Public Data for Text Entry Rate of Access Interfaces used by People with Physical Disabilities

This spreadsheet includes a description of the methods used in our systematic review of text entry rate (TER) with access interfaces and provides public access to the data.

This is a companion to our manuscript by the same title, published in Assistive Technology.

We hope this will provide useful insight into some of the details of our approach and serve as a resource for answering other similar research questions.

The sheets in this workbook correspond to stages of the full-text screening and quantitative synthesis for each of the 7 interface types.

Please note that this workbook is made available for viewing only, not editing. (If you find that you are able to edit the document, please contact the authors.)

Please contact the authors with any questions.

Authors: Heidi Koester (Koester Performance Research, hhk@kpronline.com) and Sajay Arthanat (University of New Hampshire, sajay.arthanat@unh.edu)

Research Question: What are the reported text entry rates of various text entry methods that are relevant to people with physical limitations?

Literature Search and Initial Screening

1. Search each database with keywords customized to the database (e.g., use MeSH terms with Medline). (1 author) (N=3687).

2. Scan abstracts, Part 1. manually sift through the results for an initial screening, based on title and abstract, to determine which citations to save. Save in Zotero citation library. (1 author)

Initial Screening criteria based on title: (adapted from Davies 2010). Topic focuses on assistive device or technology to enhance or facilitate computer use or communication. Access method appears to be a focus. N=617

3. Scan abstracts, Part 2. A more thorough screen of the abstracts, to remove those that definitely are irrelevant. (If we’re not sure, keep the reference for subsequent full paper analysis.) (1 author - with sample reviewed by 2nd author)

Exclude abstracts where specific DV’s are mentioned but do not reflect some measure of text entry performance with the access method. (e.g., only metrics were for target acquisition or survey instrument). N=362

Ultimately 18 additional records from manual search were added to this, yielding 380 articles for full-text screening.

Starting N = 380, with 362 from databases and 18 from manual additions.

A sheet for each round of full-text screening is included in this workbook.

Results anecdotal, duplicate, or unclear

The first round, then, excluded 297 articles, leaving (380 - 297) = 83 articles. This matches what we see in the spreadsheet labeled 'Step 9 (N=83). First round of full-text screening'.

Second round of full-text screening

This leaves 57 articles for data extraction, which agrees with the sheet labeled 'Step 10 (N=57). 2nd round of full-text screening.'

In the process of data extraction and validation, we realized that some of those 57 studies really should have been removed during Step 10.

2 (Todman 1996, McCormack)

5 (Leung, Adams, Thoumie, Hurlburt, Bay)

5 (Gibbons, Betke, Klein, Wobbrock 2006, Ripat)

Results anecdotal, duplicate, or unclear

2 (Bhattacharya, Simpson)

This agrees with sheet 'Step 10a (N=43). 3rd round of full-text screening.'

Summary across all rounds of full-text screening

Combining Steps 9, 10, and 10a, we get the following exclusion table:

Results anecdotal, duplicate, or unclear

In total, then, we’ve excluded 334 articles, leaving (380 - 337) = 43 articles.

Goal is to mathematically combine TER results across studies, to obtain an overall average text entry rate for each type of interface.

Starting N = 43, from full-text assessment above.

Removed 4 articles where standard deviation of text entry rate was unavailable. (Standard deviation is required to combine the quantitative results across studies.)

Remaining 39 articles for quantitative synthesis are shown in sheet 'Step 10b (N=39). Articles for quant synthesis.'

Quantitative Synthesis for each interface is shown in all sheets beginning with 'Step 11' in their name.

The sheet 'Step 12. Summarize across interfaces' summarizes the results.

Details of selecting TER data entries from included studies are shown in the sheets named 'Quant analysis interface counts' and 'Quant TER entry tracking'.