Exploratory data analysis (EDA) is a well-established statistical tradition that provides conceptual and computational tools for discovering patterns to foster hypothesis development and refinement. These tools and attitudes complement the use of significance and hypothesis tests used in confirmatory data analysis (CDA). Although EDA complements rather than replaces CDA, use of CDA without EDA is seldom warranted. Even when well-specified theories are held, EDA helps one interpret the results of CDA and may reveal unexpected or misleading patterns in the data. This article introduces the central heuristics and computational tools of EDA and contrasts it with CDA and exploratory statistics in general. EDA techniques are illustrated using previously published psychological data. Changes in statistical training and practice are recommended to incorporate these tools.

No abstract

Competing interpretations of the structure of the White Racial Identity Attitude Scale (WRIAS; J. E. Helms & R. T. Carter, 1990) were assessed in 2 investigations. First, a meta-analysis of scale intercorrelalions and internal reliability estimates indicated that, after correction for measurement error, intercorrelations between some scales were equal to unity, suggesting that the structure of the WRIAS is less complex than the theory of White racial identity it is assumed to measure. Second, confirmatory factor analysis of 2 data sets likewise revealed that scale structures found in the data are more parsimonious than those suggested by theory. Although it remains unclear which construct or constructs are actually measured by the instrument, interpretations of the WRIAS as composed of 5 meaningful dimensions are unsupported.I gratefully acknowledge Sandra Choney and Wayne Rowe for providing the data collected at the University of Oklahoma and Jane Swan son and David Tokar for providing the data collected at Southern Illinois University. I also acknowledge the assistance in collecting and organizing the meta-analytic data provided by Dan Huston and Yu Chong-Ho.A preliminary version of the meta-analytic results was presented at the 103rd Annual Convention of the American Psychological Association, New York, August 1995. The bibliography of studies considered for the meta-analysis, along with computer programs and data used in all analyses reported here, is available on the World Wide Web at http://seamonkey.ed.asu.edu/~behrens/wrias.

In contrast to statistical approaches aimed at testing specific hypotheses, Exploratory Data Analysis (EDA) is a quantitative tradition that seeks to help researchers understand data when little or no statistical hypotheses exist, or when specific hypotheses exist but supplemental representations are needed to ensure the interpretability of statistical results. In this way, EDA seeks to answer the broad scientific questions of “what is going on here” and “how might I be fooled by my statistical results.” The techniques of EDA are discussed following the “4 Rs” of Revelation (graphics), Re‐expression (scale transformation), Residuals (model building and assessment), and Resistance (using summaries unaffected by unexpected values). The philosophical justification for EDA is presented in terms of C.S. Pierce's concept of abduction and the recognition of a broad range of analytic needs that arise throughout the research process. Several previously published datasets from psychological literature are re‐analyzed to illustrate the interpretive errors that can occur when techniques of EDA are omitted. In general, these errors occur because researchers unwittingly assume the existence of structure that is not supported by the data. Using the techniques of EDA, however, underlying structure is brought to the researcher's attention and appropriate interpretation can be obtained.

Prevalence and severity of depression and related negative cognitive self-statements were assessed in a sample of 465 junior and senior high school learning disabled (LD) and seriously emotionally disturbed (SED) adolescents receiving special education services in public school resource room programs. Twenty-one percent of the adolescents sampled experienced severe depressive symptomatology. Some senior high females exhibited a more negative cognitive style than their male peers, although no differences were found at the junior high level. There were no differences in severity of depressive symptomatology and related dysfunctional cognitive self-statements between LD and SED students. Results indicate that depression is a prevalent condition among many LD and SED adolescents, a finding that warrants increased attention among special educators. Implications for expanding school-based identification and intervention procedures are discussed.