Improper sampling in natural experiments: limitations on the use of meta-analysis results in Bayesian updating
Author: Price, Lydia ; Vanhonacker, Wilfried R.INSEAD Area: Marketing Series: Working Paper ; 90/47/MKT Publisher: Fontainebleau : INSEAD, 1990.Language: EnglishDescription: 20 p.Type of document: INSEAD Working Paper Online Access: Click here Abstract: The natural experiment hypothesis underlying meta-analyses and their extensive designs give rise to many empty or scarcely populated cells. The implications of this improper sampling can be severe when the results are incorporated as prior knowledge in a Bayesian estimation framework. Using existing meta-analyses in marketing and a known recursive framework for updating estimates in linear regression models, the practical limitations of such priors are discussed and illustrated. Suggestions are provided to alleviate some of the problems| Item type | Current location | Collection | Call number | Status | Date due | Barcode | Item holds |
|---|---|---|---|---|---|---|---|
|
Digital Library | Available |
The natural experiment hypothesis underlying meta-analyses and their extensive designs give rise to many empty or scarcely populated cells. The implications of this improper sampling can be severe when the results are incorporated as prior knowledge in a Bayesian estimation framework. Using existing meta-analyses in marketing and a known recursive framework for updating estimates in linear regression models, the practical limitations of such priors are discussed and illustrated. Suggestions are provided to alleviate some of the problems
Digitized
There are no comments for this item.