Reader Comments

Post a new comment on this article

Conclusions do match results

Posted by Nicolas_Claidière on 29 Jun 2012 at 14:52 GMT

In a comment posted 25th May 2012, Richard McElreath suggests that we draw conclusions in our article that contradict the results of the statistical analyses of our data that we present. He suggests that contrary to our title, the analyses support hyper-conformity. In this response we attempt to provide further clarification of why that is not the case.

First, we would like to record our thanks to McElreath for suggesting the additional statistical analyses that we included in our paper. These were offered anonymously at the time, but if we had known their source we would have been happy to record our acknowledgments to McElreath in the paper and we are pleased to do that now.

In our statistical analyses, the AIC weight of 61% to which McElreath draws attention is for a cubic model (our Table 2). In principle this could reflect either hyper-conformity (referred to elsewhere in the literature as a “conformist bias”) or weak conformity, terms we define unambiguously on page 2 of our article, alongside an explanatory graphic. We provide a fuller justification and explanation of our terminology in a recent review article [1], that we encourage readers interested in this issue to consult.

Inspection of our figure 3 shows that the cubic effect, having an inverted-S form, can be consistent only with weak conformity and not hyperconformity. We note that our analyses suggest the difference in fit of the linear and cubic models is not substantial – the AICc statistics that give a measure of fit are respectively 172.6 and 171.7. However, the important point is that the curve corresponding to the cubic model is the "opposite" of that predicted by hyper-conformity. Indeed, in the terminology used earlier by McElreath and colleagues [2], this is called ‘non-conformity’ [2] and elsewhere ‘anti-conformity’ [3]. Here we present a new diagram that directly juxtaposes the forms of conformity distinguished in our figure 1 and the results described in our figure 3 to make clear the logic we describe above.

To see the diagram please follow this link (we are very grateful to Krista Hoff and her team for facilitating the inclusion of this figure in our response): http://www.plosone.org/at...

To avoid doubt, we would like to finish this response by emphasizing that in designing our study we held no prejudice against the possible occurrence of hyperconformity; we simply aimed to test for it, given that Boyd, Richerson and colleagues have made it such a theoretically important construct, yet empirical tests remain few. However, we cannot but report the results we obtained in this particular study.

Nicolas Claidière and Andrew Whiten

References
[1] Claidière N, Whiten A (2012) Integrating the study of conformity and culture in humans and non-human animals. Psychological Bulletin 138: 126-145.
[2] Efferson C, Lalive R, Richerson PR, McElreath R, Lubell M (2008) "Conformists and mavericks: the empirics of frequency-dependent cultural transmission." Evolution and Human Behavior 29: 56-64.
[3] Mesoudi A, Lycett SJ (2009) "Random copying, frequency-dependent copying and culture change." Evolution and Human Behavior 30: 41-48.

No competing interests declared.