Galton’s Problem is one of identifying functional relationships in a set of observations where the observations may be related through borrowing or descent. The problem has long been recognized in the fields of anthropology and cross-cultural research, but has been neglected in economics, even in cases where it is likely to be most acute, such as in regression models where each observation is a nation. The appropriate treatment for Galton’s Problem was developed in the 1980s by Malcolm M. Dow, Douglas R. White, and Michael L. Burton. Their solution is to estimate spatial models, where weight matrices for physical distance control for relationships of borrowing, and weight matrices for linguistic similarity control for relationships of descent. This paper documents a method for estimating a weight matrix based on language phylogenetic relationships. The method is applied to the 186 cultures of the Standard Cross-Cultural Sample, and applied to a set of 216 nations. The resulting weight matrices are made available for any researchers wishing to control for Galton’s Problem.

Cross-cultural researchers often combine several component variables into a composite index or “scale.” The value of a scale for a particular observation is sensitive not only to the values of its component variables, but also to the values of the weights used to combine the components. This sensitivity to weight values is unfortunate, given that the choice of weighting scheme is in some ways arbitrary. A method is presented here, based on linear programming, which reduces the sensitivity of a scale to the component weights. An example scale is produced, for the prevalence of markets and property rights in the societies of the Standard Cross-Cultural Sample. A program, written for GAMS, is included.

Changes in R necessitate updated R scripts for Eff, E. Anthon, & Dow, Malcolm M. (2009). How to Deal with Missing Data and Galton’s Problem in Cross-Cultural Survey Research: A Primer for R. Structure and Dynamics, 3(3). Retrieved from: http://www.escholarship.org/uc/item/7cm1f10b

Using the Standard Cross-Cultural Sample, Roes and Raymond (2003) find that large societies are more likely to be located in resource-rich environments, engage in warfare, and hold beliefs in gods actively supporting human morality (“moralizing gods”). We revisit the Roes and Raymond study, using the methods presented in a series of papers by Dow and Eff. Our findings suggest that moralizing gods are less likely to be found in resource-rich environments or amongst societies frequently engaged in external warfare. We find that cultural transmission over geographic space is the most significant force in conditioning belief in moralizing gods; that moralizing gods are more likely to be found in pastoral societies; and that the relationship between society size and moralizing gods is non-linear, with both very large and very small societies less likely to have moralizing gods. We explain this non-linearity by arguing that the functions of moralizing gods can also be performed by the state, and we also argue that moralizing gods play an important role in stabilizing property rights.

Parental investment represents expenditures of time and resources for the purpose of increasing the biological fitness of one’s offspring. We examine whether parental investment has incidental effects on per capita GDP, using a cross-section of 209 countries and territories. Our work is a revisiting of a 2002 paper by Nigel Barber, with some notable methodological improvements: we use a spatial lag model to control for Galton’s problem, use multiple imputation to handle the issue of missing data, and consider the implications of endogeneity. Our results show that variations in parental investment explain nearly half of the variation in per capita GDP. We find the role of health investments to be especially critical: increases in offspring health raise the rate of return to parental investment, which prompts even more investment, creating a deviation-amplifying process.

Explanations of the causes of war fall roughly into two schools: those arguing for the primacy of environment and technology, and those arguing for the primacy of sociopolitical factors. We re-examine two hypotheses from the former school, viz, societies are more likely to engage in war when they have: 1) more productive subsistence technology; and 2) higher population density. Using data from the Standard Cross-Cultural Sample, and up-to-date multivariate modeling methods, we find only qualified support for the first hypothesis and find the reverse relationship for the second: higher population densities lead to less war, not more. We show that omitted variable bias can explain the failure of previous studies to discover this relationship. Finally, we show that the two schools seem to be equally correct, in that each explains about the same proportion of the variation in frequency of external war.

Multiple imputation (MI) has become the preferred method for dealing with missing data in survey research. MI involves three steps: creating m multiply imputed complete datasets; estimating models on each of the m datasets using any standard statistical procedure; combining the resulting multiple estimates of each statistic of interest. This paper provides R programs for MI, and offers some advice for employing MI with data drawn from the Standard Cross-Cultural Sample (SCCS). A second set of R programs combines estimates from the m imputed data sets, and also deals with the problem of network autocorrelation effects, i.e., Galton’s Problem or the non-independence of cases, using two-stage instrumental variables (IV) regression. The objective of the paper is to provide programs, advice and explanations that will help researchers employing cross-cultural survey data, especially the SCCS, to deal with the twin problems of missing data and network autocorrelation effects, using the open source statistical package R. The paper is intended to complement a recent suite of publications by Dow and Eff where both theoretical and empirical issues underlying these two problems are discussed in detail.