2023-01-15: A Summary of "Methodology for heuristic evaluation of the accessibility of statistical charts for people with low vision and color vision deficiency"

The research work addressed in the paper "Methodology for heuristic evaluation of the accessibility of statistical charts for people with low vision and color vision deficiency," published by Rubén Alcaraz Martínez, Mireia Ribera, Toni Granollers Saltiveri (2021), focuses on formulating a set of heuristics to assess statistical chart's usability while concentrating on the requirements of those with low vision (LV) and color vision deficiency (CVD). To do this, the authors created a set of heuristics based on the methodology defined by Quiñones et al. in the paper "A methodology to develop usability/user experience heuristics". The authors conducted two evaluations after establishing the first version of the set of heuristics (17 indicators). Following the assessments, the authors expanded the list to 18 indicators, and other changes were made, including a more straightforward scoring system and additional documentation for evaluators. This study was the first step in making charts accessible for those with limited vision, a user group that is widely overlooked in studies on digital accessibility.

Statistical charts condense and explain complex data in a more straightforward format that allows for rapid and easy comparisons, analysis, and consumption in academic and professional settings and also reduces the cognitive load associated with reading textual or tabular information. A person has low vision (LV) if their eyesight cannot be fully corrected using corrective lenses. Visual acuity, light sensitivity, contrast sensitivity, the field of vision, and color vision deficiency are the five categories into which LV issues are divided.

The Web Content Accessibility Guidelines serve as the industry standard for digital accessibility, and they have even been recognized as an ISO standard. Many nations have embraced these guidelines as the minimum degree of compliance for public (and in some circumstances, even private) websites. Perceivable, operable, intelligible, and resilient are the four theoretical pillars that form the foundation of WCAG, which organizes itself around every facet of accessibility. Each principle is described in various specific guidelines, which are then converted into immediately evaluable criteria under three levels of conformity. An extensive absence of articles and guidelines focused on the accessibility of statistical charts for patients with LV and CVD was revealed by a prior literature review. This gap has been found, and it exacerbates the marginalization of a user group that represents 97% of persons with visual difficulties.

To fill this gap, a formal and systematic methodology proposed by Quiñones et al. is adopted as the framework of reference in this study, and it is corroborated with the metrics proposed by Jiménez et al. in order to validate the effectiveness of the suggested indicators in comparison to an already-in-use heuristic list control. Eight stages make up this approach. Each one is listed below with a full description of the outcomes:

1) Exploratory stage: This stage aims to conduct a literature review with the intention of gathering data for the heuristic list through a review of WCAG 2.1 and associated materials and an attempt to compile all criteria pertinent to the topic of this work (results are shown in the table below). 

Data Collected from WCAG 2.1
(Source: https://doi.org/10.1007/s10209-021-00816-0)

The authors' second step involved conducting a review of the literature on chart accessibility for LV users (results are shown in the table below).

Data collected from research conducted on the accessibility of statistical charts
(Source: https://doi.org/10.1007/s10209-021-00816-0)

2) Experimental stage: This stage's goal is to study the data from earlier tests in order to find new information that was missed in stage one. However, the authors were unable to locate any prior studies that concentrated on chart accessibility for users with LV; hence, this step was temporarily skipped.

3)  Descriptive stage: This stage concentrates on choosing and ranking the most crucial inquiries from the data gathered in stages 1 (exploratory stage) and 2 (experimental stage). The previous figures show all the relevant information that had been collected.

4)  Correlational stage: A number of publications have already suggested tying accessibility and usability standards together, mostly by relating WCAG and Nielsen's heuristics and coming to the conclusion that there is a definite association between the two. In this stage, researchers attempt to harmonize user experience concepts, associated user experience qualities, and existing heuristic indicators with domain features and functionalities. The next step is to categorize the resultant heuristics into five groups: excellent practices, textual alternatives, color and contrast, readability, and additional features and functionalities.

5) Selection stage:  In this stage, the goal was to analyze the list of indications that had been developed up to this point and determine whether to maintain, modify, or remove them. According to its potential to address accessibility issues with statistics charts and to meet the demands of the various user profiles, an indicator's "Applicability" column was created, which indicates how significant it is within the context of this research. The three categories of importance are useful, important, and critical, in that order.

6)  Specification stage: In this stage, the results of the earlier stage's indicators were clearly specified. Following this, a set of 17 heuristics were defined, as shown sequentially below with their respective ID numbers. The heuristics were also scored on a 7-point Likert scale and weighted into three categories based on their severity or influence: low impact (level 1), average impact (level 2), and strong impact (level 3). Additionally, it is important to know that a less detailed scale may enable quicker responses and more distinct categories, but it may also lead to bias on the part of the evaluators when their choice is not included. A very granular Likert scale, on the other hand, is more likely to include extensive and inclusive categories, allowing for the collection of more accurate data and more significant statistical results, with higher reliability and validity, and less neutral and "uncertain" responses.

Heuristic 1.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 2.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 3.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 4.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 5.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 6.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 7.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 8.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 9.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 10.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 11.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 12.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 13.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 14.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 15.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 16.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

Heuristic 17.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

7)  Validation stage:  In this stage,  the set of heuristics was confirmed at this point by a heuristic evaluation. The process was now completed by the computation of four quality indicators: the ratios of unique problems, problem dispersion, severity, and specificity to compare the domain with control heuristics. These indicator values allow us to deduce that the proposed heuristics identify more distinct problems, the problems are better distributed, more severe, and more specific than in the control set, and as a result, the new set of heuristics is much more suitable for assessing the usability of statistical charts.

8) Refining stage: In this stage, the Likert scale was decreased,  a new heuristic (as shown below) was added, and descriptions of all the heuristics were enhanced to facilitate evaluation.

Heuristic 18.
(Source: https://doi.org/10.1007/s10209-021-00816-0)

The authors of this study developed a brand-new accessibility tool that complements WCAG and focuses on the usability of statistics charts for people with impaired vision and CVD. The developed tool and the indicators (18  heuristics) included in this work are contributions to the evaluation of chart accessibility  (35 statistical charts), with the main priority given to people with low vision and people with CVD, which tries to make up for a lack of research in guidelines, standards, and recommendations for LV subjects. There are more efforts being made to consider the challenges faced by various disability groups and provide them with accessible solutions; however other groups, such as those with cognitive impairment or low vision, are frequently overlooked. In particular, institutions and major corporations are unaware of low vision as a handicap, which prevents this group from being taken into account when attempting to solve the access hurdles for people with visual disabilities. Statistical charts are already a fundamental type of daily information and are increasingly a part of general digital literacy. Hence, the 18 heuristics proposed in this paper aims to fundamentally create better accessible charts. (It is also clearly mentioned in the paper that this new set of heuristics does not aim to replace WCAG criteria; instead, both tools are complementary to one another).To further confirm the set of heuristics, the research team is working on incorporating users as part of future work. 

(Source: http://hdl.handle.net/2445/182301)

The figure above shows an accessible chart that complies with the heuristics: Tiresias font family, a font size range of 16–20 px, a legend, safe colors and the use of various patterns for better differentiation, a contrast ratio of 6.9:1 and 6.4:1 for marks, information about the data source, short alternative text, a complementary caption with a brief comment, keyboard navigation compatibility, focus visibility, tooltips with data associated, customizable (SVG format), and an equivalent data table are all present. We live in a time where data is of utmost importance for myriad tasks. Statistical charts play a vital role in visualizing data to draw inferences effortlessly. It is easier for sighted users to visualize these charts as compared to low-vision users who have to continuously zoom and pan through charts making it tedious to analyze charts effectively. Hence, guidelines to make accessible statistical charts for low-vision users are listed (18 Heuristics). It is necessary to follow these guidelines whilst creating charts for the benefit of all. 

Alcaraz Martínez, R., Turró, M.R. and Granollers Saltiveri, T., 2022. Methodology for heuristic evaluation of the accessibility of statistical charts for people with low vision and color vision deficiency. Universal access in the information society, 21(4), pp.863-894. https://doi.org/10.1007/s10209-021-00816-0

- YASH PRAKASH  (@LunaticBugbear)