Schulte, A. C., Stevens, J. J., Elliott, S. N., Tindal, G., & Nese, J. F. T. (2016). Achievement gaps for students with disabilities: Stable, widening, or narrowing on a state-wide reading comprehension test? Journal of Educational Psychology, 108, 925–942.
An achievement gap is a persistent pattern of differences in level of academic achievement between groups of students who differ on a demographic characteristic. Previous research has documented achievement gaps based on factors that include socioeconomic status, limited English proficiency, and disability status. These achievement gaps have been observed in students’ level of academic ability at the start of school and in the pace of their gains in academic proficiency over time. Closing or reducing the magnitude of these gaps is among the goals of many education reform initiatives at the state and federal levels. Given that reading comprehension proficiency is critical for learning across the curriculum, addressing achievement gaps in reading is particularly important, especially for students with reading disabilities.
Schulte, Stevens, Elliott, Tindal, and Nese (2016) analyzed longitudinal data from the North Carolina state reading assessment for students in grades 3 through 7 to look for patterns of differences in initial level of reading ability in grade 3 and in rate of growth over time for students with disabilities, academically gifted students, and typically achieving students. They also examined the data for differences based on demographic characteristics such as socioeconomic status, race/ethnicity, gender, and limited English proficiency.
There are a number of theories regarding the way in which students grow from being nonreaders to being proficient at comprehending text. One theory, known as the simple view of reading, states that reading comprehension is the result of two skills: decoding and language comprehension. Skills such as phonemic awareness, inference-making, vocabulary knowledge, and others provide a foundation for developing word recognition and language comprehension abilities. According to Schulte et al. (2016), students with learning disabilities (LD) most often have difficulty with word recognition and its underlying skills, and students with intellectual disabilities typically struggle with both word recognition and language comprehension. Delays in the acquisition of either or both skills result in delays in developing reading comprehension. Once students with delays develop the needed underlying skills for comprehending text, it is possible that a period of rapid growth in comprehension skills would follow. This period of growth might occur later for students with disabilities than for typical students. Schulte et al. noted that this delayed growth might account for differences in reading comprehension growth rates between typical students and those with disabilities at different points in time.
Schulte et al. describe three patterns of growth over time that differ in their implications for achievement gaps.
Pfost, Hattie, Dörfler, and Artelt (2014), in a meta-analysis summarized here, synthesized the results of 78 studies that looked at these reading growth patterns. They found that patterns differed depending on the reading skill examined and the quality of the reading assessment, with evidence for each of the three patterns found for one or more reading skills. However, they excluded studies of students with disabilities from their analysis, so less is known about how these patterns might be present in reading growth data where students with disabilities are compared to general education students and academically gifted students. Additionally, most previous studies have not controlled for the effect of demographic variables, such as socioeconomic status, that may contribute to differences in patterns of growth. Studies that included students with disabilities and controlled for demographic differences have generally not found a fan-spread pattern of increasing achievement gaps over time, but results have been mixed.
Schulte et al. sought to add to the knowledge base on patterns of reading growth over time for students with disabilities, general education students, and academically gifted students. Their purpose was to compare the growth of these groups of students over multiple grades, controlling statistically for demographic differences. They wanted to determine which of the three patterns of growth characterized the growth of students with disabilities compared to general education students and the growth of academically gifted students compared to general education students.
To address their research questions, Schulte et al. analyzed scores on the North Carolina End of Grade Reading Comprehension Tests, the state’s reading assessment, starting with all students who took the test in 2003 when they were in grade 3 and tracking their scores through their grade 7 assessment. The state changed the reading assessment the following year, making it difficult to compare scores across a longer span of time. Assessment results from nearly 100,000 students were included in the analysis. About 13% of the students had disabilities and nearly 6% were academically gifted in reading. Students with reading LD comprised 4.7% of the total sample. To be designated as having LD in North Carolina at the time the data were collected, a student had to have a discrepancy of 15 standard score points or more between their IQ and achievement test scores.
To analyze the data across the 5-year timespan, Schulte et al. used an approach known as hierarchical linear modeling. This approach accounts for the fact that students’ scores over time are correlated. Scores from each year of testing (at level 1 of the model) are nested within students (at level 2 of the model) to reflect the relationship between the same student’s score on the assessment at different points in time. Using hierarchical linear modeling, Schulte et al. analyzed longitudinal growth models to discover and compare the patterns of reading growth over time for students with disabilities, gifted students, and general education students. These models allowed the researchers to determine the effect of demographic variables (gender, race/ethnicity, socioeconomic status, and limited English proficiency) and control for their influence on growth patterns.