BESE Pres Chas Roemer: Opt-out Zeroes Won’t Affect School Scores (??)
Let’s talk “opting out” of Louisiana’s “PARCC” test on the first day (Monday, March 16, 2015) as reported on the Louisiana Radio Network blog:
Calcasieu, Jackson, and Central School systems lead the state in the percentage of opt-outs during Monday’s first day of statewide PARCC testing. Jackson had an opt-out rate of 13-percent, followed by 12-percent for Calcasieu and eight-percent for Central.
Central School Superintendent Michael Faulk says 153 students in the district did not participate.“It’s really higher at Central Middle. It’s 10-percent at Central Middle and a little bit lower at Central Intermediate.”Statewide about one-percent of students opted-out of the Common Core testing. Opt-outs will be given a zero on the exams and that will transfer to the school and school district scores. Faulk says he’s concerned that his school system will be judged with no consideration to the percentage of opt-outs.
I do not have the Louisiana “PARCC” opt-out numbers for Tuesday, March 17, 2015; however, such was apparently concern enough for Central school district to seek a word from Louisiana Board of Elementary and Secondary Education (BESE) president, Chas Roemer.
The following message appeared around 6 p.m. on Tuesday, March 17, 2015, on the Central City News Facebook page:
BESE CHAIRMAN CHAS ROEMER Says Central Schools Will Not Be Penalized Because of Opt/Out.
By Woody Jenkins, editor.
In an exclusive interview with the Central City News this afternoon, Board of Elementary and Secondary Education chairman Chas Roemer said the board will not allow the Central Community School System to be penalized because a large number of students have opted out of the PARCC test being administered this week.
Roemer said the board had anticipated that many parents would opt out of the tests around the state but the large number in Central came as a surprise.
“We’ve told state Supt. John White to give us an analysis of how many students have opted out of each school district. But the fact that they have opted out will not affect the school’s performance score. In other words, if you’re an ‘A’ school, you’re not going to become a ‘B’ school because of students who opted out.
“Students have every right to opt out, and we support that. They also have a right to opt out of LEAP and iLEAP. But PARCC has been more controversial, and more students have chosen to opt out of this exam.”
Roemer said the board’s policy is that students who do not participate will get a zero on the test for statistical purposes but that will not affect the student or the school’s performance rating. The rest results will not be released until this fall, he said. [Emphasis added.]
That might sound good in an exclusive interview; however, “getting a zero on the test for statistical purposes” will most certainly affect statistical calculation of a school performance score.
Even though Louisiana is not using the PARCC test contracted to most PARCC consortium states, the test vendor supplying Louisiana’s “PARCC” test, Data Recognition Corp (DRC), will likely model Louisiana “PARCC” scoring scales to resemble those of official Pearson-developed PARCC tests. Sample PARCC scoring reports can be viewed in this New Jersey PARCC Power Point. The scores do not go to zero; however, the scoring is continuous, which means that entering a zero score would likely be analyzed as “zero correct.”
Therefore, entering a zero “for statistical purposes” for students who have opted out would indeed lower statistical calculations of school performance scores.
In order for a student’s opt out to not count against school performance score calculations, a researcher would have to set some value that could not otherwise be a score as a place marker for missing data. It is efficient to set a missing data numeric marker that could not be counted as a true value for any other variable in an analysis.
In the case in which variables are continuous, a value of zero often means the lowest possible score or measure. This could be true even if a scoring scale does not begin at zero. For example, even though the PARCC assessment scoring range might extend from 150 to 230 (which is what seems to be true based on the NJ PARCC info) and zero is not a value in that range, a score of zero could still mean “none correct.”
The only way for a researcher to ensure that a zero would not mean “none correct” is if the researcher intentionally defined zero as a missing data marker, one not meant to be analyzed. And frankly, such a choice would be rare; other continuous variables in a complex analysis could have true values of zero, and a researcher would not want to block these by using zero as a missing data marker.
The researcher determines what value to use as a missing data marker. Moreover, the researcher must intentionally define any chosen value as a place marker for missing data. A common missing data marker for continuous variables is the number 9, or 99, or 999, depending upon which of these could not possibly be the value of a true score. Again, the researcher intentionally chooses this missing data marker.
Even though Roemer has stated that zeroes will not negatively affect 2014-15 school performance score calculations, such a promise directly contradicts the language explaining Louisiana school performance score calculations in Louisiana’s 2014 No Child Left Behind (NCLB) waiver (see page 57):
NOTE: In the old and the new system [of school performance score calculation], 100% participation is required; schools receive a zero for non‐ participants. Also, because ELA and Mathematics are core competencies, student performance in these subjects will receive double the weight given to Social Studies and Science performance. …
… A zero is assigned to the assessment index of a school for every test and subject for students who do not test. The zeros are included in the calculation of the school performance score and directly, negatively affect the school’s letter grade. [Emphasis added.]
Contrary to what Roemer implies in his exclusive interview with Central City News, according to Louisiana’s 2014 NCLB waiver, a zero for “nonparticipation” is a score with a true value, one intended to negatively impact a school score “for statistical purposes.”
In order for opt-out “nonparticipant” scores to not count against a school, the “zeroes” need to be recoded as missing data using some value specifically designated such that an opt-out student’s missing score does not hold the meaning of “none correct.” This is the only way that Roemer’s words to Central schools could possibly be true.
Either Louisiana is following that NCLB waiver or defying it. Both are not possible.