ECRISS
Educational Consultants and Research Associates (ECRA)
Barrington Community Unit School District 220 has been working with the ECRA group for several years to develop a local growth model that reflects student performance in relation to performance of students in the previous years. The local growth model takes into account the pattern of the achievement of students over the past several years to build a model for comparison. This is important for district and school improvement purposes because students in Barrington perform better than students across the state. Metrics that have been provided by the state have been inadequate to assist schools with decision making. By contrast the data provided by ECRA’s Local Growth Model (LGM) have the benefit of combining information from between five and seven hours of testing with multiple measures to produce a reliable and valid index of student performance as well as a prediction of how well students might be expected to grow based on this index.
By design, ECRA’s growth model expects students to grow as students with similar growth profiles have grown in the past. This means that a reasonable expectation for growth for one student may be different than the expectation for another student based on what we know about that student’s past performance. While this may not appear to be an equitable model at first glance, digging a little deeper into the limitations of our assessments we find that this is more equitable than the expectation that all students would grow the same amount. For example, students who score toward the top of a scale may not grow as much because a test may not be able to reflect growth beyond a certain point. Similarly, a student who performs poorly may not grow as much as a typical student because the test is not sensitive enough to capture the growth that may be occurring. For example, if a student were learning how to add single digit numbers but the test were measuring long division, even if the student mastered addition and subtraction of three digit numbers with regrouping, the test might still indicate that the student could not divide. Alternatively, some students who score below expectations may be able to demonstrate more growth because tests are particularly well suited to measure their performance.
The model employed is realistic about the variability in test data. Specifically, the model assumes a growth of 0 +/-.29 standard deviations is typical. Growth that is below .30 standard deviations but above -.59 standard deviations is lower than expected, and lower than -.60 is unacceptable. Growth above .30 standard deviations is considered higher than expected growth. This growth is depicted with the analogy of a stoplight, i.e., green, yellow and red, for expected, below and unacceptable performance with the addition of a blue light for higher than expected growth relative to the past performance of Barrington 220 students. ECRA’s LGM growth update present district, subgroup, school and grade level performance summaries in the pages that follow. These data are useful for multiple purposes because they synthesize complex relationships between scores to make the data manageable and actionable.
The model was built from archival data from Barrington to set the parameters for the level of achievement that is expected for each student on combined scores derived from MAP, ISAT, EPAS and/or PSAE, based on the performance of students on available assessments from the previous year. With this in mind we can see that the ECRA summaries uses data from the fall, winter and spring of the previous school year to predict outcomes on assessments in the fall, winter and ultimately the spring of the current school year. The effects that are reported are the combined result of teaching, learning and testing over the course of two years. These results help us to understand the strengths and weaknesses of our curriculum and instruction, and plan for changes in the materials we purchase and the professional development we provide.
Quick Tips:
