and other classification systems. Elias cautioned against spending too much money on a new system, urging revision rather than complete rebuilding. His other points of counsel included: establish a long-term program rather than attempt a one-shot revision, use criteria for revising the classification system but be prepared to ignore them in the interests of practicality, use a committee of producers of occupational statistics and description to make the revisions but have consultation with users, be prepared to handle the advocacy of specific occupational groups, seek advice from industry-based organizations, expect strong academic criticism of any revision effort, put resources into automated coding and user-friendly guides, and, finally, know that both the enumerative and the descriptive systems are essential and both must be maintained. In agreement with Elias, Wootton does not think ISCO-88 is useful for U.S. needs. First, it offers no added value over existing U.S categorical systems; second, its use of skill level and type as an aggregating principle has produced a grouping structure that is inconsistent across nations. However, she believes it "provides some useful general guidelines" for the U.S. revision process—one is the move toward "replacement of multiple, fragmented classification systems with a unified structure, another is the shift in the basis of classification towards an explicit recognition of occupational skills" (1993:329).

Several enumerative systems are used in the United States:

• the Classified Index of Occupations and Industries (U.S. Department of Commerce, 1990a),
• the Occupational Employment Statistics Structure (U.S. Department of Labor, 1992),
• the Standard Occupational Classification (U.S. Department of Commerce, 1980).

In the past, a patchwork of cross-walks has been required to link the occupational categories across these systems (National Occupational Information Coordinating Committee, 1993), and