Book contents
- Frontmatter
- Contents
- Series Editor's Preface
- Acknowledgements
- Abbreviations
- Part I Basic concepts and statistics
- 1 Basic concepts and terms
- 2 Describing test scores
- 3 Investigating relationships among different sets of test scores
- Part II Statistics for test analysis and improvement
- Part III Statistics for test use
- Bibliography
- Appendix: Statistical tables
- Index
3 - Investigating relationships among different sets of test scores
Published online by Cambridge University Press: 05 May 2010
- Frontmatter
- Contents
- Series Editor's Preface
- Acknowledgements
- Abbreviations
- Part I Basic concepts and statistics
- 1 Basic concepts and terms
- 2 Describing test scores
- 3 Investigating relationships among different sets of test scores
- Part II Statistics for test analysis and improvement
- Part III Statistics for test use
- Bibliography
- Appendix: Statistical tables
- Index
Summary
Introduction
In the previous chapter I discussed procedures for describing the characteristics of univariate distributions – the scores from a single measure. However, in many situations we have scores from two or more different tests from the same group of individuals and may want to find out whether there is a relationship between these different scores. When we look at the relationship between two different score distributions for the same group of individuals, we use the term bivariate distribution. Suppose, for example, that we had observed that students who did well in classroom speaking tasks also tended to perform well in listening tasks, and wanted to investigate this further. We could give the class a test of speaking and a test of listening, and then look at their scores. We might notice that students who did well in speaking also tended to do well in listening. If the order of scores on the speaking test corresponded more or less to the order of scores on the listening test, we might have some sense that these two sets of scores are related to each other. Or suppose we had developed tests of reading, writing, grammar and vocabulary, and wanted to know whether the scores on these tests were related to each other. As with the previous example, we could give these tests to a group of test takers and take a look at their scores, noting whether students who did well tended to do so on all four tests. We might find that, in general, this seemed to be the case.
- Type
- Chapter
- Information
- Statistical Analyses for Language Assessment Book , pp. 78 - 116Publisher: Cambridge University PressPrint publication year: 2004