Buzick, H. M. (2019). Testing accommodations and the measurement of student academic growth . Educational Assessment , 24 (1), 57–72. https://doi.org/10.1080/10627197.2018.1545571

Journal Article

Buzick, H. M. (2019). Testing accommodations and the measurement of student academic growth. Educational Assessment, 24(1), 57–72. https://doi.org/10.1080/10627197.2018.1545571

Tags

Disabilities Not Specified; Elementary; Emotional/Behavioral disability; Hearing impairment (including deafness); Intellectual disabilities; K-12; Language arts; Learning disabilities; Math; Middle school; Multiple accommodations; Multiple ages; Multiple content; Multiple disabilities; Physical disability; Speech/Language disability; Traumatic brain injury (TBI); U.S. context; Visual impairment (including blindness)

URL

https://www.tandfonline.com/journals/heda20

Summary

Accommodation

Accommodations were not specified; the focus was on the assignment of accommodations in math and ELA in only one of two consecutive years.  The relationship between accommodation assignment and student proficiency levels and aggregate growths were reported. 

Participants

An extant data set from a total of approximately 65,000 students from 2 unspecified U.S. states was examined. This total was compiled from the number of English language arts (ELA) and mathematics scores from students with various disabilities across grades 3–8. The number of students in each disability category were not tabulated, nor were student score data compared by disability categories. Nonetheless, disability categories for the population included: emotional-behavioral disability, hearing impairments (including deafness), intellectual disability, multiple disabilities, physical disabilities, specific learning disabilities, speech or language impairment, traumatic brain injuries, and visual impairments (including blindness).   Student data were sorted into four groups for comparison: students receiving no accommodations (in either year), students receiving accommodations in both years, students receiving accommodations in the first year only, and students receiving accommodations in the second year only. Other factors calculated in logistic regression model included the whiteness racial category, socioeconomic level (free- and reduced-lunch benefit eligibility), and students without disabilities.

Dependent Variable

A set of extant data from 2005–2009 administrations of two unidentified states’ grades 3–8 ELA and mathematics tests was analyzed for accommodations use patterns in single years and across years and to estimate relationships between inconsistencies in use and student growth across years.

Findings

The researcher observed that "nontrivial percentages of students [were] assigned accommodations inconsistently across two consecutive years" (p. 69). Specifically, among two state populations of students with disabilities across grades 3 through 8, the largest proportion were students who were assigned accommodations in both years, and the smallest proportion for many grades was students with accommodations in the first of two years. Relatively speaking, the second-largest proportion were students with disabilities who were not assigned any accommodations in the first year but were assigned accommodations in the following year. Further, the researcher found correlations between inconsistent use of accommodations across academic years and different rates of learning growth in English language arts and math. Specifically, students assigned accommodations only in the second year averaged the most academic performance improvement, in comparison with other groups of students with disabilities. These growth rate differences were found when using two different measurement approaches. These statistical links were observable in average test performance in English language arts and math across large data sets, such as those obtained at the state level, and not easily recognized in averages in individual schools due to their small sample sizes. The implications of these findings were discussed in consideration of assignment consistency, inferences about growth scores, and whether systematic policy change might threaten validity of score interpretation. Limitations of the study were reported, and future research directions were also suggested.