Continuous Improvement Model

In addition to the increased focus for schools on setting and meeting high standards of student performance, NCLB emphasizes the importance of evidenced-based interventions. Educators can not rely on tradition, anecdotal evidence and unsubstantiated claims from vendors for determining the effectiveness of products and programs. Discovery Education weaves the concepts of research-based and data-driven decisions into all aspects of product development and school support to ensure that teachers have what works.

Based on scientific design principles that ensure reliability and validity, Discovery Education applies the Continuous Improvement Model (shown right) to its own product development. Using content research, items matched specifically to each state's standards and evidence-based analysis, revision continues at every stage of program development.

This is an ongoing process; each year Discovery Education performs extensive psychometric analyses to ensure teachers receive the most accurate and reliable, state specific feedback possible. Teachers see what their students know and are then able to teach what they need to improve proficiency.

Discovery Education starts with the creation of a matched state specific test and predicts student proficiency by state. Following pilots with students using the ThinkLink test and comparing the results to these students' actual state test, we revise and refine our predictions. Even with the first year's administration, teachers can be assured that they are using a content valid and reliable instrument whose predictions get better with more student use, just as with any research-based standardized test.

Match Standards and Test

Discovery Education Assessment matchs the state standards and high stakes tests before offering the Predictive Benchmark series in a specific state.

Educators can be certain that Predictive Benchmark tests and mini-assessment items are carefully aligned by grade and subject matter experts. They are reviewed multiple times to assure a true match to each state with careful examination of:

• Released item samples
• Published "State Specific" testing benchmark definitions
• State-defined content standards and limits, and
• State publication of proficiencies and previous year's high stakes' scores

This assures content validity and statistically reliable tests and items with the appropriate range of difficulty levels and item formats.

Matching procedures ensure that Discovery Education Assessment benchmark tests meet the requirements for scientifically based research defined by NCLB. In other words, in a math test, the overall average is useful, but the average within a skill, such as number sense or geometry is most useful to teachers. Similarly in each subject, data must be available on each skill to guide teachers and students on where improvement can occur. With state-published averages, a testing company must show that its assessments match state difficulty levels by skill.

Thus, predictive assessments can only have content validity when test items and skill results are state specific. Anybody can claim content validity. Only by viewing the proposed tests and reports that compare actual state results to test company results is it possible to determine if an assessment demonstrates content validity.

Predict Results

Discovery Education Assessment benchmark tests are reliable predictors of student performance on norm- and criterion referenced tests.

Industry standard statistical techniques are used and presented to demonstrate the high correlation between student results on ThinkLink tests compared with their actual performance on national and state standardized tests.

Discovery Education Assessment tests are just as difficult as-and often a bit harder than-nationally normed and state-specific high-stakes tests. Discovery Education Assessment results can be used to gauge progress toward "Proficiency" levels as defined by each state under the requirements of NCLB.