Fundamental principles in drawing inference from sequence analysis
Fundamental principles in drawing inference from sequence analysis
Individual life courses are dynamic and can be represented as a sequence of states for some portion of their experiences. More generally, study of such sequences has been made in many fields around social science; for example, sociology, linguistics, psychology, and the conceptualisation of subjects progressing through a sequence of states is common. However, many models and sets of data allow only for the treatment of aggregates or transitions, rather than interpreting whole sequences. The temporal aspect of the analysis is fundamental to any inference about the evolution of the subjects but assumptions about time are not normally made explicit. Moreover, without a clear idea of what sequences look like, it is impossible to determine when something is not seen whether it was not actually there. Some principles are proposed which link the ideas of sequences, hypothesis, analytical framework, categorisation and representation; each one being underpinned by the consideration of time. To make inferences about sequences, one needs to: understand what these sequences represent; the hypothesis and assumptions that can be derived about sequences; identify the categories within the sequences; and data representation at each stage. These ideas are obvious in themselves but they are interlinked, imposing restrictions on each other and on the inferences which can be drawn
sequence analysis, dissimilarity algorithms, stochastic models, event histories, inference, visualisation, clustering
Southampton Statistical Sciences Research Institute, University of Southampton
King, Tom
e7ee0b03-a6ed-4ab7-9836-7d564e317e29
15 March 2010
King, Tom
e7ee0b03-a6ed-4ab7-9836-7d564e317e29
King, Tom
(2010)
Fundamental principles in drawing inference from sequence analysis
(S3RI Applications & Policy Working Papers, A10/02)
Southampton, GB.
Southampton Statistical Sciences Research Institute, University of Southampton
39pp.
Record type:
Monograph
(Working Paper)
Abstract
Individual life courses are dynamic and can be represented as a sequence of states for some portion of their experiences. More generally, study of such sequences has been made in many fields around social science; for example, sociology, linguistics, psychology, and the conceptualisation of subjects progressing through a sequence of states is common. However, many models and sets of data allow only for the treatment of aggregates or transitions, rather than interpreting whole sequences. The temporal aspect of the analysis is fundamental to any inference about the evolution of the subjects but assumptions about time are not normally made explicit. Moreover, without a clear idea of what sequences look like, it is impossible to determine when something is not seen whether it was not actually there. Some principles are proposed which link the ideas of sequences, hypothesis, analytical framework, categorisation and representation; each one being underpinned by the consideration of time. To make inferences about sequences, one needs to: understand what these sequences represent; the hypothesis and assumptions that can be derived about sequences; identify the categories within the sequences; and data representation at each stage. These ideas are obvious in themselves but they are interlinked, imposing restrictions on each other and on the inferences which can be drawn
Text
s3ri-Workingpaper-A10-02.pdf
- Other
More information
Published date: 15 March 2010
Keywords:
sequence analysis, dissimilarity algorithms, stochastic models, event histories, inference, visualisation, clustering
Identifiers
Local EPrints ID: 79404
URI: http://eprints.soton.ac.uk/id/eprint/79404
PURE UUID: 016788c1-6af3-46df-a148-0d80dc983f88
Catalogue record
Date deposited: 15 Mar 2010
Last modified: 29 Jan 2020 13:40
Export record
Contributors
Author:
Tom King
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics