When we begin to receive end-of-year assessment data, we begin processing. When the deadline arrives for us to post RFAI results, we do, even though all data are not in. Because data continue to be not only submitted, but updated and corrected, almost all results can change, but especially state and LEA means and ranges. When the second version of STAR is released, additional data or corrections appear there and are incorporated, sometimes changing a school's data, but frequently changing an LEA's, especially those with year-round schools, which test too late to appear in STAR version 1.
This year (2007), when we compared our RFAIs with those calculated by EDS, we traced some slight discrepancies to a difference in how we rounded component numbers. We at CTAC rounded the five STAR components to 1 digit before adding them to the fluency and kindergarten numbers to produce an RFAI. EDS does not round those component numbers. When we reran our STAR components to 2 digits, we found a significant enough difference statewide to prompt rerunning our reports. Those differences did not affect ranks, produced no large changes in any one score, and did not affect significant progress status. CTAC's rounding technique was, in effect, making scores slightly higher. Across the state's 886 Reading First schools, our rounding added about 55 points total. Spread across all those schools, that number is not large, but large enough to make us want to pursue even greater precision.