Findings from the Reliability Study
All students whose data were included in the analyses completed the Reading Measure in two separate administrations. The Lexile measures obtained from these two separate administrations were used to calculate the test-retest reliability of the Reading Measure . Due to the nature of a CAT, the same student may not receive the same set of questions across two administrations of the Reading Measure ; thus the test-retest reliability of the current Reading Measure relies on the assumption that the different items administered in both administrations are still measuring the same underlying reading ability. The initial data sets included 11,329 entries for Administration 1 and 10,506 entries for Administration 2 of the Reading Measure . After cleaning the data set, the final matched sample included 7,425 students who completed the Reading Measure for both administrations and had valid final Lexile measures. This data was matched with a data set that included demographic information provided by the district. The demographic information used for subsequent analyses included grade, gender, race, free and reduced-price lunch eligibility, English learner, special education, and gifted student status information. The independent research firm calculated Pearson correlation coefficients to determine the test-retest reliability of the Reading Measure . The overall reliability for the Reading Measure across the two administrations was found to be high: r = .90. The breakdown of the test-retest reliability, based on different demographic groups, is presented in Table 1. In every instance, the test-retest reliability of the Reading Measure remained significant and desirable (ranging from r = .76 to r = .94). These results are strong evidence of the test-retest reliability of the Reading Measure .
6
SCHOLASTIC RESEARCH & VALIDATION
Made with FlippingBook - Online catalogs