Using measures of student growth has become more popular in recent years—especially in the context of high stakes testing and accountability. While these methods have advantages over historical status measures, there is still much evidence to be gathered on patterns of growth generally and in student subgroups. To date, most research studies dealing with student growth focus on the effectiveness of specific interventions or examine growth in a few urban areas. This project explored math, reading, and English language arts (ELA) growth in the students of two rural school districts in Utah. The study incorporated hierarchical and latent growth methods to describe and compare these students’ growth in third, fourth and fifth grades. Additionally, student characteristics were tested as predictors of growth. Results showed student growth as complex and patterns varied across grade levels, subjects and student subgroups. Growth generally declined after third grade and students experienced summer loss in the second summer more than the first. Females began third grade ahead of their male peers in ELA and reading and began at a similar level in math. Male students narrowed the gap in reading and ELA in fourth and fifth grade and pulled ahead of their female peers in math in third grade. Low SES students were the most similar to their peers in math and ELA growth but were ahead of their peers in reading. Hispanic and Native American students started consistently behind white students in all subjects. Hispanic students tended to grow faster during the school year but lost more over the summer months. Native American students had more shallow growth than white students with a gradual decline in growth in fourth and fifth grades. ELA and reading growth were more closely related to each other than with math growth. Initial achievement estimates were more highly correlated with subsequent growth than previous years’ growth. A cross-classified model for teacher-level effects was attempted to account for students changing class groupings each school year but computational limits were reached. After estimating subjects and grade levels separately, results showed variance in test scores was primarily due to student differences. In ELA and reading, school differences accounted for a larger portion of the overall variance than teacher differences.



College and Department

David O. McKay School of Education; Educational Inquiry, Measurement, and Evaluation



Date Submitted


Document Type





student growth, teacher effects, principal effects, hierarchical linear modeling, growth modeling, summer loss