Understanding Strategies and Challenges of Conducting Daily Data Analysis (DDA) Among Blind and Low-vision People
Chutian Jiang, Wentao Lei, Emily Kuang, Teng Han, Mingming Fan · 2023 · Proceedings of the 25th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2023) · doi:10.1145/3597638.3608423
Summary
This paper investigates how blind and low-vision (BLV) people perform what the authors term Daily Data Analysis (DDA) — the common, everyday tasks of analyzing and deriving insights from data such as splitting expenses, computing stock portfolio changes, and calculating averages. While prior accessibility research has focused primarily on how BLV people access and navigate data, this study examines the full analytical process from opening a dataset through to deriving conclusions. The researchers conducted a mixed-methods study with 16 BLV participants who had experience using Microsoft Excel with screen readers, combining semi-structured interviews about their general DDA strategies with think-aloud task sessions where participants worked through cross-sectional data analysis, time-series analysis, and free exploration tasks on two curated spreadsheets. All sessions were conducted online due to COVID-19 restrictions, lasting 90-120 minutes total across two phases separated by 1-2 weeks. The study used thematic analysis with open coding performed by two coders, followed by affinity diagramming to organize findings. Participants ranged from ages 17 to 41, with varying levels of vision from totally blind to low vision, and all relied on screen readers as their primary assistive technology for spreadsheet work.
Key findings
The study identified five key approaches BLV people use for DDA: (1) overview obtaining, where participants traverse spreadsheet elements to understand structure, layout, and content types; (2) column comparison, using either horizontal alternating traversal for adjacent columns or vertical full-column traversal for distant columns; (3) key statistics identification, employing a three-step process of data splitting, within-group statistics calculation, and overall statistics comparison; (4) note-taking, using either external Word documents or inline spreadsheet annotations to reduce memory load; and (5) data validation, checking calculations and formulas for errors. Critical challenges emerged at each stage: participants struggled to understand spreadsheet layout and element quantities without visual overview, found it nearly impossible to read line charts even with OCR, had difficulty manipulating non-adjacent columns for comparison, lost within-group trends when splitting data into summary statistics, faced inconvenience switching between Excel and Word for note-taking, and found identifying outliers extremely difficult without visual reference points. Participants frequently traded accuracy for efficiency, skipping validation steps or using coarser data groupings to reduce cognitive load. The study also found that 15 of 16 participants could not meaningfully extract information from line charts.
Relevance
This research fills an important gap between data access and data analysis for BLV users. While much prior work has addressed making data visualizations accessible through sonification, haptic displays, and alternative text, this study reveals that BLV people often bypass visualizations entirely, working directly with raw tabular data through screen readers. This finding suggests the accessibility community should invest not only in making charts readable but in improving the raw data analysis workflow itself. The identified strategies — particularly data splitting and navigational patterns — provide concrete design targets for assistive tool developers. The authors suggest dynamic hierarchical overviews adapted for screen readers, automated error detection integrated with non-visual feedback, and AI-powered natural language interfaces for data querying as promising directions. For practitioners building data-rich applications, the findings emphasize the importance of well-structured spreadsheet layouts, consistent element positioning, and meaningful programmatic labels that screen readers can convey to support the overview-obtaining process.
Tags: blind and low vision · data accessibility · screen readers · spreadsheets · data analysis · think-aloud · qualitative study