Morphy, P., & Graham, S. (2012). Word processing programs and weaker writers/readers: A meta-analysis of research findings . Reading and Writing , 25 (3), 641–678. https://doi.org/10.1007/s11145-010-9292-5

In this meta-analysis, grades 1–12 student writing performance while using word processing as a support was compared with performance under typical paper-and-pencil composition conditions. A small number (n=3) of the 27 studies comprising the data set incorporated writing composed through dictated response, employing voice-recognition accommodations.

A total of at least 1,019 students from across grades 1–12 participated in the 27 studies, which were published in the time frame 1984–2005. All participants were prescreened on writing and reading performance with either norm-referenced or researcher-created tests, and demonstrated performance at least one standard deviation below the mean in comparison to peers. Participants—termed "weaker writers/readers"—had difficulties that were attributed to characteristics including learning disabilities (15 studies), English learner status (2 studies), or low socioeconomic status (3 studies), and several studies had students considered at-risk of persistent low performance. Students with other disabilities—such as neurological, motor, or sensory impairments—were not included in the participant pool. The researchers stated, "The typical student in the collected studies was in grade 7, not-at-risk as LD, ELL, or low SES, but identified as having reading problems (n = 11), writing problems (n = 8), or both (n = 8)" (p. 656). Nearly all studies were apparently completed with participants from U.S. school settings; only one of the 18 dissertations was completed by a doctoral candidate in a non-U.S. institution (in Canada); all of the eight journal articles and the one report appeared in U.S. contexts.

To be included in this meta-analysis of performance data, the 27 studies were required to employ quasi-experimental (37% of studies), experimental (41%), or within-participant designs (22%). A total of 1,995 writing samples included 976 control paper-and-pencil compositions and 1,019 word-processed compositions; many studies included compositions in more than one condition from the same participants, while other studies compared between similar participants. Written composition performance was scored with a range of dependent variables: quality, length, organization, grammar, mechanics, vocabulary use; some studies were also examinations of students' motivation or expressed preferences for word processing (rather than handwriting). Genres of the written products included stories/creative writing, expository descriptions or comparisons, opinion or persuasive, or a mix of these genres. The researchers examined the assembled data set (a) for whether use of word processing enhanced writing performance and student motivation, (b) for whether performance varied for different forms of word processing, (c) for the degree to which the quality effects of the written products were attributable to the length of compositions, and (d) for the degree of meaningfulness of the text quality effects of word processing use.

Across the findings of up to 27 studies, use of word processing improved the performance of readers and writers with weak skills, as well as increased their motivation, in comparison to handwriting compositions. The improvements were reflected in statistically significant effect sizes, and included longer compositions, better organization, higher quality, and greater mechanical correctness; all other aspects of performance (i.e., grammar, in 2 studies; vocabulary, in 3 studies) were not statistically significantly different between test conditions. Many of the studies each reported similarly significant improvements. Across the studies examining motivation, students were demonstrably more motivated to write when word processing their compositions; writers with weak skills also mostly (74%) preferred word processing over handwriting compositions. Word processing testing conditions did not seem to have significant writing quality or mechanical correctness effects beyond the benefit of word processing alone; for instance, students receiving additional orientation to or instruction on word processing did not further improve performance. Inclusion of voice-recognition support during writing tasks did not benefit student performance in writing quality beyond the effect sizes associated with word processing. The researchers noted that word processing with both instruction and additional embedded supports during writing showed significant benefits, yet noted that there were only three studies with these testing conditions. The quality of the written compositions using word processing were found to be strongly related to the fact that the word-processed products were also longer. When comparing the word processing effects to previous (2007) NAEP writing performance data from students with disabilities, the researchers estimated that using word processing could be projected to benefit "the typical writing quality of such weaker writers (i.e., disabled students) to a level roughly half the distance between their current level and the 2007 U.S. National average for all students" (p. 673).