More-or-less elicitation (MOLE): Testing a heuristic elicitation method

Files

hdl_54280.pdf (191.14 KB)
  (Published version)

Date

2008

Authors

Welsh, M.
Lee, M.
Begg, S.

Editors

Advisors

Journal Title

Journal ISSN

Volume Title

Type:

Conference paper

Citation

Proceedings of the 30th Annual Meeting of the Cognitive Science Society (CogSci 2008), 23-26 July, 2008: pp. 493-498

Statement of Responsibility

Matthew B. Welsh, Michael D. Lee and Steve H. Begg

Conference Name

Annual Meeting of the Cognitive Science Society (30th : 2008 : Washington DC)

Abstract

Elicitation of people’s knowledge is a central methodological challenge for psychology, with important impacts in many technical disciplines and industrial settings. The need to convert an expert’s beliefs into a useable format is of particular importance when judgments and decisions are made under uncertainty. Simply asking a person for their best estimate or to estimate a range is subject to many biases – e.g., overconfidence – and methods for eliciting information that avoid these effects are required. This paper presents a heuristic-based elicitation method, More-Or-Less Elicitation (MOLE) which, rather than requiring people make absolute judgments, asks them to make repeated relative judgments. MOLE uses these, along with confidence statements, to construct probability distributions (pdfs) representing a person’s beliefs. We evaluate MOLE by comparing these subjective pdfs with ranges elicited using traditional methods. The central finding is that use of MOLE greatly improves the accuracy and precision of elicited ranges, thereby reducing overconfidence. The benefits of this and other possible heuristic-based methods of elicitation are discussed.

School/Discipline

Dissertation Note

Provenance

Description

The document attached has been archived with permission from the copyright holder.

Access Status

Rights

© the authors

License

Grant ID

Published Version

Call number

Persistent link to this record