Book contents
- Frontmatter
- Contents
- Preface
- Part I Concepts of configural frequency analysis
- Part II Applications and strategies of CFA
- Part III Methods of longitudinal CFA
- Part IV Strategies of CFA and computational issues
- 7 Exploratory and confirmatory search for types and antitypes
- 8 CFA and log-linear models
- 9 Computational issues
- Appendix A Computational issues. The estimation of tail probabilities for the standard normal and the F distributions
- Appendix B Estimation of expected frequencies in 2 × 2 × 2 tables under the assumption that main effects and first order interactions exist
- Appendix C Critical alpha levels under Holm adjustment for up to 330 cells and a priori alphas 0.05 and 0.01
- References
- Subject index
- Author index
7 - Exploratory and confirmatory search for types and antitypes
from Part IV - Strategies of CFA and computational issues
Published online by Cambridge University Press: 04 August 2010
- Frontmatter
- Contents
- Preface
- Part I Concepts of configural frequency analysis
- Part II Applications and strategies of CFA
- Part III Methods of longitudinal CFA
- Part IV Strategies of CFA and computational issues
- 7 Exploratory and confirmatory search for types and antitypes
- 8 CFA and log-linear models
- 9 Computational issues
- Appendix A Computational issues. The estimation of tail probabilities for the standard normal and the F distributions
- Appendix B Estimation of expected frequencies in 2 × 2 × 2 tables under the assumption that main effects and first order interactions exist
- Appendix C Critical alpha levels under Holm adjustment for up to 330 cells and a priori alphas 0.05 and 0.01
- References
- Subject index
- Author index
Summary
The following sections discuss exploratory and confirmatory strategies for CFA application. First exploratory then confirmatory approaches are introduced.
Exploratory and hybrid CFA
The exploratory search for types and antitypes has two main characteristics:
(1) The configurations for which one expects types or antitypes are not specified a priori. Rather, simultaneous CFA tests are applied to all cells. The obvious disadvantage of this strategy lies in the large number of tests performed. For large tables the adjustment of the critical level of alpha can lead to prohibitively small critical alphas and thus, result in the identification of only a very small number of types and antitypes.
(2) Because no explicit configuration predictions are set forth, emerging types and antitypes can be interpreted only tentatively. Interpretation of types and antitypes emerging from exploratory CFA typically relies on plausibility arguments rather than confirming theory generated hypotheses.
The second of these two characteristics can only be changed by introducing confirmatory characteristics in one's data analysis (cf. Abt 1987). The first characteristic led to three approaches to define more efficient test strategies. The first approach is to derive more powerful tests for the identification of types and antitypes (e.g., Lehmacher 1981; Lindner 1984), or to use more powerful tests based on residual analysis of loglinear models within a CFA framework (Lehmacher 1980b). The second approach involves the application of more efficient strategies of alpha-adjustment.
- Type
- Chapter
- Information
- Introduction to Configural Frequency AnalysisThe Search for Types and Antitypes in Cross-Classification, pp. 207 - 215Publisher: Cambridge University PressPrint publication year: 1990