reasonable content analysis. The remainder of the chapter reviews primary examples of content analysis and delineates a set of dimensions for review that might help make the use of content analysis evaluations more informative to curriculum decision makers.

We identified and reviewed 36 studies of content analysis of the supported and commercially generated National Science Foundation (NSF) mathematics curricula. Each study could include reviews of more than one curriculum. Table 4-1 lists how many studies were identified in each program type (NSF-supported, University of Chicago School Mathematics Project [UCSMP], and commercially generated), the total number of reviews in those studies, and the breakdown of those reviews by elementary, middle, or high school. These reviews allowed us to consider various approaches to content analysis, to explore how those variations produced different types of results and conclusions, and to use this information to make inferences about the conduct of future evaluations.

The content analysis reviews were spread across the 19 curricular programs under review. The number of reviews for each curricular program varied considerably (Table 4-2); hence our report on these evaluations draws on reports by some programs more than others.

Table 4-3 identifies the sources of the studies by groups that produced multiple reviews. Those classified as internal were undertaken directly by an author, project evaluator, or member of the staff of a publisher associated with the curricular program. Content analyses categorized as external