Carlos A. Botero

From InterSciWiki
(Redirected from Botero, Carlos A.)
Jump to: navigation, search


j<-EA[rownames(SCCS),c("stability","abundance")]
setDS("SCCS")
dx<-data.frame(dx,j)
tail(names(dx)) # to check results
EA34d4 - Write csv for Biodiversity.Rasters.Rdata questions for Michael.

EA Variables and Supplementary Materials

The following variables were obtained from descriptions in the Ethnographic Atlas (references 36, 37): geographic location (v104 and v106), religious beliefs [v34, recoded as supportive of human morality (category 4) versus not (categories 1–3)], agriculture [v28, recoded as present (categories 2–6) or absent (category 1)], animal husbandy [v40, recoded as presence (categories 2–7) or absence of large domestic animals (category 1)] (this is replaced by 1-5 versus 6-7 in the DEf EA34d4 model), and political complexity (v33, number of jurisdictional hierarchy levels beyond local communities). In accordance with ref. 9, we used the presence of animal husbandry as a proxy for movable property and explored nonlinear effects of political complexity in our models.

Supplementary Materials. -- Methods: AIC - Akaike information criteria PDF for Supplementary Materials

Author affiliations

  • a-b Initiative for Biological Complexity,
  • a-b Department of the Interior Southeast Climate Science Center,
  • c Department of Forestry and Environmental Resources, North Carolina State University, Raleigh, NC 27695;
  • b Department of Biology, Washington University in St. Louis, St. Louis, MO 63130;
  • d Department of Ecology & Evolutionary Biology and Department of Geography and Program in Planning, University of Toronto, Toronto, ON, Canada M5S 3E8;
  • e School of Art History, Classics and Religious Studies, Victoria University of Wellington, Wellington 6140, New Zealand;
  • f Department of Human Dimensions of Natural Resources, Colorado State University, Fort Collins, CO 80523;
  • g School of Psychology, University of Auckland, Auckland 1142, New Zealand;
  • h School of Philosophy, Research School of the Social Sciences, Australian National University, 0200 Canberra, Australia; and
  • i Department of Linguistic and Cultural Evolution, Max Planck Institute for History and the Sciences, 07745 Jena, Germany

Assistant Professor, Department of Biology, Washington University in Saint Louis Evolution, Evolutionary ecology, Behavioral Ecology, Ornithology Verified email at email.wustl.edu - Homepage

Publications

http://cabotero.weebly.com/publications.html

Publications and Botero blog

EA (Ethnographic Atlas) Map

Carlos A. Botero et al. 2014 PNAS 2014 Screen Shot 2014-11-11 at 6.47.14 AM.png

Abstract

Significance: Here we show that the spatial prevalence of human societies that believe in moralizing high gods can be predicted with a high level of accuracy (91%) from historical, social, and ecological data. Using high-resolution datasets, we systematically estimate the relative effects of resource abundance, ecological risk, cultural diffusion, shared ancestry, and political complexity on the global distribution of beliefs in moralizing high gods. The methods presented in this paper provide a blueprint for how to leverage the increasing wealth of ecological, linguistic, and historical data to understand the forces that have shaped the behavior of our own species.

Abstract: Although ecological forces are known to shape the expression of sociality across a broad range of biological taxa, their role in shaping human behavior is currently disputed. Both comparative and experimental evidence indicate that beliefs in moralizing high gods promote cooperation among humans, a behavioral attribute known to correlate with environmental harshness in nonhuman animals. Here we combine fine-grained bioclimatic data with the latest statistical tools from ecology and the social sciences to evaluate the potential effects of environmental forces, language history, and culture on the global distribution of belief in moralizing high gods (n = 583 societies). After simultaneously accounting for potential nonindependence among societies because of shared ancestry and cultural diffusion, we find that these beliefs are more prevalent among societies that inhabit poorer environments and are more prone to ecological duress. In addition, we find that these beliefs are more likely in politically complex societies that recognize rights to movable property. Overall, our multimodel inference approach predicts the global distribution of beliefs in moralizing high gods with an accuracy of 91%, and estimates the relative importance of different potential mechanisms by which this spatial pattern may have arisen. The emerging picture is neither one of pure cultural transmission nor of simple ecological determinism, but rather a complex mixture of social, cultural, and environmental influences. Our methods and findings provide a blueprint for how the increasing wealth of ecological, linguistic, and historical data can be leveraged to understand the forces that have shaped the behavior of our own species.

religion, cultural evolution, environmental effects, ecological risk, supernatural beliefs

Reviews Physics.org

Physics.org

The belief in moral, high gods may be advantageous because it fosters cooperative behavior, especially in harsh environments. Just as physical adaptations help populations prosper in inhospitable habitats, belief in moralizing, high gods might be similarly advantageous for human cultures in poorer environments. A new study from the National Evolutionary Synthesis Center (NESCent) suggests that societies with less access to food and water are more likely to believe in these types of deities.

"When life is tough or when it's uncertain, people believe in big gods," says Russell Gray, a professor at the University of Auckland and a founding director of the Max Planck Institute for History and the Sciences in Jena, Germany. "Prosocial behavior maybe helps people do well in harsh or unpredictable environments." Gray and his coauthors found a strong correlation between belief in high gods who enforce a moral code and other societal characteristics. Political complexity—namely a social hierarchy beyond the local community— and the practice of animal husbandry were both strongly associated with a belief in moralizing gods. The emergence of religion has long been explained as a result of either culture or environmental factors but not both. The new findings imply that complex practices and characteristics thought to be exclusive to humans arise from a medley of ecological, historical, and cultural variables. "When researchers discuss the forces that shaped human history, there is considerable disagreement as to whether our behavior is primarily determined by culture or by the environment," says primary author Carlos Botero, a researcher at the Initiative for Biological Complexity at North Carolina State University. "We wanted to throw away all preconceived notions regarding these processes and look at all the potential drivers together to see how different aspects of the human experience may have contributed to the behavioral patterns we see today." Of gods and men

This figure illustrates the distribution of societies that believe in moralizing, high gods (blue) and those that do not (red). Light gray shading indicates lower potential for plant growth with the darker areas signifying high potential. Credit: Carlos Botero

The paper, which is now available online, will be published in an upcoming issue of the Proceedings of the National Academies of Science. To study variables associated with the environment, history, and culture, the research team included experts in biology, ecology, linguistics, anthropology, and even religious studies. The senior author, Gray, studies the intersection of psychology and linguistics, while Botero, an evolutionary ecologist, has examined coordinated behaviors in birds.

This study began with a NESCent working group that explored the evolution of human cultures. On a whim, Botero plotted ethnographic data of societies that believe in moralizing, high gods and found that their global distribution is quite similar to a map of cooperative breeding in birds. The parallels between the two suggested that ecological factors must play a part. Furthermore, recent research has supported a connection between a belief in moralizing gods and group cooperation. However, prior to this study, evidence supporting a relationship between such beliefs and the environment was elusive.

Reviews Business news

http://jewishbusinessnews.com/2014/11/10/evolutionary-synthesis-center-societies-living-in-harsh-environments-more-likely-to-believe-in-moralizing-gods/

Just as physical adaptations help populations prosper in inhospitable habitats, belief in moralizing, high gods might be similarly advantageous for human cultures in poorer environments. A new study from the National Evolutionary Synthesis Center (NESCent) suggests that societies with less access to food and water are more likely to believe in these types of deities.

“When life is tough or when it’s uncertain, people believe in big gods,” says Russell Gray, a professor at the University of Auckland and a founding director of the Max Planck Institute for History and the Sciences in Jena, Germany. “Prosocial behavior maybe helps people do well in harsh or unpredictable environments.”

Gray and his coauthors found a strong correlation between belief in high gods who enforce a moral code and other societal characteristics. Political complexity–namely a social hierarchy beyond the local community– and the practice of animal husbandry were both strongly associated with a belief in moralizing gods.

The emergence of religion has long been explained as a result of either culture or environmental factors but not both. The new findings imply that complex practices and characteristics thought to be exclusive to humans arise from a medley of ecological, historical, and cultural variables.


“When researchers discuss the forces that shaped human history, there is considerable disagreement as to whether our behavior is primarily determined by culture or by the environment,” says primary author Carlos Botero, a researcher at the Initiative for Biological Complexity at North Carolina State University. “We wanted to throw away all preconceived notions regarding these processes and look at all the potential drivers together to see how different aspects of the human experience may have contributed to the behavioral patterns we see today.”

The paper, which is now available online, will be published in an upcoming issue of the Proceedings of the National Academies of Science. To study variables associated with the environment, history, and culture, the research team included experts in biology, ecology, linguistics, anthropology, and even religious studies. The senior author, Gray, studies the intersection of psychology and linguistics, while Botero, an evolutionary ecologist, has examined coordinated behaviors in birds.

This study began with a NESCent working group that explored the evolution of human cultures. On a whim, Botero plotted ethnographic data of societies that believe in moralizing, high gods and found that their global distribution is quite similar to a map of cooperative breeding in birds. The parallels between the two suggested that ecological factors must play a part. Furthermore, recent research has supported a connection between a belief in moralizing gods and group cooperation. However, prior to this study, evidence supporting a relationship between such beliefs and the environment was elusive.

“A lot of evolutionists have been busy trying to bang religion on the head. I think the challenge is to explain it,” Gray says.

“Although some aspects of religion appear maladaptive, the near universal prevalence of religion suggests that there’s got to be some adaptive value and by looking at how these things vary ecologically, we get some insight.”

Botero, Gray, and their coauthors used historical, social, and ecological data for 583 societies to illustrate the multifaceted relationship between belief in moralizing, high gods and external variables. Whereas previous research relied on rough estimates of ecological conditions, this study used high-resolution global datasets for variables like plant growth, precipitation, and temperature. The team also mined the Ethnographic Atlas– an electronic database of more than a thousand societies from the 20th century– for geographic coordinates and sociological data including the presence of religious beliefs, agriculture, and animal husbandry.

Comment by DRW: 20th Century and earlier.

“The goal became not just to look at the ecological variables, but to look at the whole thing. Once we accounted for as many other factors as we could, we wanted to see if we could still detect an environmental effect,” Botero says. “The overall picture is that these beliefs are ultimately shaped by a combination of historical, ecological, and social factors.”

Botero believes that this study is just the tip of the iceberg in examining human behavior from a cross-disciplinary standpoint. The team plans to further this study by exploring the processes that have influenced the evolution of other human behaviors including taboos, circumcision, and the modification of natural habitats.

“We are at an unprecedented time in history,” Botero says. “Now we’re able to harness both data and a combination of multidisciplinary expertise to explore these kinds of questions in an empirical way.”

NC State University News

http://news.ncsu.edu/2014/11/religion-ecology/ "Global distribution of societies that exhibit beliefs in moralizing high gods (blue) or not (i.e., non-moralizing deities or atheism in red). The underlying map depicts the mean values of net primary productivity in gray scale. Darker localities reflect places with greater potential for overall plant growth." 11/10/2014 should have said "or no high gods"

Explanation of method

Dec 12 2014 9:24AM Carlos

Dear Doug and Anthon,

Thank you for your messages. I wanted to let you know that all the data in the paper on religious beliefs as well as values for all other societies in the EA and Binford databases that were not included in that paper will soon be available for free in a new community resource called ‘D-PLACE’ (short for "Database of Peoples, Languages, Cultures, and Ecology”; future URL: www.dplace.org). Please note that the variables ‘Resource Abundance’ and ‘Climate Stability’ in the PNAS paper are simply composite (unit-less) variables derived from principal components analysis and that the ’true’ data comes from a suite of environmental variables including the mean, variance and predictability of temperature and precipitation cycles as well as measurements of net primary production and the richness of vertebrate species and vascular plants (as we explain in the methods for the PNAS paper). D-PLACE will aggregate all of these raw environmental variables with ethnographic data from various sources including the EA. It will also house cultural phylogenies for different language families (and eventually for the entire set of societies). Our hope is that the kind of analysis we presented in the religious beliefs paper will be easily performed by anyone in the community and that the analyses we have (and will) publish in our papers will be easily replicated by students or other researchers that wish to do so.

In that sense, I also want you to know that I have written an R routine that easily replicates the analyses we ran for the PNAS paper on new data sets. I’m calling it ‘MuMACE’ for ‘Multi Model Analysis of Cultural Evolution’ and will be distributing it with upcoming papers that our group will submit very soon. I am hoping to also release these routines as an R package that extends MuMACE to allow for multinomial regression and for the inclusion of covariance structures from an actual phylogeny.

By the way… I am also happy for you guys to include any and all variables from the PNAS paper in the databases you maintain so please let me know what exactly you still need from us to do so and I will happily oblige (e.g., Anthon’s message indicates that our "agriculture variable isn't quite right”… could you please elaborate? If there are any problems I will speak to our data curator to make sure that they are corrected immediately.

Thanks again for your messages. I hope that this is the first of many… would love to consider follow ups and collaboration on potential new projects with you two!

All the best,

Carlos

Dec 12 2014 11:36AM Carlos

Dear Doug,

The dichotomy we used was indeed category 1 v. cats. 2 to 7 combined (i.e., absence v. presence of any form of animal husbandry). I assume that you ran a single model with all predictors and that you mean that in that particular case Animal Husbandry is not significant… is that right? Please remember that we ran not only that fully parameterized model but also all other possible models with all the different parameter combinations in order to account for model uncertainty. It turns out that, as you found, Animal husbandry is significant in some of these models and not significant in others. Just for kicks try simply removing political complexity from the full list of predictors and you’ll see that AH becomes significant. In a multimodel context — and by that I mean when you consider the effects of each predictor across all models in the candidate set and weigh them by the strength of evidence for each of these candidate models— it is the case that Animal husbandry indeed has a meaningful effect (i.e., the posterior distribution of its estimate does not include zero).

I hope this helps…

C

Dec 12 2014 11:39AM Carlos

Quick clarification… NONE of the Binford data were used for the PNAS paper or have been used in any of our papers so far (hence we have only acknowledge what we have used).

Dec 12 2014 11:56AM Carlos

Sorry other quick clarifications:

On Dec 12, 2014, at 1:50 PM, Doug White <douglas.white@uci.edu> wrote:

Dont you think we should ask him now for the ’true’ data from his "suite of environmental variables including the mean, variance and predictability of temperature and precipitation cycles as well as measurements of net primary production and the richness of vertebrate species and vascular plants" (explained in the methods for the PNAS paper)?

Happy to provide that… will do so later today… C

Dec 12 2014 12:07PM Carlos

Dear Kate (and Mike),

I’ve been approached by Doug White and his team of colleagues re our paper in PNAS. One of the issues we’ve discussed are the data we included in our analysis. I've explained that we will soon be releasing a new and more comprehensive database called dplace and that it will include ethnographic data from the EA and the Binford databases. I've explained that we did not use Binford for the PNAS paper... would you mind further clarifying the use and provenance of the Binford data in dplace? Thanks,

C

Dec 12 2014 12:47PM Doug

Dear Carlos,

Of course I know you did not use Binford data in the PNAS article, you mentioned it only in an email. For the future, tho, several in our group have put in effort to build out the Binford data on our sites, similarly for the much expanded SCCS. So those people would appreciate acknowledgement of their articles or construction of data sources from our projects. Your group cited Brown and Eff in PNAS but you need broader acknowledgements of the series of articles e.g., such as might be chosen from http://intersci.ss.uci.edu/wiki/index.php/Dow-Eff_References or elsewhere including the the Dow-Eff development of what went into the DEf software. Apart from EA, which was simply corrected by Pat Gray, as you cited, datasets developed or expanded like the Binford data -- involving myself, Anton, and the most extensive final version as well as 2001 version completion by Amber Johnson, programmer, PhD student and academic colleague of Binford -- are discussed in my initial draft of the introduction (attached) to the Wiley Companion that we are editing. There we'd also appreciate a chapter from you! I recently invited Andrey Korotayev who coded the Christianity and Islam codes for the SCCS, to code in addition the non-SCCS societies of the EA that are in your PNAS sample. That would make it clear that most of the high or moral god societies in the EA sample are either deep or superficially Christian or Muslim (6% exceptions in the SCCS), which would be good to include to your EA model. One of your PNAS reviewers (at NCSU it happens) had the mistaken idea that some of the non high god societies were atheists, which is untrue but also not a fair representation of the indigenous peoples included in coded cross-cultural datasets from ethnographic reports. Cross-cultural research should not be in the business of spreading misunderstanding -- which is why we need eventually to merge the SCCS dataset with the EA dataset so that the cultural information is much broader. That's not easy to do but Eff may be working on that once his problem of sufficient covariates is solved. Or maybe you have those skills in merging datasets that Anthon also has. What do you do with missing data? I think thats a probably the biggest problem for Anthon when there might be 748 or so EA societies and only 170 SCCS societies coded for high gods.

Anyway dont worry about cross-citations and acknowledgements, that will all work itself out, we all have the same goals and I'm very happy to learn of your projects and want to be supportive of them and your group.Dear Carlos,

Of course I know you did not use Binford data in the PNAS article, you mentioned it only in an email. For the future, tho, several in our group have put in effort to build out the Binford data on our sites, similarly for the much expanded SCCS. So those people would appreciate acknowledgement of their articles or construction of data sources from our projects. Your group cited Brown and Eff in PNAS but you need broader acknowledgements of the series of articles e.g., such as might be chosen from http://intersci.ss.uci.edu/wiki/index.php/Dow-Eff_References or elsewhere including the the Dow-Eff development of what went into the DEf software. Apart from EA, which was simply corrected by Pat Gray, as you cited, datasets developed or expanded like the Binford data -- involving myself, Anton, and the most extensive final version as well as 2001 version completion by Amber Johnson, programmer, PhD student and academic colleague of Binford -- are discussed in my initial draft of the introduction (attached) to the Wiley Companion that we are editing. There we'd also appreciate a chapter from you! I recently invited Andrey Korotayev who coded the Christianity and Islam codes for the SCCS, to code in addition the non-SCCS societies of the EA that are in your PNAS sample. That would make it clear that most of the high or moral god societies in the EA sample are either deep or superficially Christian or Muslim (6% exceptions in the SCCS), which would be good to include to your EA model. One of your PNAS reviewers (at NCSU it happens) had the mistaken idea that some of the non high god societies were atheists, which is untrue but also not a fair representation of the indigenous peoples included in coded cross-cultural datasets from ethnographic reports. Cross-cultural research should not be in the business of spreading misunderstanding -- which is why we need eventually to merge the SCCS dataset with the EA dataset so that the cultural information is much broader. That's not easy to do but Eff may be working on that once his problem of sufficient covariates is solved. Or maybe you have those skills in merging datasets that Anthon also has. What do you do with missing data? I think thats a probably the biggest problem for Anthon when there might be 748 or so EA societies and only 170 SCCS societies coded for high gods.

Anyway dont worry about cross-citations and acknowledgements, that will all work itself out, we all have the same goals and I'm very happy to learn of your projects and want to be supportive of them and your group.

Dec 12 2014 11:26PM Carlos

Dec 12 2014 12:07PM Carlos

Dear Kate (and Mike),

I’ve been approached by Doug White and his team of colleagues re our paper in PNAS. One of the issues we’ve discussed are the data we included in our analysis. I've explained that we will soon be releasing a new and more comprehensive database called dplace and that it will include ethnographic data from the EA and the Binford databases. I've explained that we did not use Binford for the PNAS paper... would you mind further clarifying the use and provenance of the Binford data in dplace? Thanks,

C

Dec 12 2014 12:47PM Doug

Dear Carlos,

Of course I know you did not use Binford data in the PNAS article, you mentioned it only in an email. For the future, tho, several in our group have put in effort to build out the Binford data on our sites, similarly for the much expanded SCCS. So those people would appreciate acknowledgement of their articles or construction of data sources from our projects. Your group cited Brown and Eff in PNAS but you need broader acknowledgements of the series of articles e.g., such as might be chosen from http://intersci.ss.uci.edu/wiki/index.php/Dow-Eff_References or elsewhere including the the Dow-Eff development of what went into the DEf software. Apart from EA, which was simply corrected by Pat Gray, as you cited, datasets developed or expanded like the Binford data -- involving myself, Anton, and the most extensive final version as well as 2001 version completion by Amber Johnson, programmer, PhD student and academic colleague of Binford -- are discussed in my initial draft of the introduction (attached) to the Wiley Companion that we are editing. There we'd also appreciate a chapter from you! I recently invited Andrey Korotayev who coded the Christianity and Islam codes for the SCCS, to code in addition the non-SCCS societies of the EA that are in your PNAS sample. That would make it clear that most of the high or moral god societies in the EA sample are either deep or superficially Christian or Muslim (6% exceptions in the SCCS), which would be good to include to your EA model. One of your PNAS reviewers (at NCSU it happens) had the mistaken idea that some of the non high god societies were atheists, which is untrue but also not a fair representation of the indigenous peoples included in coded cross-cultural datasets from ethnographic reports. Cross-cultural research should not be in the business of spreading misunderstanding -- which is why we need eventually to merge the SCCS dataset with the EA dataset so that the cultural information is much broader. That's not easy to do but Eff may be working on that once his problem of sufficient covariates is solved. Or maybe you have those skills in merging datasets that Anthon also has. What do you do with missing data? I think thats a probably the biggest problem for Anthon when there might be 748 or so EA societies and only 170 SCCS societies coded for high gods.

Anyway dont worry about cross-citations and acknowledgements, that will all work itself out, we all have the same goals and I'm very happy to learn of your projects and want to be supportive of them and your group.

Dec 12 2014 12:51PM Carlos

Dear all,

As promised, I am attaching a csv file with the entire set of ecological variables used in our analysis. I am also including information on units of measure and sources for each variable below. Very soon we will make available to everyone these same environmental variables (plus a few more) for all the EA and Binford societies. All the best,

C

Precipitation and temperature:

  EntireDataset....csv

Dec 12 2014 12:54PM Doug

Its not a problem. I know that. I discussed future uses of the Binford dataset with Carlos a few minutes ago, by email. Am delighted about development of D-PLACE and all your group is doing.

Dec 12 2014 1:06PM Carlos

Dear Doug…

Yes of course!! I and, in fact ALL of the people involved in the development of DPLACE, are very aware that we stand in the shoulders of giants and that none of these work would be possible without the fantastic work that has been done before us. We really do wish to make every effort to cite and acknowledge all sources. Please excuse any omissions that you feel we did in the PNAS paper (we truly thought we were citing everything we needed for that particular paper). FYI: Kate Kirby is the point person in charge of curating and developing the ethnographic side of DPLACE’s data and I am copying her here to make sure that she is aware of this. I think she is away for a few days but I’m sure she will get respond as soon as she’s back. Also, please know that Michael Gavin (also cc’ed) is leading the development of the database itself (along with Russell Gray) and that I am simply providing ecological data and expertise in spatial/statistical analyses...

As for the chapter you mention… how can I/we help? What exactly do you have in mind and what would be the time frame for this?

C ps. re these comments/questions:

> One of your PNAS reviewers (at NCSU it happens) had the mistaken idea that some of the non high god societies were atheists, which is untrue but also not a fair representation of the indigenous peoples included in coded cross-cultural datasets from ethnographic reports.

Sorry about that… I thought we had explicitly corrected this potential misunderstanding by stating in the legend for figure 1 that we were distinguishing between "societies that exhibit beliefs in moralizing high gods (blue) or not (i.e., atheism or beliefs in nonmoralizing deities or spirits in red).”

> Or maybe you have those skills in merging datasets that Anthon also has. What do you do with missing data? I think thats a probably the biggest problem for Anthon when there might be 748 or so EA societies and only 170 SCCS societies coded for high gods.

So far we have simply eliminated missing data from all of our analyses. There are many techniques for data ‘imputation’ but I’ve preferred to stay away from those since we've enough samples at the global level to test the questions we are pursuing…

Cheers,

C

Dec 12 2014 1:14PM Doug

That's great, Carlos, very much appreciated and important. Thank you.

Carlos and Anthon - maybe you two can work out the appropriate way to report the R squares -- the autocorrelation R sq, the R sq for the model once autocorrelation is controlled for and -- does it make sense to report a total R sq? Anyway having two independent ways in respective R programs to run models is outstanding we just need to work out ways to compare them by the comparable methods.

Muy altos saludos ambos, a los dos.

Dec 12 2014 3:59PM Doug

Lets see whether Andrey Korotayev wants to do the coding of Christian and Muslim societies from the EA, it may not be difficult because he probably has a map of the distributions of religions that apply to your EA v34 coded societies excluding SCCS cases and there are probably not so many more than he coded for the SCCS. If so you could do a joint article, your team and him, but have recommended that Anthon or Malcolm contact you on installing an R package that would do imputation in R quite quickly and we can give you the same dataset that we created that forms the basic for the imputation. Then the two approaches should give fairly similar results except for how the autocorrelation control is done, your methods and ours may differ within certain bounds of comparability and the questions about the best methods to use are still open. My ideal improvement would be whether contiguous ecological patches, as in Binford's work, could be used in controls for autocorrelation alongside distance and language phylogeny where your phylogenies probably give better (and more interpretable) results than Eff's language distances. I have a team of three of his students working on our own controls for language phylogeny as trees (not language distance) and then finding the best phylogenetic placement of the variables in a given model in different parts of the world language phylogenetic structure.

If Korotayev doesnt elect do the Christian and Muslim coding, then: Could your chapter be worked out by your own team or would a better chapter be worked out between your team of Eff's, including his students working on the idea that I posed to them. I would not need to be a coauthor in any case but I would propose some variant of the options given above, and I'm cc:ing Anthon on this too so you could talk it out and accomodate Korotayev's coding input (he wrote a book on Religion from the SCCS sample data that he coded and from other material).

Dec 15 2014 7:16AM Kate

Hi Doug and Anthon,

I've been leading the compilation of the cultural data for D-PLACE. The Binford data we have been working on incorporating are based on my own scans of the tables in Binford's 2001 book, which I digitally translated and compiled. However, as you know, many of the variables in those tables are poorly defined in the book. So, I have been working (on a very part-time basis) on trying to resolve what they are.

Binford and Johnson's ENVCALC2 database (2006) has provided me with some additional definitions (and also focal year data, which I wasn't able to find in the book). Many of my searches have led me to your published papers and website, which I have found to provide some of the clearest explanations available on the history of major cross-cultural datasets and the variables they contain. However, in the case of the Binford dataset, I have not included in d-place any variable definitions that are not published and citable.

In addition to the Ethnographic Atlas and Binford datasets, Bill Divale sent me Jorgensen’s WNAI and SCCS datasets on cd-rom. So, I have also cross-referenced these with the above.

We have designed d-place so that data sources (including sources for codes, codebooks, bibliographies, and corrections) are clearly referenced. The dplace database is also meant to be expandable, so that as additional variables and societies become available for the different datasets, we can potentially include them, linking as always to the published source.

Please let us know if you have any concerns about how the Binford-related data will be presented on d-place and in the paper that introduces it. We plan to launch the database in the middle of 2015 and want to ensure that we address any concerns before the paper is published and the database goes public.

More generally, I am happy to have made contact. I have a great respect for your contributions to cross-cultural research and methods, and hope we’ll cross paths at some point in the future.

Best regards,

Kate Kirby

Dec 15 2014 10:40AM Doug

Thanks for this, Kate. I am adding Amber Johnson to the email list. The datasets that Anthon, Amber and I have collaborated on to provide online access are SCCS and LRB (foragers) at http://intersci.ss.uci.edu/wiki/index.php/Main_Page; these have been massively updated from the Divale distributions that we used to update and distribute the World Cultures eJournal and CD.

Our project has been running for 2.5 years accompanied by a draft in process of a Wiley Companion to Cross-Cultural Research which you already know from Carlos. WNAI and our EA were updated by Anthon to include standard language data and global ecological data. Otherwise they had not changed from the World Cultures eJournal version distributed by Divale -- until Carlos sent us the data for EA which Anthon has started to integrate (only abundance and stability so far).

Amber Johnson is writing a book of results from data in the Binford book and her/our new database. You should know: Amber was Binford's RA, probably a TA, also the project programmer and integrator of the environmental data, and did all that recoding in the book. The publisher asked that it be shortene and a different PhD student of Binford was given the task (Amber by 2001 had taken a faculty job at Truman State). That student gutted the appendix which is why it was so hard for readers to get clear versions of the variable definitions, etc. I invited Amber 2 years ago to our Santa Fe Institute working group on causal analysis for CCResearch focusing then on compilation of the integration of data with the Dow-Eff R functions for our Cross-Cultural datasets and solutions for autocorrelation as well as imputation. You should also know that Amber at Truman State recruited Binford into her faculty is his last years; she became Chair of Anthro at some point and at some point were husband and wife.

RE: imputation - you should definitely try to implement MD and MID (e.g., http://intersci.ss.uci.edu/wiki/index.php/Imputing_data_for_Regression_Analysis - http://intersci.ss.uci.edu/wiki/index.php/Donald_B._Rubin). We use MID, which applies to imputing variables for those societies that are coded for the dependent variable. Without imputation you will really have alot of unknown biases affecting your results. This is also a serious omission because results from the same model using DEf functions (and Galaxy CoSSci which give the same results) our teams will not be able to reproduce the same results, or know where the differences come from since you use different routines for analysis: that is 1) in the data (which of your societies are omitted); 2) the methods of control for autocorrelation (which will be useful to compare), and 3) your software code for analysis. Item (1) here is very serious and can be fixed by using the standard MID R library (with the alternative being the standard MD R library). There is lots of missing data in all the cross-cultural datasets, so fixing this should have high priority. All the major national databases use these standard Rubin (1976, 1987) methods. This is not the place to be sloppy about missing data if we want to have CCR emerge as a strong science.

Back to Amber and LRB:

Dec 15 2014 12:57PM Anthon

Hi Kate and Carlos. Your plans sound great. It looks like we can probably help you a bit. Our materials are found here: http://capone.mtsu.edu/eaeff/Dow-Eff%20functions.html --including codebooks, examples, and a manual.

Our data and functions are kept in an R workspace. You should use our version of the data, since Doug, Pat Gray, and I have made many amendments and corrections--and these are continuing. The workspace can be loaded with this command:

load(url("http://capone.mtsu.edu/eaeff/downloads/DEf01f.Rdata"))

Like yourselves, we have used GIS to add many variables. Our coordinates have been significantly rectified, so you will want to use the coordinates found in our data. The attached Rdata file contains a single data.frame called p. It contains the coordinates and the GIS-derived data for all 1461 societies in the current version of our data (the variables are described in the codebooks). If you use these coordinates, your data will be available for all four of the ethnological datasets: EA, SCCS, LRB, and WNAI.

The data.frame p does not include a significant number of new GIS variables that will appear in the next version of our workspace--for example, we have already produced variables for CRU TS 3.22--so we might communicate to make sure that there is no duplication of effort.

I'm particularly eager to get the biodiversity variables--I could not find geo-referenced rasters for the Jenkins et al. work or the Kreft and Jetz work. Would you have time to rerun them with the new coordinates? Or, alternatively, perhaps you could send me a link to the geo-referenced data?

best wishes, --Anthon

Dec 16 2014 1:32PM Carlos

Dear Anthon,

Wow… I wish we had known you guys had revised the coordinates in these databases. We spent a lot of time and money doing the same for the EA. In any case, I can easily extract biodiversity data from the new lat/longs but I am sending you instead the actual data rasters because I know you will appreciate having them at hand in case coordinates change or new societies are added in the future. The Jenkins et al. rasters I got directly from Clinton (a friend of mine) and the Kreft and Jetz I downloaded from data supplements to the paper. I hope this helps,

Carlos

Biodiversity.Rasters.Rdata

Dec 17 2014 1:32PM Kate

Thanks Doug, Amber and Anthon for your replies and for the information and links you sent. Amber, I'm very happy to make contact and I'm looking forward to talking more.

I haven't yet had a chance to go through the links you sent in detail, but will write again once I've done this, with thoughts on areas in which we may be able to contribute, and no doubt with a few questions as well!

So far we have not tried to impute missing data in any of our analyses, and our plan is to leave these as missing values in DPLACE. However, it will be great to be able to refer to your approach if at some point we use analytical methods that require imputation.

I'll write again soon but wanted to get this message out first. I'm looking forward to ongoing discussion,

Kate

xxx

xxx

xxx

xxx

xxx

xxx

xxx

xxx

Dec 17 2014 1:32PM Kate

xxx

xxx

xxx

Mar 3 2015 1:32PM Kate

Dear Dr. White,

I have now read the manuscript section you sent and shared it with my co-authors. After careful debate and consideration, we find it very unfortunate that you continue to mischaracterize our study and criticize it based on an improper understanding of our methods. Taking into account that we have already pointed out these mistakes recently (and shared with you both data and algorithms to properly replicate our analyses), we believe it is no longer productive to continue this discussion by email. Let’s continue this debate in print. Thanks,

Carlos

Mar 3 2015 6:59AM From Doug

Dear Carlos,

Yes, I'd like to resolve this too. I checked http://cabotero.weebly.com/publications.html but perhaps you can steer me to where you pointed out these mistakes as I thought you would. What I dont have are the independent variables for distance and language and the algorithms, so that the analysis can be replicated.

Thanks, this would be very helpful. I am very pro what you are doing and have appreciated your sharing most of the variables.

Best regards,

Douglas R. White

Mar 3 2015 9:29AM Explanation from Carlos

Doug: Re Dec 12 2014

The coordinates identifying the location of each society in our study are publicly available and provided in the data set that accompanies our paper and the dataset I sent directly to you in 12 Dec 14 (see third and fourth columns). You can obtain pairwise distances between societies by simply computing the greater circle distance between their geographic coordinates. Please note that, as I explained in earlier correspondence to you (and in the PNAS paper), we did not use distance itself as a covariate in our analysis. We used geographic distance simply to identify the 10 nearest neighbors to each society from which we then computed an average neighborhood belief score (which was the actual covariate used as a proxy to test for a potential effect of spatial diffusion).

As for the language variable you say you don’t have, I will refer you again to the dataset we have already provided (column 2). In the absence of a complete language phylogeny (i.e., a cultural phylogenetic tree analogous to the molecular phylogenies we use in biology) we explain in the paper that the next best thing to test for vertical transmission is to use language family as a random effect in our models (which is what we did).

Thus, we explicitly addressed two different versions of Galton’s problem by distinguishing between and accounting for potential dependencies among societies that are created by the (a) vertical and (b) horizontal transmission of cultural traits. I’ll add to this that we believe just as strongly as you do that exogeneity is important to consider, and that we checked that the residuals of our models were not spatially autocorrelated.

One more thing… you continue to miss a point that I've already made before: we did not use R-square in our paper or our analysis. We used instead the area under the curve of receiver operating characteristic curves (AUC ROC) to assess the predictive ability of our models. AUC values are conceptually and numerically different to R-square and it is simply incorrect to compare them directly.

I want to end this conversation by saying that I appreciate your investment in this topic and your desire to find the most appropriate methodology to address these kinds of questions. That is our goal too and I believe our team has demonstrated to you that we are willing to engage in an open and constructive discussion of ideas. However, please realize that your prior correspondence to me (one of the most junior members of the D-PLACE team) has at times come across as abrasive and bully-ish. We may disagree on the utility of our methods, but I resent the implication that we have committed any type of misconduct. I stand absolutely confident in the knowledge that my team and I have maintained the highest standards of ethics and scholarship, that we are giving to this our honest best effort, and that we have been and always will be transparent about our work.

Good day,

Carlos

sent directly to you 12 Dec 14 (see third and fourth columns).

From Anthon 12:57PM

Hi Kate and Carlos. Your plans sound great. It looks like we can probably help you a bit. Our materials are found here: http://capone.mtsu.edu/eaeff/Dow-Eff%20functions.html --including codebooks, examples, and a manual.

Our data and functions are kept in an R workspace. You should use our version of the data, since Doug, Pat Gray, and I have made many amendments and corrections--and these are continuing. The workspace can be loaded with this command:

load(url("http://capone.mtsu.edu/eaeff/downloads/DEf01f.Rdata"))

Like yourselves, we have used GIS to add many variables. Our coordinates have been significantly rectified, so you will want to use the coordinates found in our data. The attached Rdata file contains a single data.frame called p. It contains the coordinates and the GIS-derived data for all 1461 societies in the current version of our data (the variables are described in the codebooks). If you use these coordinates, your data will be available for all four of the ethnological datasets: EA, SCCS, LRB, and WNAI.

The data.frame p does not include a significant number of new GIS variables that will appear in the next version of our workspace--for example, we have already produced variables for CRU TS 3.22--so we might communicate to make sure that there is no duplication of effort.

I'm particularly eager to get the biodiversity variables--I could not find geo-referenced rasters for the Jenkins et al. work or the Kreft and Jetz work. Would you have time to rerun them with the new coordinates? Or, alternatively, perhaps you could send me a link to the geo-referenced data?

best wishes, --Anthon

1:57PM from Carlos

Dear Anthon,

Wow… I wish we had known you guys had revised the coordinates in these databases. We spent a lot of time and money doing the same for the EA. In any case, I can easily extract biodiversity data from the new lat/longs but I am sending you instead the actual data rasters because I know you will appreciate having them at hand in case coordinates change or new societies are added in the future. The Jenkins et al. rasters I got directly from Clinton (a friend of mine) and the Kreft and Jetz I downloaded from data supplements to the paper. I hope this helps,

Carlos

> On Dec 16, 2014, at 3:57 PM, Anthon Eff <Anthon.Eff@mtsu.edu> wrote:

I'm particularly eager to get the biodiversity variables--I could not find geo-referenced rasters for the Jenkins et al. work or the Kreft and Jetz work. Would you have time to rerun them with the new coordinates? Or, alternatively, perhaps you could send me a link to the geo-referenced data?

Eff and All

https://www.google.com/search?client=safari&rls=en&q=A+study+from+NEScent+suggests+that+societies+with+less+access+to+food+and+water+are+more+likely+to+believe+in+these+types+of+deities.&ie=UTF-8&oe=UTF-8

Doug: Thanks for sending me the EA data, and I will add it soon.

  • The SCCS.html file is created using markdown, in Rstudio. If you take a look, the text wraps (resize the browser window and you will see the number of lines change), so I can’t place # at the beginning of each line. But I can put a link in the page to a script that can be downloaded and run, and will do that as soon as I can.
  • Will take a close look at Botero soon.

Anthon


From: Doug White [1]
Sent: Friday, November 14, 2014 10:26 AM
To: Anthon Eff; Malcolm
Subject: Fwd: National_Evolutionary_Synthesis_Center / PNAS / paper on Moral Gods / HiGod

Anthon

  • Your http://capone.mtsu.edu/eaeff/DEf_SCCS.html would be a good model for users to run R for high gods if you put a blank followed by # in front of each text paragraph. Only takes a couple of minutes to make the modification. DEf_SCCS as an example has scale construction and other features that allow users to run the full complement of R for DEf01f and make it easier for researchers using your latest R code to copy and paste the code directly into RStudio, especially if you put a blank followed by # in front of each text paragraph. That way its easier to see that all the code is running correctly.

Most will not have the SCCS book collection as does SAR. I think its good to have a comparison of your http://capone.mtsu.edu/eaeff/DEf_SCCS.html for high gods and the PNAS Carlos Botero model that I use as an example. I sent you his codes for EA used in his model. Would it be difficult to add those needed to your EA dataset to make a direct comparison of DEf results and his results using his own methods of evaluating autocorrelation?

  • The SAR type of posting might help to disseminate information about DEf usage in R. Your suggestions would be welcome. Next I will do something similar for posting at SFI.

Thanks, Doug