Statistical Development and Application of Cultural Consensus Theory

Abstract

Abstract : The grant was to develop new models and Bayesian statistical inference for Cultural Consensus Theory (CCT). CCT is an approach to pooling response data from informants (experts) to estimate their consensus knowledge, unknown apriori to the researcher. During the period of the grant, the PI made progress in all five areas of the proposed research, namely (1) Constructing a catalog of CCT models for different testing formats; (2) Developing and implementing Bayesian computational inference for the models; (3) Determining the minimal number of informants needed to achieve confidence in the pooled information; (4) Developing an approach to aggregating expert views of the ties in a digraph that imposes prior constraints on the consensus digraph; and (5) Developing CCT models that detect cultural variation and/or prevarication among the informants. Progress on the grant was evidenced by seven new publications on CCT; 15 invited talks on CCT by the PI; and two new grants on CCT, one from the Army Research Office (ARO) and the other from the Intelligence Advanced Research Projects Activity (IARPA).