From InterSciWiki
Jump to: navigation, search
patrilocality, intercommunity marriage, v598 v963 restrictions on extramarital sex, 
v961 restrictions on premarital sex, v973 participants in the marriage celebration. 
v69.d3,v598 v963.v961,v973
Other calculations of interest include restrictions on divorce, finality at marriage, grounds for divorce.

Suzanne's Model marriage celebration (v973) as the dependent variable. Among the three stars noted are: patrilocality, intercommunity marriage, restrictions on extramarital sex, restrictions on premarital sex, participants in the marriage celebration. Other calculations of interest include restrictions on divorce, finality at marriage, grounds for divorce.



Suzanne Final V973 Model v974 MarrCelebrationParticipants R2=0.44 Suzanne Frayser

The ** for the Hausman Test is irrelevant because Wy is not significant (pval=0.519). Therefore this is an excellent model for depvar Degree of Marriage Celebration, SCCS v973, with Rsq=0.587. That's about as high as Rsq gets outside of depvar variables on religion. The big evolutionary effect in Wy is language, about 84%, distance 16%.

v216.d1     dummy
v973        depvar
bio15Sq,bio.2Sq,sdalt,v1,v216.d1,v72,v961   indepvar restricted
v2002.d2,v1845,v1    indepvar UNrestricted
Suzanne's variables when significant

There were a couple more that worked with the marriage celebration (973): v12 (***) [gathered foods] and v17 (***) [money and credit]. I'd put them in as unrestricted independent variables, but I didn't have a theoretical framework for them beyond looking at subsistence variables. The money and credit could relate to exchange and gathered foods, which are generally women's work, could relate to exchange of women. But it seems to me that money and credit involve much more sophisticated social organization than gathering. So, I'm not sure why they work.

Suzanne's variables minus v5, v35, v935 not significant

Showing that a new Variable doesn't work

http://socscicompute.ss.uci.edu/u/drwhite/h/v973marriagecelebration model shows how a New Variable dx$v213.d3*dx$bio.15Sq  is not working


see model at http://socscicompute.ss.uci.edu/u/drwhite/h/unnamed-history-2-1-1#31:_DEf01f_h but up to CoSSci #19 is the model of V1649 Internal war using http://socscicompute.ss.uci.edu/

SCCS v973 degree of marriage celebration n=59

Steps in analysis: I copied into the blanks in CoSSci the variables suggested except for substituting the actual dependent variable v973:
New Indep vars: v1649,v1127.d2,v2137,v279.d5,v213.d3,v1265,v234  indep
First IndVars:bio.15Sq,sdalt,v1265,v213.d3  ## but not effect of Islam v2002.d2 or v2001.d2
New Indep UN:   v2002.d2,v1845,v1  (indepUn = UNrestricted vars) indepUn
Run results then start and continue from: (those CoSSci #s may vary)
CoSSci #22 Replied: v1127.d2, v67.d2 not significant but ToTry had: 
  sdalt (input adjusted accordingly and run - with no ecology variables test, distance and language only)
CoSSci #25 Replied: model works plus ToTry had: bio.15Sq , was added
CoSSci #28 Replied: model works plus ToTry had:
  v213.d3 , added but dropv1649,dropv1127.d2,dropv2137,dropv279.d5, v213.d3, v1265,   dropv234   added   sdalt,bio.15Sq
CoSSci #31 final model (cannot do further ToTry Variables)
New Indep vars:              v1265,v213.d3,sdalt,bio.15Sq (alphabetized) called indep  (YOU COULD TRY ADDING OR SUBSTITUTING v216.d1 (virilocal first year) for v213.d3) 
New Indep UNrestricted vars: v2002.d2,v1845,v1 (unchanged)  this is also called indepUn
There were further ToTry Variables which could be defined as New Variable: v213.d3Xbio.15Sq
Defined as: v213.d3*bio.15Sq  #but this does not work
Screen Shot 2014-12-29 at 3.07.08 PM.png

For a new and better model

v1649,v1127.d2,v2137,v279.d5,v213.d3,v1265,v234 NO NO
bio.2Sq,sdalt,v213.d3,v216.d1  ## No effect of Islam v2001.d2 or v2002.d2
ADDING v216.d1 (virilocal first year) because this does work it is apparent that *.d dichotomies are fetched by DEf01f only if there are included in the list of dummies
ToTry: bio.2Sq replaces bio.16Sq -- but renders v1265 (No) Occurrence of Famine the least significant (and there are no new ToTry Variables). So delete v1265 among indepvars
Screen Shot 2014-12-29 at 3.10.58 PM.png

But effects of the addition of patrilineality are not so strong as might have been expected

Screen Shot 2014-12-31 at 8.29.00 AM.png File:Screen Shot 2014-12-29 at 7.23.00 PM.png]]

Delete Patrilineal Descent - Repeats New and Better Model v973.1.xlsx

Screen Shot 2015-01-01 at 8.56.00 AM.png

Since Suzanne has read these ethnographies she might have a better idea of other missing variables

For now, send your ideas to Doug. Ignore this comment: Comments249 where we're trying to simplify the Hausman test

Comments on v973 by Doug

On 2/11/15 7:49 PM, sassy frayser wrote: Hi Doug,

The denouement of my paper (chapter/presentation) is testing variables and, hopefully, finding new avenues for exploration. Would you have any time this weekend to explore this? I apologize for it being so late in the game to request this. So much has changed since I wrote you last. People's plans have changed and changed and changed. The upshot is that I've had no time to work on the paper/presentation, and I won't until Saturday and Sunday. I really want to showcase the utility of using the variables to explore new hypotheses. Just a couple of examples will suffice.

D: Yes, I agree. You have one already, v973, celebration, but if its not in the set you want of emotional aspects of sexuality you might rethink that too: the experience of a marriage ceremony is intensely emotional and personal. My impression of the patrilocality aspect is that high formality of marriage in this case bonds the women into a network where there are intense relations between the women in locationally different male kin groups and in many middle eastern lineages where patrilateral cousin marriage (pcm) is allowed the wife effectively becomes an inside member of the lineage not an incoming member from an outside group. Alot of emotional aspects tie into he forms of family structures. Might be interesting to test whether a dummy pcm variable is another predictors of v973. This kind of wife, as with Ancient Judaism, effectively becomes an inside core member of a male lineage, as in the family of Abraham. And so forth, interweaving emotional ties with network structure for some of the variables you study.

D: You write well and have a great objective

D: couldnt tell what was missing with "as if postition" on p13

If I can find time to send possibilities to you before then, I will, but I think it's hope over reality. Yes, send possibilities -- I have time

Trying to find an error in a computed variable

If you can solve this, Lukasz, its a good example for Eric
Input Parameter    Value
Dataset    SCCS
Dummy variables    v213.d3,v216.d1,v279.d1
Dependent variables    v973
Independent variables in restricted model bio.2Sq,sdalt,v1,v216.d1,v213d3bio15Sq
Independent variables in UNrestricted model    v1,v1845,v279.d1
Exogenous variables
Additional variables to consider    bio.15Sq,v213.d3
Name    dx$v213d3bio15Sq 
Definition    dx$v213d3bio15Sq<-dx$v213.d3*dx$bio.15Sq
The error has to be in the last two lines:  is dx$v213d3bio15Sq TOO LONG A VARIABLE - i.e., some limit on variable length in CoSSci windows?
What does work is this
Input Parameter    Value
Dataset    SCCS
Dummy variables    v213.d3,v216.d1,v279.d1
Dependent variables    v973
Independent variables in restricted model bio.2Sq,sdalt,v1,v216.d1
Independent variables in UNrestricted model    v1,v1845,v279.d1-- Doug White http://intersci.ss.uci.edu/wiki/index.php/DW_home

Many thanks, Doug

Feb 2015 R code

DEf01f 2015 data are now in the cloud
bio.15Sq,v213.d3  #additional
#--set working directory and load needed libraries--
library(mapproj) #added