Ben Backes and James Cowan from CALDER have published a working paper on the differences between computer- and paper-based tests.
In 2015, Massachusetts introduced the new PARCC assessment. Districts could choose whether to use the computer or paper versions of the test, and in 2015 and 2016, districts were divided fairly evenly between the two. The authors use this division to compare results for students in Grades 3-8.
Students who took the online version of PARCC scored about 0.10 standard deviations lower in math and about 0.25 standard deviations lower in English language arts (ELA) than students taking the paper version of the test. When students took computer tests again the following year, these differences reduced by about a third for math, and by half for ELA.
The study also looked at whether the change to computer tests affected some students disproportionately. There were no differences for math, but for ELA there was more of an effect on students at the bottom of the achievement distribution, English language learners, and special education students.
The authors point out that these differences not only have consequences for individual students, but for other decisions based on the data, including teacher and school performance measures, and the analysis of school wide programs.