Techniques to identify the degree to which a response pattern is unusual and the pattern's relationship to examinee background are presented. Sato's caution index and modified caution index use clusters of items for comparisons of observed performance outcomes and do not require the use of item response theory. Correlation between the indices was nearly one for the indices computed for the students and the problems. The algebraic formulations of the indices are discussed with their empirical interrelationships and relationship with total number-right scores for data collected by the National Assessment of Educational Progress (NAEP) from 13-year-olds in the 1978 Mathematics Assessment. High Caution indices of unusual response patterns for individual test examinees are discussed. The indices may be useful in interpreting a total test score and may identify particular strengths and gaps in examinee knowledge. The distributional and relational properties of the indices with total scores, indentification of student background characteristics with unusual response patterns, and school characteristics associated with aberrant response patterns are shown. The variables associated with the modified caution index and comparisons of subgroup caution index values are discussed. Primary type of information provided by the report: Results (Secondary Analysis). (CM)