At the dawn of the 21st century, North Carolina maintained a respected testing program that consisted of end-of-grade (EOG) tests in reading and mathematics for grades 3-8 and end-of-course (EOC) tests in selected high school courses. The EOG tests were vertically linked on a developmental scale to facilitate growth calculations for the state’s accountability program in grades 3-8. The EOC tests were scaled independently, each with its own unique scale.
After years of implementing an annual accountability program based on year-to-year gains, the state had accumulated a substantial longitudinal data base. Consequently, the state considered incorporating a longer-term focus on growth into its accountability reform efforts. To support that initiative, research was conducted to explore the possibility of describing growth in reading and mathematics throughout the elementary, middle and high school years, utilizing available measures in ways they had not been used before.
In a study conducted by MetaMetrics and the NC DPI (2011), a combination of psychometrics and statistical modeling were employed to examine the EOG and EOC data. Specifically, EOG and EOC test scores were strategically expressed on common scales (The Lexile Framework for Reading and The Quantile Framework for Mathematics) and the resulting measures were analyzed with a multilevel growth model. The goal was to test the feasibility of estimating statewide average growth curves across grades 3-11 using existing measures while satisfying six pre-specified criteria for extending the developmental trend beyond the grade 3-8 time frame.
The results confirmed that the analysis of growth can be facilitated by the use of both EOG and EOC scores when they are linked to the Lexile and Quantile scales. The Lexile and Quantile measures derived from EOC test scores behaved developmentally in the sense that: a) they identified the higher levels of reading comprehension and mathematics understanding that resulted from additional instruction and study; and b) this was predictably reflected in changes observed in the average growth curve. This finding provides the basis for a practical strategy for states who may wish to explore a long-term approach to growth by combining measures from developmental and non-developmental tests in their accountability program(s). This approach could be used to set long-term developmental growth standards for a state based on longitudinal panel data, rather than the usual practice of setting only year-to-year growth standards based on non-developmental (e.g., status projection) or short-term growth (e.g., gain score) formulations.