A discussion of selected applications of new tests developed within the context of a large-scale curriculum for educable mentally retarded (EMR) children, the Social Learning Curriculum (SLC), is presented in this paper which investigates three types of reliability that need to be demonstrated in order to provide a basis of these applications. The three reliability coefficients refer to differences among students, classrooms, and tests. The SLC model is based on social environment levels (the Self, the Home and Family, the Neighborhood, and the Community). The SLC Survey Test is an experimental set of test items developed in an effort to tap samples of the conceptual skills contained in each of the 11 phases of the Self level. For this study, five phase tests were randomly selected and administered to ten randomly selected students from two samples of EMR primary classrooms and, in the analysis of data provided, three reliability coefficients derived. A discussion of the over-all findings suggests that the test measures are not adequate for their purposes at their present level of development. References and tables are included. Related documents are ED 075 972, ED 084 658, SO 006 684, 686 and 688. (Author/KSM)