Requested Information
Revised data used for evaluation of redistricting scenarios continues to be incorrect. The data on 2016 PARCC test performance by polygon is to insure balance in test scores before and after redistricting. Previous MPIA requests #2018-047 and #2018-056 showed that test performance described in this dataset differed significantly from previously published data summaries. The answer to the previous request indicated that percentage passing (>=4) on PARCC tests was not based on the percentage of students taking the test, but rather was based on the percentage who passed one of the two tests. In other words, percentage passing English 10 = number passing English 10 / (number passing English 10 or Algebra 1 or Algebra 2). Students who failed both tests were not counted. Therefore the percentage passing was artificially inflated.
However, recalculation of the percentage of students taking the English 10 test calculated as 100 times number passing English 10 divided by number of students who would have taken it (number of 10th grade students), did not yield correct results. For example, percentage passing averaged more than 180% for many schools, and was more than twice the reported averages for PARCC 2016 reported online in Maryland Report Card. Other data were equally skewed.
Please verify and correct the data set again, please also verify data submitted to the State of Maryland.
However, recalculation of the percentage of students taking the English 10 test calculated as 100 times number passing English 10 divided by number of students who would have taken it (number of 10th grade students), did not yield correct results. For example, percentage passing averaged more than 180% for many schools, and was more than twice the reported averages for PARCC 2016 reported online in Maryland Report Card. Other data were equally skewed.
Please verify and correct the data set again, please also verify data submitted to the State of Maryland.
Status
Complete
Response Notes
Clarification provided from Assessments Office: The differences are related to actual enrollment vs enrollment at the time of the test. The initial data run off of 2015 proficient students. The models are based off of 2016-2017 school year enrollment. When matching the available data to the enrolled students you get data points for 9th and 10th graders from their middle school tests.