CommonLit works hard to ensure that scores produced by the Assessment Series assessments are both reliable and valid. Before any students take CommonLit’s assessments, the assessment development team carefully screens passages to ensure they meet CommonLit’s high standards and do not contain any content that could potentially be biased toward certain groups of students. The team then writes and tests questions in the field to make sure they meet strict performance criteria. Previous research has shown that CommonLit’s assessment scores developed by the team were highly associated with standardized state test scores, like the Florida Standards Assessment (FSA). Additional research is being conducted to evaluate how well students’ scores on the Assessment Series predict performance on standardized state test scores.
Internally, CommonLit’s data analysts run a series of statistical analyses of student assessment results to ensure that students’ scores on the assessment are consistent and can be trusted. A reliability estimate above 0.8 suggests that students are answering questions correctly or incorrectly in consistent and predictable patterns, based on the difficulty of the questions. CommonLit’s Pre- and Post-Assessments consistently surpass that benchmark.
To estimate the difficulty and reliability of assessments, a sample of over 90,000 students from 3rd-12th grade completed the assessment questions. The students who completed the assessments were from all 50 states and representative of the general population of U.S. students on factors tracked by CommonLit. Responses from at least 1,000 students per set of assessment questions were collected, and the data was then screened for quality prior to running analyses.