Book contents
- Frontmatter
- Contents
- Figures
- Preface
- 1 Introduction to Observational Methods
- 2 Coding Schemes and Observational Measurement
- 3 Recording Observational Data
- 4 Representing Observational Data
- 5 Observer Agreement and Cohen’s Kappa
- 6 Kappas for Point-by-Point Agreement
- 7 The Intraclass Correlation Coefficient (ICC) for Summary Measures
- 8 Summary Statistics for Individual Codes
- 9 Cell and Summary Statistics for Contingency Tables
- 10 Preparing for Sequential and Other Analyses
- 11 Time-Window and Log-Linear Sequential Analysis
- 12 Recurrence Analysis and Permutation Tests
- Epilogue
- Appendix A Expected Values for Kappa Comparing Two Observers
- Appendix B Expected Values for Kappa Comparing with a Gold Standard
- References
- Index
8 - Summary Statistics for Individual Codes
Published online by Cambridge University Press: 05 June 2012
- Frontmatter
- Contents
- Figures
- Preface
- 1 Introduction to Observational Methods
- 2 Coding Schemes and Observational Measurement
- 3 Recording Observational Data
- 4 Representing Observational Data
- 5 Observer Agreement and Cohen’s Kappa
- 6 Kappas for Point-by-Point Agreement
- 7 The Intraclass Correlation Coefficient (ICC) for Summary Measures
- 8 Summary Statistics for Individual Codes
- 9 Cell and Summary Statistics for Contingency Tables
- 10 Preparing for Sequential and Other Analyses
- 11 Time-Window and Log-Linear Sequential Analysis
- 12 Recurrence Analysis and Permutation Tests
- Epilogue
- Appendix A Expected Values for Kappa Comparing Two Observers
- Appendix B Expected Values for Kappa Comparing with a Gold Standard
- References
- Index
Summary
After data collection is complete, and before data are analyzed, many measurement methods require intervening steps – some sort of data reduction – even if it is only computing a score from the items of a self-report measure. Behavioral observation, however, seems to require more data reduction than most measurement methods. Rarely are the coded data analyzed directly without intervening steps. First, summary scores of various sorts are computed from the event, timed-event, and interval sequential data produced by the coders. In other words, the data-as-collected, which usually reflect categorical measurement, are transformed into summary scores for which ratio-scale measurement can usually be assumed.
As with scores generally, the first analytic steps for summary scores derived from behavioral observation involve quantitative description. Descriptive results for individual variables (e.g., means, medians, and standard deviations, as well as skewness and distribution generally) are important – first, of course, for what they tell us about the behavior we observed, but also because they may define and limit subsequent analyses. Limited values or skewed distributions, for example, may argue against analysis of variance or other parametric statistical techniques. But what summary scores should be derived and described first?
- Type
- Chapter
- Information
- Publisher: Cambridge University PressPrint publication year: 2011