Multivariate methods in historical data: An Optimal Scaling Analysis with Missing Categorical Data
- Cecilio Mar Molinero 1
- Fabiola Portillo 2
- 1 University of Kent, UK, and Universitat Autonoma de Barcelona, Spain,
-
2
Universidad de La Rioja
info
- Christos H. Skiadas (ed. lit.)
Publisher: ISAST: International Society for the Advancement of Science and Technology
ISBN: 978-618-80698-2-4
Year of publication: 2013
Pages: 151
Congress: 15th Applied Stochastic Models and Data Analysis International Conference With Demographics 2013 Workshop. 25 – 28 June 2013 Mataró (Barcelona), Spain
Type: Conference paper
Abstract
The information contained in a database from the Spanish railwaycompany MZA for the period 1882-1889 is explored by means of OptimalScaling methods. The sample includes 992 employees who joined theMadrid-Atocha workshop during that period and contains missingqualitative values. A technique that combines correspondence analysiswith the k-means clustering algorithm is implemented to impute thesevalues, maximizing internal consistency as measured by Guttman’ssquared correlation ratio. The results show two characteristics observedin other studies of labour relations: the existence of «ports of entry» forworkers at low levels of qualification and long-term labour relations