Typologies, taxonomies, and the benefits of policy classification.

Policy typologies are one of the most durable analytical frameworks in political science. Originally conceived by Lowi nearly 40 years ago (1964, 1972), numerous variants of policy typologies provide theoretical structure for a broad range of scholarship (e.g., Schneider & Ingram, 1997; Lowi, 1998; Mooney 1999; McCool, 1995, pp. 244-250). The typological approach, however, has well-known flaws. Typologies draw their theoretical strength from the idea that public policies can be systematically classified and that associated with each policy category are distinct and predictable patterns of political behavior. The central difficulty for typologies is establishing some means to objectively assign policies into conceptually distinct categories. Once considered a natural precursor to a general understanding of politics, the theoretical promise of typologies has been consistently frustrated by the inability to construct a system for valid and reliable classification (Greenberg, Miller, Mohr, & Vladeck, 1977; Steinb erger, 1980). 1 argue a potential solution to the classification problem lies in the empirical (as opposed to conceptual) construction of policy categories. This is an argument for a taxonomy rather than a typology of policy The article proceeds with an overview of the typology literature and the central importance of the classification problem, examines the merits of adopting taxonomic methods, and presents an expository example by constructing a taxonomy based on morality policy theory, a typology-based framework at the heart of broad and newly aggressive research literature (for surveys, see Mooney, 1999; Tatalovich & Daynes, 1998).

Typology Theory: Promise and Problems

Lowi's (1964, 1972) original typology theory was anchored in two basic ideas: (1) Policy causes politics, which reversed (and continues to reverse) the causal path assumed by most political scientists. The argument is that different types of policy spawn different power relationships among individuals and groups, and these relationships can be described and predicted on the basis of policy type. (2) The central characteristic of government is coercive power, the ability to force individuals and groups into certain activities or behaviors. The latter was used as the basis for a classification scheme to specify how policy determined power relationships.

Lowi's typology was based on a two-dimensional table, with the dimensions indicating the likelihood of government applying its coercive powers (immediate or remote) and the target of its coercion (the individual or the environment of conduct). This produced Lowi's four well-known policy categories: distributive policy (likelihood of coercion remote, coercion applied to individuals); regulatory policy (likelihood immediate, applied to individuals); redistributive policy (likelihood immediate, applied through environment); and constituent policy (likelihood remote, applied through environment). Once classified, detailed explanations and predictions of the political patterns for any policy could be extracted with relative ease. The potential contribution to a general theory of politics inherent in such a claim seemed obvious.

Spurred by this promise, Lowi's typology spawned numerous empirical tests, and as some initial problems were encountered, a set of rival typologies were created (Froman, 1967; Edelman, 1974; Wilson, 1973; Smith, T. A., 1975, 1982; Heclo, 1972; Steinberger, 1980; McCool, 1995, pp. 244-250). Although scoring some successes and illuminating a number of interesting policy problems, it soon became apparent that typology frameworks had inherent problems. These applied to typologies generally, though given its popularity and centrality to the field, Lowi's work attracted the most critical attention. Lowi's work was a prominent target in two most of the most important critiques of policy typologies that more than 20 years ago raised a set of questions never satisfactorily answered (Greenberg et al., 1977, Steinberger, 1980). Common to Greenberg et al. and Steinberger's assessments was the observation that it was virtually impossible to objectively classify policy: Scholars classified the same policies differently, so me policies overlapped categories, and others seemed to shift categories over time or in response to changes in the broader political environment. Both critiques concluded that if policies cannot be clearly classified, any policy typology loses its claim to be supporting the search for a general understanding of politics. The key obstacle in addressing this issue was identified as the ambiguity of policy as a concept, that its status as a particular type was a mental construct and not an observable and measurable empirical quality.

Several scholars have responded to this problem, with mixed results. Some try to modify Lowi's original typology making categories more permeable or adding other dimensions to the process of classification (Anderson, 1997; Heckathorn & Maser, 1990; Miller, 1990; Spitzer, 1987, 1989). Others accept Lowi's premise that policy causes politics but abandon a rigidly delineated classification scheme, arguing that the struggle to classify or define policy is itself at the center of political conflict (Stone, 1997; Schneider & Ingram, 1997; Smith, T. A., 1982). These revisionist approaches suffer from one or more of the following drawbacks: They mix policy categories or increase their number, muddying the conceptual clarity of the original scheme; they reinvent the problem rather than solve it by creating a different set of classifications rather than specifying how to assign policies to those categories; in making classification a subjective operation of the groups involved in a given policy subsystem, they are forc ed to all but abandon the ambitious general theoretical objectives originally conceived to justify the project. Lowi (1997, 1998) has made a number of these points explicit in published assessments of these efforts.

Thus far, this is where the issue stands. Lowi's policy typology and its derivatives continue to be gainfully and usefully employed as explanatory frameworks and are periodically refashioned in an attempt to respond to the problems identified by Greenberg et al. and Steinberger. But the all-important classification question remains unresolved.

Typology Versus Taxonomy

There are two basic approaches to classification. The first is typology, which conceptually separates a given set of items multidimensionally Lowi's typology is a classic example. For present purposes, the key characteristic of a typology is that its dimensions represent concepts rather than empirical cases. The dimensions are based on the notion of an ideal type, a mental construct that deliberately accentuates certain characteristics and not necessarily something that is found in empirical reality (Weber, 1949). As such, typologies create useful heuristics and provide a systematic basis for comparison. Their central drawbacks are categories that are neither exhaustive nor mutually exclusive, are often based on arbitrary or ad hoc criteria, are descriptive rather than explanatory or predictive, and are frequently subject to the problem of reification (Bailey, 1994). Typologies generally--not just policy typologies--suffer from the sorts of problems Steinberger (1980) and Greenberg et al. (1977) attribute to Lowi's scheme. In other words, the typological approach is virtually incapable of producing the sharp policy distinctions needed to fully support the explanatory and predictive resources scholars want to extract from it.

A second approach to classification is taxonomy. Taxonomies differ from typologies in that they classify items on the basis of …