MATERIAL FOR NCME 2019

A researcher has proposed submitting the following op-ed to the New York Times.

1. What are some problems with the op-ed?
2. How could the op-ed be improved? Write a new first paragraph.

Op-Ed

Results of a recent study (Jones, 2019)show that a random coefficient hierarchical logistic regression model(RC-HLR)  predicts differential item functioning (DIF) more accurately than standard logistic regression (LR) DIF models. Bias and Type I error rates were reduced in the RC-HLR model relative to the standard logistic regression model. Since 2003, Florida has required students to be retained if they are not testing at  grade level in reading by the end of the third grade. Using the LR model, two of 100 items on the 2018 Florida grade 3 reading assessment were flagged for potentially having DIF favoring male students. Using the RC-HLR model, 80 of 100 items were flagged for DIF favoring male students. Supporters of this law frequently cite test score analysis as evidence of its effectiveness. Approximately 80 percent of Florida schools are expected to have uniform DIF one standard deviation or more below the estimated value of−1.084, corresponding to a DIF coefficient of −2.061 favoring male students. It is recommended that the state of Florida stop retaining students in grade based on its third-grade state reading assessment.