YT Identity Survey #7/7 - final influence network results

This is my final results for the YT religious identity survey (I hope). I'm including as many qualities as I can for the purpose of trying to find the most likely probabilistic networks that describes the data. Gender is used as a lever in order to find the direction of influence for the qualities. In light of the previous clip, the results further shows that life after death belief is the prime influencer after gender.

The study is done for three divisions of religious identity, namely the whole set of grouped religious identities, only agnostics vs atheists and Christians vs the rest.

The type of statistical analysis done here is relatively new. Probabilistic networks were formalized in the 80s, though tools for doing inference on them seems to be of more recent date, though I do not know how recent (1995-1999?).

I'm using a tool for doing inference on categorical data called LearnBayes:
http://compbio.cs.huji.ac.il/LibB/programs.html
My thanks to the programmers, Nir
Friedman and Gal Elidan for making it
available for the public. The world
needs more scientists like that.

Category:

Science & Technology

Tags:

• religious
• identity
• survey
• inference
• probabilistic
• Bayesian
• networks
• BIC
• LearnBayes

[9macrina9]

yes...i think according to this data, an exploration of the basis for the belief in an afterlife may be fertile ground for further analysis...very interesting work here, trond.

[trondreitan]

Thank you!

[trondreitan]

Thanks! You are right. I haven't the knowledge necessary to make such a program on my own and going through them manually would be the death of me. (I'm counting 10^37 directed graphs, though not all are separate models. Or acyclic.) So it was great luck that I found the LearnBayes program.

[michaelispan]

Takk!

[randyhelzerman]

Wow. Great analysis. No doubt you would have gone blind trying to do all the calculations necessary to generate these influence diagrams had you not found the LearnBayes. What influences what is highly highly nonintuative to me. Amazing.