Child care programs call for research that supports the healthy development and school readiness of children. Approaches to Learning is one of the five domains of early learning embraced by federal and state school readiness frameworks. Child care programs often use assessments that are completed by teachers, classroom aides, or professional support staff to monitor children's progress in school readiness domains and support children's learning (hereafter referred to as "teacher-report assessments"). However, widely-used traditional statistical techniques do not remove the variance associated with the teacher assessor of the child's functioning. Failure to differentiate this "assessor variance" from the child variance undermines a measure's ability to produce a truly child-centered assessment free of assessor variance. In addition, the only available teacher-report assessments of Approaches to Learning are too long for routine program use to monitor children's progress and guide classroom intervention. The proposed study will use advanced statistical methods to develop and validate a scientifically-based, practical teacher-report assessment of young children's Approaches to Learning for use in early child care classrooms. The study will accomplish this goal through four objectives: (1) determine the amount of assessor variance in the Learning-to-Learn Scales (LTLS), the most highly developed multidimensional assessment of Approaches to Learning; (2) illustrate the psychometric advantages of removing assessor variance and focusing only on child variance to determine validity; (3) design and validate a shorter version of the LTLS based on child variance, and (4) disseminate research findings to policymakers, child care professionals, and parents to stimulate a discussion about purposeful assessment in early childhood and to determine ways to improve assessment for children from low-income households.