Final published version
Research output: Contribution to conference - Without ISBN/ISSN › Abstract › peer-review
Research output: Contribution to conference - Without ISBN/ISSN › Abstract › peer-review
}
TY - CONF
T1 - Bilinear models for scaling and score building: when should they be used?
AU - Francis, Brian
AU - Davies, Elouise
PY - 2019/2
Y1 - 2019/2
N2 - This talk looks at the utility of bilinear models for scorebuilding in contingency tables and contrasts it with thecorrespondence analysis approach. The groundwork for usingbilinear models for score building was laid down byClogg(1982), who specified a set of rules for the instrumentalvariable against which the target variable is classified.Typical bilinear models used for this purpose include the logmultiplicativemodel and the correspondence analysis model.While this approach seems at first sight to be promising,there are issues relating to empty cells and sample sizewhich often mean that the model fails to form exactly asintended. We discuss whether Clogg’s rules needextending and determine whether similar rules are neededfor correspondence analysis. An example is used from theproblem of scaling crime harm and impact from survey data.Clogg, C. (1982) Using Association Models in SociologicalResearch: Some Examples. American Journal of SociologyVol. 88(1) 114-134
AB - This talk looks at the utility of bilinear models for scorebuilding in contingency tables and contrasts it with thecorrespondence analysis approach. The groundwork for usingbilinear models for score building was laid down byClogg(1982), who specified a set of rules for the instrumentalvariable against which the target variable is classified.Typical bilinear models used for this purpose include the logmultiplicativemodel and the correspondence analysis model.While this approach seems at first sight to be promising,there are issues relating to empty cells and sample sizewhich often mean that the model fails to form exactly asintended. We discuss whether Clogg’s rules needextending and determine whether similar rules are neededfor correspondence analysis. An example is used from theproblem of scaling crime harm and impact from survey data.Clogg, C. (1982) Using Association Models in SociologicalResearch: Some Examples. American Journal of SociologyVol. 88(1) 114-134
KW - log-multiplicative models, bilinear models, seriation, criminology, harm
KW - bilinear model
KW - harm
KW - score building
M3 - Abstract
T2 - Correspondence Analysis and Related Methods 2019
Y2 - 4 February 2019 through 6 February 2019
ER -