that heavy drinking by either parent substantially reduces offspring survival to age 15. Bayesian regressions revealed a significant effect of heavy drinking on maternal fertility by decreasing the number of living children and reducing the proportion of live children at the end of maternal reproductive life. Although the absence of data prior to monetary support precludes a direct assessment of long-term demographic trends, relatively stable population sizes over the last decades and a fertility rate close to the replacement rate rule out an imminent population crash. Further studies are required to elucidate the Rautes’ origins and relationship with other nomadic people in the region.

Previous research in the evolutionary and psychological sciences has suggested that markers or tags of ethnic or group membership may help to solve cooperation and coordination problems. Cheating remains, however, a problem for these views, insofar as it is possible to fake the tag. While evolutionary psychologists have suggested that humans evolved the propensity to overcome this free rider problem, it is unclear how this module might manifest at the group level. In this study, we investigate the degree to which native and non-native speakers of accents – which are candidates for tags of group membership – spoken in the UK and Ireland can detect mimicry. We find that people are, overall, better than chance at detecting mimicry, and secondly we find substantial inter-group heterogeneity, suggesting that cultural evolutionary processes drive the manifestations of cheater detection. We discuss alternative explanations and suggest avenues of further inquiry.

Natural selection should favour litter sizes that optimise trade-offs between brood-size and offspring viability. Across the primate order, the modal litter size is one, suggesting a deep history of selection favouring minimal litters in primates. Humans, however – despite having the longest juvenile period and slowest life-history of all primates – still produce twin births at appreciable rates, even though such births are costly. This presents an evolutionary puzzle. Why is twinning still expressed in humans despite its cost? More puzzling still is the discordance between the principal explanations for human twinning and extant empirical data. Such explanations propose that twinning is regulated by phenotypic plasticity in polyovulation, permitting the production of larger sib sets if and when resources are abundant. However, comparative data suggest that twinning rates are actually highest in poorer economies and lowest in richer, more developed economies. We propose that a historical dynamic of gene–culture co-evolution might better explain this geographic patterning. Our explanation distinguishes geminophilous and geminophobic cultural contexts, as those celebrating twins (e.g. through material support) and those hostile to twins (e.g. through sanction of twin-infanticide). Geminophilous institutions, in particular, may buffer the fitness cost associated with twinning, potentially reducing selection pressures against polyovulation. We conclude by synthesising a mathematical and empirical research programme that might test our ideas.

Humans show remarkable differences in social behaviour between families, groups, communities and cultures, whereas such group-level within-species variation in socio-behavioural propensities is typically overlooked in other species. Studies on intraspecific variation in animal social structures are needed to inform an evolutionary account of human sociality. Here, we study multiple independent bonobo populations (n = 6) in zoological settings to investigate if and how bonobos (n = 70) show group-specific signatures in sociality. By applying tailored Bayesian statistical methods, we find that beyond individual and dyadic variation, the groups substantially differ from each other in core dimensions of great ape sociality: social proximity, grooming and play. Moreover, the groups’ network structures are distinct regarding cohesiveness and clustering, with some groups forming cohesive wholes, while others showcasing high levels of sub-grouping. Overall, while there is consistent evidence of differences in sociality between the groups, the patterns of cohesiveness and clustering are not consistent across the networks. This suggests that rather than groups having different levels of sociality, different patterns of sociality exist in each group. These findings warrant caution with characterising bonobos’ behavioural phenotype at the species level, and identify an essential source of variation that needs to be integrated in phylogenetic analyses.