A Bayesian truth serum for subjective data.

by: D Prelec

Science, Vol. 306, No. 5695. (15 October 2004), pp. 462-466.

View FullText article

DOI, Pubmed, Hubmed

Reviews [Write a review of this article]

There are no reviews of this article

Find related articles from these CiteULike users

Yanno, klouie, stefanherzog, jrw, garyfeng, schaal, Borelli, ddahlem, eibin, proportional, suizan, Glimcher_Lab (group), ReadingLab (group), NoncommutativeGeometry (group), ACS-Basel (group), SamMartin, quianominorleo

Find related articles with these CiteULike tags

2004, bayesian, baysian, belief, computation, datamining, decision, information-theory, judgment, method, modeling, modelling, no-tag, opinion, pdf, review, statistics, subjective_data, subjectivity, to-do, trust, truth

Abstract

Subjective judgments, an essential information source for science and policy, are problematic because there are no public criteria for assessing judgmental truthfulness. I present a scoring method for eliciting truthful subjective data in situations where objective truth is unknowable. The method assigns high scores not to the most common answers but to the answers that are more common than collectively predicted, with predictions drawn from the same population. This simple adjustment in the scoring criterion removes all bias in favor of consensus: Truthful answers maximize expected score even for respondents who believe that their answer represents a minority view.

@article{citeulike:488160, abstract = {Subjective judgments, an essential information source for science and policy, are problematic because there are no public criteria for assessing judgmental truthfulness. I present a scoring method for eliciting truthful subjective data in situations where objective truth is unknowable. The method assigns high scores not to the most common answers but to the answers that are more common than collectively predicted, with predictions drawn from the same population. This simple adjustment in the scoring criterion removes all bias in favor of consensus: Truthful answers maximize expected score even for respondents who believe that their answer represents a minority view.}, address = {Massachusetts Institute of Technology, Sloan School of Management, E56-320, 38 Memorial Drive, Cambridge, MA 02139, USA. dprelec@mit.edu.}, author = {Prelec, D. }, citeulike-article-id = {488160}, doi = {http://dx.doi.org/10.1126/science.1102081}, issn = {1095-9203}, journal = {Science}, keywords = {baysian, belief, judgment, method, modelling, truth}, month = {October}, number = {5695}, pages = {462--466}, posted-at = {2006-03-17 11:08:16}, priority = {2}, title = {A Bayesian truth serum for subjective data.}, url = {http://dx.doi.org/10.1126/science.1102081}, volume = {306}, year = {2004} }

RIS record

TY - JOUR ID - citeulike:488160 L3 - citeulike-article-id:488160 TI - A Bayesian truth serum for subjective data. JF - Science VL - 306 IS - 5695 SP - 462 EP - 466 AD - Massachusetts Institute of Technology, Sloan School of Management, E56-320, 38 Memorial Drive, Cambridge, MA 02139, USA. dprelec@mit.edu. SN - 1095-9203 N2 - Subjective judgments, an essential information source for science and policy, are problematic because there are no public criteria for assessing judgmental truthfulness. I present a scoring method for eliciting truthful subjective data in situations where objective truth is unknowable. The method assigns high scores not to the most common answers but to the answers that are more common than collectively predicted, with predictions drawn from the same population. This simple adjustment in the scoring criterion removes all bias in favor of consensus: Truthful answers maximize expected score even for respondents who believe that their answer represents a minority view. KW - baysian KW - belief KW - judgment KW - method KW - modelling KW - truth AU - Prelec, D PY - 2004/10/15/ UR - http://dx.doi.org/10.1126/science.1102081 ER -

RIS BibTeX