Book contents
- Frontmatter
- Contents
- Preface
- 1 Why do linguists need statistics?
- 2 Tables and graphs
- 3 Summary measures
- 4 Statistical inference
- 5 Probability
- 6 Modelling statistical populations
- 7 Estimating from samples
- 8 Testing hypotheses about population values
- 9 Testing the fit of models to data
- 10 Measuring the degree of interdependence between two variables
- 11 Testing for differences between two populations
- 12 Analysis of variance – ANOVA
- 13 Linear regression
- 14 Searching for groups and clusters
- 15 Principal components analysis and factor analysis
- Appendix A Statistical tables
- Appendix B Statistical computation
- Appendix C Answers to some of the exercises
- References
- Index
14 - Searching for groups and clusters
Published online by Cambridge University Press: 05 June 2012
- Frontmatter
- Contents
- Preface
- 1 Why do linguists need statistics?
- 2 Tables and graphs
- 3 Summary measures
- 4 Statistical inference
- 5 Probability
- 6 Modelling statistical populations
- 7 Estimating from samples
- 8 Testing hypotheses about population values
- 9 Testing the fit of models to data
- 10 Measuring the degree of interdependence between two variables
- 11 Testing for differences between two populations
- 12 Analysis of variance – ANOVA
- 13 Linear regression
- 14 Searching for groups and clusters
- 15 Principal components analysis and factor analysis
- Appendix A Statistical tables
- Appendix B Statistical computation
- Appendix C Answers to some of the exercises
- References
- Index
Summary
To this point in the book the methods introduced and discussed have been appropriate to the presentation and analysis of data consisting of a single variable observed on each of several sample units or subjects. We have discussed situations which involved several experimental conditions, or factors, in the ANOVA chapter, but the data itself consisted of observations of the same variable under each of the experimental conditions. It is true that in chapter 13 examples were introduced where several variables were measured for each subject, but one of these variables was given a special status (the dependent variable) and the others (the independent variables) were used to assist in its analysis via multiple regression.
A rather different situation arises when several variables are observed on each sample unit or subject and we wish to present the data or to extract the information in the complete set of variables without giving a priori a special status to any of them. An example of this, which we shall develop in the next chapter, would be the set of scores of a number of subjects on a series of foreign language tests, with no reason for the scores on any one test to be regarded differently from the scores on the remainder of the tests. It is in such cases that multivariate statistical analysis is appropriate. Before looking at any technique in particular it may be helpful to make some general remarks about multivariate analysis.
- Type
- Chapter
- Information
- Statistics in Language Studies , pp. 249 - 272Publisher: Cambridge University PressPrint publication year: 1986