Human Social Complexity - World Cultures 2010

From InterSciWiki
Jump to: navigation, search
Map of Indigenous Language families of 186 societies in the SCCS (Standard Cross-Cultural Sample)

Contents

Map of language families of 186 societies in the SCCS

Click images to the right to open the maps: (1) map of Indigenous Language families of societies in the SCCS (upper) or (2) Atlas of Contemporary Religions (lower). The first can be copied under the Creative Commons license.

Map of Indigenous Language families of 186 societies in the SCCS (Standard Cross-Cultural Sample)
Atlas of Contemporary Religions
Note that the data on human societies used for the assignments in this class are not contemporary societies but representative of the earliest speakers of indigenous languages that are well described ethnographically, i.e., Amerindians rather than Euro-, Afro-, and Asian-Americans, for example.

Overview: Resources, Class Objectives, Projects

  • APPLICABILITY: The methods we learn are the most advanced software to date for any problem of survey data analysis, from any study or discipline. They include the treatment of statistical inference, data from observations where the individuals or other units interact with one another and are thus not statistically independent, and the treatment of missing data. These approaches have virtually universal applicability to problems in industry, government, physical, biological, economic and social sciences, and the humanities. Understanding of these methods (finally perfected in 2009) is thus of essential importance to a university education and professional research training.
  • THE CLASS IN SHORT: We are working in this class with one of the great treasures of anthropology, of which I was a founder and am a key custodian: The SCCS coded ethnographic database. I stopped doing studies based on these data in 1990 when our studies made it clear -- see National Science Foundation (NSF) proposal -- that a key analytic problem had not been solved, one that crippled the social sciences and results from the observational sciences based on survey analysis. In 2009 a former student and another colleague of mine solved what has been a massive problem identified in 1889 and never solved in the intervening 130 years: "Galton's" problem of drawing scientific inferences (e.g., about causality) from observed case studies that influence one another through interaction or come from a common origin. This problem is pervasive in the natural sciences, and its solution is of major importance in obtaining major findings about human societies, whatever the field of study (economics, intl. relations, political science, social psychology, disease, sociology, and anthropology). The NSF proposal is about implementing an open access (EduMod) version for use of the new (2SLS) method in classrooms worldwide, and this and the fall 2009 classes are part of the pretest. NSF funding will enable salaries to be paid to qualified undergraduates through REU (Research Experience for Undergraduates) fellowships.
Abstract: You run a powerful regression program that does estimates of the effects of causal factors for human culture variation. You select variables for study from a database of two thousand variables on features of the full range of diverse human cultures (the SCCS). Take the open source page R home with you and the database and software. Receive tutorials to learn these methods, learn from other student projects, and start from where they left off. Present an 10-minute powerpoint on your findings, and participate with the instructor, optionally, in preparing a journal publication. Learn how to understand complex human systems and write about them (the fall 2009 writing course, but not the winter course, has a proposal, draft, and 10 page paper). We build off the
  1. Examples of work: 2009 course.
  • RESOURCES:
Codebook: Coded data for 2000 variables for the sample societies
Coded database: SCCS.Rdata.
GIS mapping: SCCS1 geodatabase (two sample maps at Indep/Depvar list - All - see also White-Veit SCCS Atlas and SCCS Maps in Spss (others available on request)
Software for analysis: Two-stage least squares (2SLS) linear regression (1) Imputing_the_data#Eff_and_Dow (2) Unzipping the Eff-Dow data download
Software templates: Initial choices from 30 Edu-Mod pages
Software package: R open source/open access software (installed in the lab along with the data)
Edu-Mod pages are chosen for use in the classroom or at home.
It is also possible to install R and the data and do analysis from home as well as in the classroom.
  • GENERAL OBJECTIVES:
Understanding, without the requirement of prior courses, of the fundamentals of data analysis using contemporary probabilistic methods of statistical inference and spatiotemporal analysis.
Practice in the usage of these methods and describing results of analyses in student projects.
  • PROJECT OBJECTIVES:
Discovery of the historical and evolutionary linkages among human cultures
Mapping of spatial clustering of cultural variables
Discovery of the causal linkages among cultural features
Similarity and differentiation among language groups
  • STUDENT PROJECTS:
Each student chooses one or several key topics variables from the codebook to investigate. This defines one or several projects.
Each student chooses 1 dependent variable from the codebook for each key topics and justifies their suitability for analysis
For each (1 or 2) project the student chooses from the Edu-Mod files a software template based on the independent (predictor) variables of that template.
This software template is run by inserting the dependent variable name and number into the template, and copying the template into R (the .
  • WORKFILES:
Each student takes one new Edu-Mod workfiles for each (1 or 2) project to:
copy, edit, and save the new software template
copy the edited template into R to run
save the results of successful runs
if you cant get your changes to work, I have csdiff on my computer in the classroom and we can compare a previous successful run with a newly edited unsuccessful run.
  • PRESENTATION OF WORK:
8 Minute powerpoint summaries presented in class.
  • GRADING
5% Class notes and participation
5% Project proposal(s) (3 pages)
25% Powerpoint presentations
5% Project maps
35% Edu-Mod pages for the project(s)
25% Report on project accomplishments
  • Hints for the wiki
  • If you are looking for something, type, for example: SCCS (in the searchbox to the left).
  • Hint on printing these pages: Don't print wiki pages directly or your page will be cluttered with url's for external links. Rather, copy what you want to print (control-A for the whole page), paste into word, and reset the margins and type sizes if needed, then print from word.

Diagram of interactive effects for human societies modeled in the R program in Eff and Dow (2009)

Eff&Dow in CHINESE BOX OF INTERACTIVE EFFECTS2.png

  • Simple cross tab to help select independent variables not already in the program

Using the codebook, you specify a DV (dependent variable) for the 186 societies in the SCCS database. This involves changing two lines of computer code on the wiki page where you copy one of the available Eff and Dow 2009 models. That model contains the IVs (independent variables). The model gives you coefficients and significance for the following:

fydd - the effects (positive coeff.) of neighbors on the DV
fyll - shared history (positive coeff) or differentiation (negative coeff.) from the language family
each IV - positive or negative effect on the DV, in addition to those of the fydd and fyll.

Readings

Core methods

General readings

(under construction)

How to get your studies of variables in Sccs - for your project(s) by author or topics

Sources for codes on the SCCS -- all authors, online links to articles in pdf

Studies of variables in Sccs by topic for variables

see: 2010 Students

after signin FirstNameLast you will have a user name. Use "Edit" to enter your User:FirstNameLast here between double square brackets

2009 student paper/powerpoint contributors

Day One Jan 5

Announce Soc Sci 240B 1.33 credits

Class notes

Review

Overview

Themes of the Course

  • Question: Are we too complex as a species to understand ourselves?
What is this complexity?
Cognitive ability to form and interpret networks of relationships
Social ability to exapt existing social modules and reintergrate (readings from Springer Innovation chapters)
Difference: (How insurgents win - Malcolm Gladwin) --- Competition: How diversity beats expertise (Scott Page) --- Resistence & egalitarianism: (Chris Boehm --- 1993 Egalitarian Behavior and Reverse Dominance Hierarchy. Current Anthropology 34:227-254. [This paper won the Stirling Prize in Psychologica Anthropology])
Prospectus: Human Complexity
How to model complexity?
e.g., Wardrop's principles of equilibrium as random routes through a network of choices to get from A to B --- Swings of Nash equilibrium and the Conservation of Information
What are examples of complex behaviors in human societies?
e.g., trade amplifies peace in the short run but economics amplifies instability and conflict dynamically (Turchin)
e.g., revenge and war are self-amplifying but not intrinsic to humans but how can they be averted?
e.g., religions unite and direct motivations but lead to divisions and oppositions
How to avoid catastrophe? Find a direction of development? Find causality? Predict outcomes?
We are notoriously bad at these
Our simplifying assumptions get in the way
Economics assumes independence of peoples choices
But people affect or copy others' choices (lead to irrational and potentially devastating bubbles of crowd behavior (Malcolm Gladwin's example of Enron behavior: copy success rather than took at data - they had no income!)
International relations models behavior by models of costs/benefits, risk and expectations
But the "laws" assumed to model behaviors change with major world events, like treaties that establish principles such as sovereignty (Westphalia), order (Napoleonic code), or human rights (United Nations).
Sociology and Politics model behavior by surveys
But people do not act independently, and their survey answers and demographics are not independent:
Galton's Problem as a theme of this course

Day Two Jan 7

Class notes

Finding and naming a dependent variable (depvar and depvarname)

  • Choose carefully, look at codebook and literature first: Sources for codes on the SCCS - find the literature where authors have contributed SCCS codes, and the literature where authors study those codes to get research findings. Copy to your project word-file the codebook number, and the full description of the variable from the codebook.
  • Then see if your variable number is found in this program: ---EduMod-52 speedy program or ---EduMod-52 working program|---. If it is, then place a # (comment out this line) IN FRONT of your variable. DO THIS FOR ANY VARIABLE THAT SEEMS EQUIVALENT, e.g., ### pctFemPolyg=SCCS$v872. If you dont do this they you will get a VIF error that says that a variable was aliased. To see the errors, move the cursor bar up to the FIRST error in the program, and ignore any later errors. Report that error to me if you cant solve the problem.
  • Now replace those such program statements as those below with your own depvar and depvarname
depvar<-apply(SCCS[,c("v473","v474","v475","v476")],1,sum) #can replace "sum" with "max"
depvar<-SCCS$v860
depvarname<-"polygyny"

Two methods for Making Maps that show how your dependent variable (depvar) varies around the Globe

You can use either of these in your reports

are societies clustered spatially? (see your Geodatabase results)
are societies clustered by language group? Language map - Native Americans
are societies clustered by religion? Religion map - use Old World Religions only - Indigenous New World religions not shown.
Requests for new Geodatabase variables to map

Day Three Jan 12

Class notes

Choose your worksite on Edumod

Go to Imputing_data_for_Regression_Analysis#EduMod_for_2010_Classroom_lab002_and_PCs_at_home

Test and run your program

  • Go to "Programs" and open R
  • Run these test commands and report any error message immediately (or email) to your instructor. If you have a load() error we can reinstall the needed data on your C:/My Documents/MI directory. If you install at home you will need to create this directory and download the datafiles from [5], [6], [7], and [8] (already installed in SST155).
setwd("C:/My Documents/MI")
library(foreign)
library(mice)
library(tripak)
library(zoo)
library(sp)
library(maptools)
library(spam)
load("vaux.Rdata",.GlobalEnv)
load("SCCS.Rdata",.GlobalEnv)
library(spdep)
library(car)
library(lmtest)
library(sandwich)
  • If no errors then click the ||Edit|| tab for your program, Ctrl-A to copy the entire text, open R, and copy into R. It takes 5-6 minutes to run, so find something else to do in the meantime.
  • If you do not get a table of results, then move the cursor bar up slowly and report to your instructor the earliest error by copy and paste, ignoring all other.
  • If you do get a table of results, then move the cursor bar up slowly and copy the entire table, from the first BBB to the last AAA.
  • Go to Programs - open TextPad. Add a space. Copy the results. Into TextPad. There should now be a leading space on every line. Use Ctrl-A again to copy this new page with leading spaces.
  • Back at the Wiki insert a new line AT THE BOTTOM of the program you just ran:
==B Results for [your depvar}==
  • The function of the double equals is to create a NEW EDIT WINDOW. If you think this is useless knowledge, remember that this is precisely the way you edit WIKIPEDIA pages because we are using precisely the same open access software ("[Wikipedia:MediaWiki]]")
  • Copy the TextPad text into this window and save.
If you cant get your changes to work, I have csdiff on my computer in the classroom and we can compare a previous successful run with a newly edited unsuccessful run.

Mark each line for the significant variables in your ==|B Results for |your depvar|== with <-- keep

Now, edit your ==B| Results for [your depvar}== page adding <-- keep to every line in which the pvalue is <.10 (for fyll and fydd <-- keep if pvalue < .25 and decide which the the variables with VIF > 2.5 are so similar that you want to <-- keep only one of them. What you are doing here is deciding which variables will go into your xR Restricted Model so that you are more likely to get results in which most of the variables you select will be significant.

If this is a 1st time - Restricted Model xR<-

Then after you save your results, create the ==A| Program page for your Unrestricted Model xUR<-==. Do this by copying the contents after xUR<- to replace the contents after xR<-. Then do a test run of your new program, put results in a ==B| Unrestricted Results for [your depvar}==

For a 1st time - Unrestricted Model cull out and keep only the significant variables in xR<-

Then create an ==A| Program page for your Restricted Model xUR<-==. Do this by editing the contents after xR<-. Then do a test run of your new program, put results in a ==B| Restricted Results for [your depvar}==

Now copy the ==A Program that created your results== and copy it, along with the ==A| xR Restricted Model Heading==

  • Copy to word or print the list of variables to <-- keep.
  • Save the edit window and press |edit| for your ==A| xR Restricted Model Heading==.
  • use "Find" to search for xR< and remove all but the variables to <-- keep.
  • Since you have already successfully run the WHOLE program, you only need to copy and paste (into R to run) the second half starting with Program 2 to the end. If there are no errors you will get your ==B| xR Restricted Model Results== so create that heading, copy your results to TextPad as before (which adds the needed spaces in column 1), then copy and paste into the ==B| xR Restricted Model Results== edit window and save.
  • If you get an error, go back to the program, do a correction, and rerun the same half as before labeled "program 2." Repeat until you get your results. If you close R and restart you have to rerun the WHOLE program, which takes 5-6 minutes longer.
  • The results under "ccc" for your results are explained at Meaning of diagnostics in the Eff-Dow Restricted Model results.

Day Four Jan 14

Sign up for Soc Sci 240B(71500 NOT SSL 122 but Anteater I&RB 3030) talk on friday, attend, and get 1.33 course credits for a 1-2 page paper discussion one of the four fall talks. First talk: Nov 15 1:30-3:20 Doug White, UCI: Anthropology. Inferential Statistics for networks of causality: Designing an Open eRepository for Anthropological Knowledge.

Class notes

How Logic Limits Causality and Choice in building your Restricted Model: Implausibles and Equivalences

  • Don't try to predict your dependent variables with an independent variable that is in some way equivalent by definition.
  • "X can never cause Y" is taken as as Implausibles: e.g., X=mud cannot cause Y=rain. So avoid independent variables that are implausible causes of your dependent variable.
  • FOR EXAMPLE: Tropical climate might help create polygyny, but polygyny cannot cause climate. So don't have climate as your dependent variable with independent variables that could not possibly cause climate.

Day Five Jan 19

Class notes

Checklist for the class

2SLS checklist - post on your page(s)

Objectives for today

  • Run your ==1A| xUR<-
  • Results: ==2B| xUR< from last time note <--keep
  • Run your ==3A| xR<-
  • Copy:- to =4B| your results using Textpad
  • Copy ==1A| to ==5A| and CHANGE depvar<--SCCS$v860 to YOUR depvar<--SCCS$XXXX
  • Also CHANGE depvarname<--"polygyny" to depvarname<--"Your depvar name"
  • check 2SLS checklist for consistency in naming
  • Run your ==5A| xUR<-
  • Copy:- to =6B| (you create this header) your results using Textpad, then markup significant results with <--keep
  • Write down those vars to <--keep
  • Copy ==6A| to ==8A| and change the header
  • use Ctrl-Find to search in ==8A| for "xR<-" and erase all but the fyll, fydd, and variables to <--keep (YOUR RESTRICTED MODEL)
  • I expect to see your results in 4B, 6B and 8B by the end of class
  • To get them you will have created programs 5A (Unrestricted model for Your Depvar) and 7A (Restricted model for Your Depvar), with the xR<- part of 7A edited down to keep only the <---keep significant variables

If you want to use R from home add directory C:\My Documents\MI [[9]] and then use http://bit.ly/8GOazd which will give you the data files.

Finding and naming your independent variables - LATER !

  • Many are already in your program.
  • Add only one at a time as needed, if you dont get significant results.
If you cant get your changes to work, I have csdiff on my computer in the classroom and we can compare a previous successful run with a newly edited unsuccessful run.

Day Six Jan 21

Class notes

Checkin under "discussion" tab

review day 5

Add 3 references to your User:page relevant to your depvar

Using a recode in defining a dependent variable

e.g., Reply to Nick User:Roblesn You mentioned v150 fixity of residence as a dependent variable which would mean trying to see if the program finds variables like "agricultural potential" or "foodtrade" for example that predict v150. Alternately if it is the consequences of "fixity of residence" you might want to choose a dependent variable like "Priests". depvar<-(SCCS$v711<-3) or depvar<-SCCS$v884. Doug 14:53, 20 January 2010 (PST) If otherw want to meet with us at 10:30 for a tutorial and if there is a class going on we'll move over to the computer lab at the other side of the elevator.

Preparing a powerpoint

The key thing in presenting your results is to copy the last ccc list of output as TEXT to the ppt, and change the format to COURIER so the columns align correctly.

Summarizing findings

Edu-Mod 2009: The Individual Studies

2009 papers

2009 contributed papers

Day Seven Jan 26

Class notes

  • money 0.097 2.705 515.768 0.101 2.282 <-- keep
  • plunder -0.259 2.432 1623.896 0.119 1.739 <-- keep
  • hunt 0.140 2.933 787.369 0.087 5.689 <-- keep
  • foodscarc 0.192 9.930 44.591 0.003 1.445 <-- keep
  • himilexp -0.314 3.622 1239.033 0.057 1.794 <-- keep

Finished examples in article form

More about the methods in this course: Causes of Polygyny and Instrumental Variables Douglas R. White - © 2010draft

Meaning of diagnostics in the Eff-Dow Restricted Model results

Meaning of diagnostics in the Eff-Dow Restricted Model results

Day Eight Jan 28

Class notes

Complexity talk

4C Soc Sci 240B talk on friday 29

Day Nine Feb 2

Class notes

Exploring *Your* topic

Avoid getting it wrong

  • Choose a topic that really interests you, something to write about. Dont just choose the first variable in the codebook that will most quickly satisfy the purposes of the course, e.g., v5 “animals” or v17 “agriculture”. These choices do not lead to focused reading or a research paper because they are so vast. What are the research questions that motivate a paper? (A,J). There is little or no chance you will get a good paper topic out of these choices.
  • Get your readings done on your chosen topic. It is the combination of the on-topic SCCS readings and your findings that will make your paper. Use the Google Scholar tools for your topic to post citations for your readings on your user page e.g., Ryan’s posts last quarter for father’s role. See Anth174-09 Day 6 DepVar and IndVars Notes for examples like these.
  • http://www.jstor.org/stable/2780365?cookieSet=1 - Father-Child Relationships and the Status of Women: A Cross-Cultural Study, Scott Coltrane. The American Journal of Sociology, Vol. 93, No. 5 (Mar., 1988), pp. 1060-1095.
  • http://www.jstor.org/pss/3773230 Parental Acceptance/Rejection and Parental Control: Cross-Cultural Codes. Ronald P. Rohner and Evelyn C. Rohner. Ethnology, Vol. 20, No. 3 (Jul., 1981), pp. 245-260.
  • Choosing an dependent variable in the database (depvar<-SCCS$v57) that does not correspond to the name of the variable depvarname<-“agriculture” (J).
  • Coming to class, asking questions, getting help in or before class is essential: this class is participatory. The 3 students last quarter who didnt come to class for one reason or another had to take incompletes because they didn't have the in-class time to finish their work.
  • Avoid putting a # sign in column 1 of the program, it is interpreted as a wiki command to number the line

Why Causality?

  • Is it dangerous to your health?
  • Isn't it something we want to know, e.g., the Challenger disaster: The tendency to think deterministically IS dangerous to your health and hugely costly to our society.
  • Why is anthropology involved in studying causality?

History of causality

See class notes and from fall 2009 1) add the Maryland government site 2) link to the ESRI 2SLS site

The Revolution in Studying causality

Wikipedia:Judea Pearl From Pearl's slide 23 on Causality, part b, of two parts (see part a)

On page 156 of Hume's "Treatise of Human Nature", we find the paragraph that shook up causation so thoroughly that it has not recovered to this day. I always get a kick reading it: "Thus we remember to have seen that species of object we call *FLAME*, and to have felt that species of sensation we call *HEAT*. We likewise call to mind their constant conjunction in all past instances. Without any farther ceremony, we call the one *CAUSE* and the other *EFFECT*, and infer the existence of the one from that of the other."

Thus, causal connections according to Hume are product of observations. Causation is a learnable habit of the mind, almost as fictional as optical illusions and as transitory as Pavlov's conditioning. It is hard to believe that Hume was not aware of the difficulties inherent in his proposed recipe. He knew quite well that the rooster crow STANDS in constant conjunction to the sunrise, yet it does not CAUSE the sun to rise. He knew that the barometer reading STANDS in constant conjunction to the rain, but does not CAUSE the rain. Today these difficulties fall under the rubric of SPURIOUS CORRELATIONS, namely "correlations that do not imply causation".

Now, taking Hume's dictum that all knowledge comes from experience, that experience is encoded in the mind as correlation, and our observation that correlation does not imply causation, we are led into our first riddle of causation: How do people EVER acquire knowledge of CAUSATION?

Review of our class projects prior to full reading

Day Ten Feb 4

Class notes

causal networks

Indep/Depvar list - All

partial network of results

update of article discussed Feb 2 relating to causal networks

That paper provides a model for how to present your tables in your final paper

  • For your Edumod page keep the diagnostics as well as the list of variables
  • For your powerpoint put the list of variables in COURIER font on one page, the DIAGNOSTICS in COURIER font on another page
  • For your paper the list of variables and diagnostics in COURIER font on one page, using as a model the article above. The discussions results in that article provide a template for you to discuss your results

We're at the halfway point in the class: time for powerpoints

  • We'll work in class on getting everyone's restricted model done, will need you in class.

Day Eleven Feb 9

Class notes

Schedule for powerpoint talks today Feb 11 thru Feb 18th Day 14 and progress on projects

Today: Help for Chelsea, Ariana, Jasiellt
2SLS checklist

Day Twelve Feb 11

Notes

copy to your page(s)

2SLS checklist -

Talk for complexity credit Soc Sci 240B

Feb 12 4C Human Complexity Soc Sci 240B talk - Xiao Ningchuan

Maps - Add variable numbers you want to map for your paper

Indep/Depvar_list_-_All - add variables that you want below not in the old page

6 Animals
90 Police
138 Famine 
155 Money
167 Premar sex norms
282 Premar sex norms
577 Mythical Founders of the Culture
595 Domestic work
649 Theories of Illness - Fate
661 Fem Partic
821 fem contr agric
892 External war
1648 Frq of war
1188 Evil eye
1189 Evil eye
1721 Wealth
1738 Formal Ed	
237 superjh
270 Class stratification
149 
150
151
152
153
154
155
156
157
158
  • new
238 Moral gods

Help in class today for

Day Thirteen Feb 16

Class notes

Powerpoint presentations

Day Fourteen Feb 18 No class - DW will be at a conference

DRW will be in Phoenix and ABQ

Day Fifteen Feb 23

Class notes

Powerpoint presentations

Day Sixteen Feb 25

Talk for complexity credit Soc Sci 240B

Feb 26 4C Soc Sci 240B talk on friday 26 on "Mind and Consciousness"

Class notes

Powerpoint presentations

Mapping with GIS

Diagnostics

Day Seventeen Mar 2

Class notes

Analysis of all results

Final Models: For review

review summaries of your results for your final paper in Edu-Mod 2009-10: The Individual Studies

Day Eighteen Mar 4

Eff&Dow in CHINESE BOX OF INTERACTIVE EFFECTS2.png

Class notes

Powerpoints

Structure of overall findings

  • clickable svg of new network of findings by project Working69Copy2 seems like the left wing of results deal with sex, rape, interpersonal violence, police, control of dvellings have one them then the right wing deals with frat-int-groups, stress on resource, war/fighting, wealth/poor differences, (low fe-)male agriculture, and individual freedom to choose a spouse. These intersect in the middle variable, money. What is needed now is to put red links for negative relations, and thicker links for greater significance. Because there are so many links, its probably necessary in order to estimate causal effects to eliminate all but the most significant variables, e.g., pvar < .01.

Day Nineteen Mar 9 the biggest lessons about causality

User:Dante Anton - Dante: there is a note on your EduMod to add 2 variables (already defined) to your model: gath and pathstress, just add to xR<-. They had pval <.25 in the xRU, should have had you add them to begin with.

Class notes

Powerpoint

Overview

Prediction is not causality!!! - errors in causal logic

EduMod-31: Imputation and Regression - depvar 892 extwar

  • This model was not properly constructed. The object is not to take variables that obviously are a part of the same phenomenon (here, a warfare complex) to predict warfare. Regressions coefficients alone are merely predictions, correlations. Regression coefficients cannot be causal if there are omitted variables. The critique here is not with Alex, but with the bad advice he was getting on causal modeling to choose variables that are obviously good PREDICTORS and ignore all others. Ren Feng was new to the problem of causality and did not give good advice. This does not lead to a causal model.
NOT A GOOD CAUSAL MODEL!!!
                       coef      Fstat        ddf     pvalue   VIF
(Intercept)     -1.73147478  1.2818655   29.07113 0.26680505    NA
fyll             0.72520771  0.8082993   83.08151 0.37122189 1.997
fydd             0.39374182  2.1956964 1209.98619 0.13865702 2.040
plunder          0.35051229 11.1305381 5297.71900 0.00085503 1.210 v912 more plunder
prestigewarrior  0.21830094  8.5840789  106.80714 0.00414711 1.302 v903 more prestige
FrqAttacked      0.36381942 23.9843690   40.26092 0.00001619 1.220 v893 being attacked (biggest effect)
superjh         -0.09865217  6.2782466 1442.75704 0.01233231 1.149 v237 more *state* external warfare
 R2:final model R2:IV(distance) R2:IV(language) 
      0.4331529       0.9820238       0.9411298 
                Fstat         df pvalue
RESET           0.613    398.637  0.434
Wald on restrs. 0.553    230.279  0.458
NCV             6.689   1920.383  0.010
SWnormal        0.001 184417.863  0.978
lagll           1.006  22781.254  0.316
lagdd           0.997  33308.388  0.318

Instead the object is to run a completely general unrestricted model and see ALL OF WHAT PREDICTS the depvar.

The concept of Implausibles and errors in causal logic

Critiquing the model above it is easy to spot the following Implausibles:

Warfare for plunder causes warfare.
The prestige of warriors does not cause warfare. It is when wars are won that warriors gain prestige.

The following are not Implausibles:

More superordinate hierarchy may cause more warfare.
Being attacked may be a cause of external war.

So here are some GOOD CAUSAL MODELS for the same project:

                   coef     Fstat        ddf     pvalue   VIF
(Intercept) -1.57119812  1.097643  6958.6180 0.29482042    NA
fyll         1.01580236  1.508930 23487.4806 0.21931468 2.034
fydd         0.31590128  1.248229 25106.4197 0.26390173 2.106
foodscarc    0.05371256  1.902185   296.8695 0.16887209 1.052 <-- remove in final model
FrqAttacked  0.53699757 58.254083   846.1294 0.00000000 1.094
localjh     -0.14211018  2.790793 86491.0865 0.09481089 1.037
superjh     -0.09934922  6.112151   387.1047 0.01385456 1.048
 R2:final model R2:IV(distance) R2:IV(language) 
      0.3679240       0.9813494       0.9358644 
                Fstat         df pvalue
RESET           0.153    200.595  0.696

Wald on restrs. 0.319     82.154  0.574
NCV             2.722    157.658  0.101
SWnormal        0.434   1157.110  0.510
lagll           1.174  23398.815  0.279
lagdd           0.915 241029.387  0.339

WITHOUT foodscarc
                   coef     Fstat       ddf     pvalue   VIF
(Intercept) -1.56678527  1.082033 5716.0767 0.29828729    NA
fyll         1.09889350  1.753672 6456.6650 0.18546271 2.024
fydd         0.28079555  0.976354 2397.3471 0.32320034 2.089
localjh     -0.12610289  2.206010 6659.4806 0.13752113 1.018
FrqAttacked  0.52783026 55.467163  274.6798 0.00000000 1.079
superjh     -0.09899757  6.141427 1295.8759 0.01333167 1.046
 R2:final model R2:IV(distance) R2:IV(language) 
      0.3587232       0.9813494       0.9358644 
                Fstat         df pvalue
RESET           0.400    268.426  0.528
Wald on restrs. 0.319     82.154  0.574
NCV             3.277    162.443  0.072
SWnormal        0.755    256.052  0.386
lagll           1.147  20448.377  0.284

Aspects of the organization, logic and data for the paper: A review

  • Nominate discussion topics (these are not paper headings)
topic and literature
evaluating hypotheses in the literature
logic of causality and implausibles
role of the unrestricted model
evaluating the restricted model
interpreting and evaluating the + or - signs of the dependent and independent variables and how each variable is ordered
evaluating the "endogenous" or social (inter-societal) variables
positive and negative language effects: what do they mean?
evaluating against the language map
positive spatial clustering
evaluating with the map by R plots and GIS maps
the logic of the overall argument
conclusions
limitations
  • Organizing the paper

Outline

Day Twenty Mar 11

Talk for complexity credit Soc Sci 240B

Mar 12 4C Soc Sci 240B talk on friday 12 - Reichardt "Large Scale Structure in Complex Networks - Methods of Detection and their Theoretical Limitations"

Final Causal graph

Indep/Depvar list - All

Class notes

summary arguments and the stories of your papers

Since we don't have any paper drafts, our last day we'll have quick summaries of the arguments for each paper, including some dialogue with the articles you are using for your paper.

This is designed to help you complete your paper both as a logical argument and in powerful storytelling form.

Email the papers

General