Perhaps more than most psychologies, the coherence of behavior analytic terms depends upon the tight interweaving of experimental techniques and data with vocabulary and interpretation. The techniques emphasize the behavior of individuals measured continuously over extended time periods rather than data combined statistically across brief observations of many individuals. This deviation from the format that provides the empirical base for most other psychologies is often credited with having precipitated the founding of separate behavior analytic journals. Distinctive characteristics of the corresponding interpretive terms and relations, however, are not so widely understood or appreciated, as amply documented in the two reviews of psychological dictionaries that begin this part.

Baum argues that Wolman's Dictionary of Behavioral Science botches the basic distinctions essential to make psychology of whatever stripe conceptually coherent. Moore sees Harré & Lamb's Encyclopedic Dictionary of Psychologyas less flawed with contradictions and omissions, but notes that it mischaracterizes behavior analytic theory and fails to do justice to its concepts. Baum and Moore both offer metatheoretical justifications for their preferred explanatory languages. Baum focuses upon definitions, making a case for anchoring terms in the circumstances of their use rather than in deductive relations. Moore highlights ways in which behavior analysis attends to its own practices and thereby includes the behavior of the theorist within the theory. Reading these essays, behavior analysts may share Baum's and Moore's indignation at the trashing of their system of carefully interrelated terms; nonspecialist readers, however, may wonder at the vehemence with which behavior analysts react to the rephrasing of their terms.

Hinson's essay illustrates how interpretive characteristics of behavior analysis flow from its distinctive research strategy. Hinson clarifies the significance of the units chosen to measure behavior-environment relations, showing how issues of scale and level of analysis bear upon both ontogenic and phylogenic selection and elucidate some differences between behavioristic, mechanistic and mentalistic approaches. Furthermore, units extended in time, place and perhaps complexity enable the analysis of complex and symbolic behavior, including behavior characterized by "trait language," which is usually viewed as beyond the behavioral pale.

Dynamic invariances, briefly addressed by Hinson, provide the key theme of Nevin's essay. Nevin develops invariance as a central concept for science in general, but with special relevance for understanding behavior-environment relations in and of themselves. Higher-order invariances comprise the rationale for pulling varied behavioral procedures and their results into a single quantitative conception. Nevin shows how quantitative formulations can work hand-in-glove with other conceptual developments, such as those identified with feedback relations, signal detection theory, the generalized matching law, and distinctions between molar and molecular analyses.

Ferster, on the other hand, sees hazards in theorizing that is not focused sufficiently on procedures and data. Echoing Skinner's critiques of the psychological theorizing of earlier decades (Skinner, 1950), Ferster questions the behavior analytic drift toward verbally interesting theoretical comment at the expense of descriptive empirical discovery. Whereas Baum and Moore examine distortions of behavioral concepts by outsiders, Ferster is concerned with distortions in the writings of behavior analysts themselves. He characterizes behavior analysis in its pioneering days as valuing playful interactions between an experimenter's procedure and a subject's behavior more highly than eloquent verbal interactions between theorists. He strongly prefers talk about data, and talk about how we talk about data, to talk about explanations of data. His essay prompts us to ask whether the fundamental nature of behavior analysis has been altered by its evolving verbal practices.

The remaining essays concentrate on innovative techniques for descriptive analyses of data and for research design, again emphasizing behavior as a s ubject matter in its own right. These techniques are described with an admiring enthusiasm reminiscent of the playful process of discovery that Ferster misses in recent work. As sketched by Church, the techniques described by Tukey in his Exploratory Data Analysisemphasize visual analyses in which graphs are used to discover new patterns as well as to communicate results. Church compares the characteristics of these graphical techniques to those of cumulative records, primary research tools during the early days of behavior analysis. Curve-fitting is included as an appropriate way to summarize the major effects of variables, but more interestingly as a preliminary to searches for subtle relationships via the graphical examination of residuals. These and scale transformations provide systematic ways to identify invariances such as those discussed by Nevin and they provide models for data-inspired rather than theory-driven analyses.

In his review of Tufte’s Visual Display of Quantitative Information, Iversen shows how graphical techniques can be used with virtuosity to portray data. While Tufte addresses the serious business of graphics excellence and the efficient portrayal of complex quantitative relations, a witty irreverence is revealed by his concise directives: "Eliminate excess ink!" Iversen shows how Minard's classic graphic portrayal of Napoleon's disastrous invasion of Russia can be emulated in portraying a behavioral experiment, and how graphic techniques can support analyses of simultaneous molecular and molar processes. Thus, contributions from a contemporary statistician, Tukey, and a political scientist specializing in graphic design, Tufte, converge on the types of data-oriented strategies that initially made behavior analysis distinctive.

In contemporary experimental psychology, mathematical, statistical and graphical techniques are taught and justified in the context of instruction on the nature of experimental design. Osborne's review of the second edition of Johnston & Pennypacker's Strategies and Tactics of Behavioral Researchand related works is a personal commentary that examines the behavior analyst's alternatives. Once again, behavior in its own right is the focus, and the behavior of the experimenter/interpreter is included within the field of view. The resulting commentary highlights the distinction between verbally formulated rules for experimental design and contingencies that operate in interactions between the experimenter's behavior and the subject's behavior.