- PhDs and Instructors: http://intersci.ss.uci.edu/wiki/pdf/Results_Grueter2.pdf <-- this (v768) is D R White's first instructive help for a publishable article (e.g., in Structure and Dynamics). I am willing to tutor PhDs and Instructors in the use of Galaxy CoSSci. Final model at v1649
"v768 (No) Conflict between Communities of the Same Society" depvar Cyril C. Grueter (No) Conflict between Communities of the Same Society mkdummy("v69", 3) creates a dichotomy for virilocal or patrilocall n=108 out of SCCS 185 coded. V72 INTERCOMMUNITY MARRIAGE (tried, not significant)
- v1649 depvar substituted for v768
- v1,v72,v779 Initial indepvars
- v1 (drop v72,779) add v1649 as Additional - doesn't work - your variables plus Ember and Ember (1971) let to choosing viri/patri locality
- add v215.d7 Virilocal only v63.d3 Virilocal or Patrilocal
- finally: Add v774 ext war missing data --> substitute v892 attacking v893 attacked v1650 ember ext war
coef stdcoef VIF relimp hpcval significance v69.d3 1.594 0.11988 1.37468 0.02609 0.21768 0.2020 Virilocal or Patrilocal v215.d7 1.38682 0.09451 1.36555 0.00625 0.32882 0.2994 Virilocal To Try v69.d3:Wy was listed in the early Galaxy csv output.
- internal warfare and patrilocal residence was postulated by Ember and Ember (1971), and in two independent samples, positive associations were found. ...
- William Divale, Frosine Chamberis, Deborah Gangloff. War, Peace, and Marital Residence in Pre-Industrial Societies. Journal of Conflict Resolution March 1976 vol. 20 no. 1 57-78. Abstract:
- Tefft and Reinhardt recently reported that peacemaking mechanisms and periods of stable peace are associated with internal warfare, while peacemaking mechanisms are absent and peace is unstable in societies with external warfare. We suggest an explanation for these findings based on the argument that in pre-state level societies internal and external war are entirely different types of conflict. Internal war (conflict between communities of the same society) is part of a system of population control in pre-state level societies which also involves female infanticide, polygyny, and patrilocal residence. Internal war is thus a regulatory type of warfare requiring several mechanisms to initiate and inhibit it. In contrast, external warfare among pre-state level societies occurs between societies which have recently migrated to their locales and adopted matrilocal residence. External war is thus an intense struggle for survival between two or more societies trying to occupy the same niche in an ecosystem that cannot support all of them. There can be no regulation of this type of warfare–only victory–and hence there can be no peacemaking mechanisms.
- Turchin, Peter. 2006. Population Dynamics and Internal Warfare: a Reconsideration. Social Evolution and History 5/2: 121–158 (contains an extra figure from the submitted version), with Andrey Korotayev. Discussed in
- I found no results that matched any of the variables in your model, but you can learn from this, at http://intersci.ss.uci.edu/wiki/index.php/V768 where my outcome did no better. These results involved iterating independent variables keeping your v1 (your only initially significant variable in the model except, as I realized later, it was negatively no positively correlated with your v768). Adding v69.d3 Virilocal or Patrilocal as the best measure of the Embers' (1971) finding on internal warfare (v1649 instead of v768), and choosing the best depvar among candidates v768, v1649 and v891 and adding a measure of reliability for v1649 (v1652) , the best measure for internal war. The result was that the coefficient for v1 was negative not positive, the Embers' independent variable of Patri/or/Virilocality was supported (v69.d3), and its reliability was supported by v1652. corr.test(dx$v1,dx$v1649) in RStudio is negative. There was no improvement i.e., possible refinements that are more predictive, when v1 is dropped.
Discussion of a good final result Rsq=0.35
If you try a new model at http://socscicompute.ss.uci.edu click DEf01f Dow Eff, select SCCS, and make sure, as in the example on V768, that you have Dummies that generate your dichotomized variables One Depvar One or More IndepVars Some Independent variables in UNrestricted mode: it doesnt matter which so long as they are valid.
The best codebook for SCCS is at http://capone.mtsu.edu/eaeff/Dow-Eff%20functions.html
Its always better that you use Galaxy CoSSci than ordinary software, even with better strategies for missing data. None will control with autocorrelation. In the model at http://intersci.ss.uci.edu/wiki/index.php/V768 (your original variable, replaced by v1649 for a better depvar), you'll see that there is a Wy term for autocorrelation that is very high and very significant. That will through off your model even if you have completely coded data.
DEf01f R Script
library(mice) library(foreign) library(stringr) library(AER) library(spdep) library(psych) library(geosphere) library(relaimpo) library(linprog) library(dismo) library(forward) library(pastecs) library(classInt) library(maps) library(dismo) library(plyr) library(aod) library(reshape) library(RColorBrewer) library(XML) library(tm) library(mlogit) #The Dow-Eff functions, as well as the four ethnological datasets, are contained in an R-workspace, located in the cloud.
setwd("/Users/drwhite/Documents") load(url("http://capone.mtsu.edu/eaeff/downloads/DEf01f.Rdata")) ls() #-can see the objects contained in DEf01.Rdata #The setDS( xx ) command sets one of the four ethnological datasets as the source for the subsequent analysis. The four valid options for xx are: “WNAI”, “LRB”, “EA”, “SCCS”. The setDS() command creates objects:
# ===list and modify variables for use in model=== # --make new variables-- dx$v768 dpV <- "v768" mkdummy("v72",5) # v72.d5 Conflict between communities # --identify variables to keep for model building-- evm<-c("v768","v72.d5","v1","v768","v774","v676","v621","v626")
smi <- doMI(evm, nimp = 5, maxit = 7) dim(smi) # dimensions of new data frame sm
# ===identify role of variables in model=== # --dependent variable-- dpV <- "v768" oxog <- c("NULL") #-independent variables in UNrestricted model-- UiV<-c("v1","v72.d5","v774","v676","v621") #--independent variables in restricted model--" RiV<-c("v1","v72.d5","v774")
#ecology NOT tested: FULL SET=FALSE #h <- doOLS(smi, depvar = dpV, indpv = UiV, rindpv = RiV, othexog = NULL, dw = TRUE, lw = FALSE, ew = FALSE, stepW = TRUE, boxcox = FALSE, getismat = FALSE, relimp = TRUE, slmtests = FALSE, haustest = NULL, mean.data = TRUE, doboot = 1000, full.set=FALSE) #, full.set=TRUE, full.set=FALSE CSVwrite(h, "IntercommunityMarriage3.noEW", FALSE)
#ecology NOT tested: FULL SET=TRUE #h <- doOLS(smi, depvar = dpV, indpv = UiV, rindpv = RiV, othexog = NULL, dw = TRUE, lw = FALSE, ew = FALSE, stepW = TRUE, boxcox = FALSE, getismat = FALSE, relimp = TRUE, slmtests = FALSE, haustest = NULL, mean.data = TRUE, doboot = 1000, full.set=TRUE) #, full.set=TRUE, full.set=FALSE #CSVwrite(h, "IntercommunityMarriage2.noEW", FALSE)
#ecology NOT tested: h <- doOLS(smi, depvar = dpV, indpv = UiV, rindpv = RiV, othexog = NULL, dw = TRUE, lw = FALSE, ew = FALSE, stepW = TRUE, boxcox = FALSE, getismat = FALSE, relimp = TRUE, slmtests = FALSE, haustest = NULL, mean.data = TRUE, doboot = 1000, full.set=FALSE) #, full.set=TRUE, full.set=FALSE #CSVwrite(h, "IntercommunityMarriage1.noEW", FALSE)
#ecology tested: h <- doOLS(smi, depvar = dpV, indpv = UiV, rindpv = RiV, othexog = NULL, dw = TRUE, lw = TRUE, ew = FALSE, stepW = TRUE, boxcox = FALSE, getismat = FALSE, relimp = TRUE, slmtests = FALSE, haustest = NULL, mean.data = TRUE, doboot = 1000, full.set=FALSE) #, full.set=TRUE, full.set=FALSE #CSVwrite(h, "IntercommunityMarriage1.noEW", FALSE)
you were right. p.s. you started out with v768 v774 which have lots of missing data. Generally better to use more fully coded variables like v1649 and v1650. Reason is that missing data are usually not missing randomly but off kilter re: more fully coded variables.
this xls is what you see in csv output without reorganizing for the kind of models that I rearrange.... my method is then to use screen save of my edit and copy the screen save into word or my wiki.
Now you can rewrite the article.... v1 is negative however. Looks like trade and war dont go together. Would lack of trade promote war? Warfare diminish opportunities for trade? Or come kind of ecological condition. Embers' measure 1685. Chronic resource problems?
Think is, now, you can just enter these variables at http://socscicompute.ss.uci.edu and get results
This is what the original Galaxy csv file looks like
Response to Alvin Wolfe Query and ABQ invitation for Cyril
The GP solution is fairly easy - I have done up to 10 such models in one day. I just did http://socscicompute.ss.uci.edu/u/drwhite/h/v1649-internal-war with an Australian anthropologist, Cyril Grueter Click 10 for the finished model. Then click the diskette image under $stepwise. Or click the upper right, "Import history" Then click "use this history" Then click "10: DEF01f" on the right and you'll see how the variables were arranged for this model. Then click the diskette image under "10: DEF01f" and a Galaxy...csv file will be uploaded to your computer. Its easier at this point to look at http://intersci.ss.uci.edu/wiki/index.php/V768 where you can see the process we went through, with results at the end.
That just gives you an introduction to a finished model thats the basis for an article.
If your interested in this online Galaxy CoSSci site as a way to do cross-cultural modeling correcting for Galton's problem, I'll send more instruction. At the moment complexity and complex networks are more complicated and I'm working right now on this project for a Wiley Companion. Thanks for your note! Hope all goes well for you in future days ... you're a model for me as I'm only 72. Great to see you in ABQ.
Cyril - do you want to present this model at our next ABQ meeting (SCCR) in ABQ in February?
Al Wolfe Query
Doug: I do want to continue to be informed about all these matters -- complexity, complex networks, Galton's problems, etc. I am not sure how much I will be able to do because at age 86, while not losing interest, I am losing some capabilities.
Alvin W. Wolfe Distinguished University Professor Emeritus Department of Anthropology 4202 E. Fowler Avenue, SOC107 University of South Florida Tampa, FL 33620-8100 813-974-0794