RESEARCH DOSSIER
Inclusive Human-System Design and Functional Accessibility
Abstract. Inclusive human-system design is the empirical and theoretical study of how interactive technologies, work systems, services, and their institutional arrangements accommodate variation in human capabilities while preserving meaningful task accomplishment, understandable control, dignity, and participation. This monograph traces the transition from generalized ergonomics and universal usability to participatory, disability-informed, and ability-based design. It synthesizes research on person–task–system fit, sensory and motor accessibility, cognitive and linguistic access, assistive technologies, adaptation, usability measurement, and sociotechnical governance. Original experiments in personalized motor interfaces and task-based evaluations with blind screen-reader users demonstrate both the value of tailored interaction and the insufficiency of standards conformance as a substitute for observed access. Qualitative research documents the social consequences of assistive technology use. The resulting evidence supports a multilayered evaluation model that measures successful goal completion, effort, recoverability, independent agency, and real-world participation across diverse people and contexts. The Artificial Era extends these questions to adaptive and agentic interfaces, where multimodal access, user control, and accountability must remain testable.
Keywords: inclusive human-system design; functional accessibility; human factors; inclusive design; universal usability; ability-based design; disability; assistive technology; human–computer interaction; accessibility evaluation; cognitive accessibility; user-centered design; participatory design; equitable task success; adaptive interfaces; Artificial Era.
The Research Object and Its Explanatory Scope
Inclusive human-system design investigates the relation between a person's functional capabilities and the demands of a technical or sociotechnical system. Its central question is whether people with differing sensory, motor, cognitive, linguistic, developmental, and situational profiles can accomplish meaningful goals through that system with acceptable effort, dependable feedback, effective recovery, and appropriate autonomy. The object is a functional interaction, observed across a specified task, environment, time, and organizational setting. An interface may satisfy numerous design requirements while still producing failed transactions, misinterpretations, exclusion from essential workflows, or dependence on assistance that the user did not choose.
The research domain lies within human factors and engineering psychology. It draws on experimental psychology, rehabilitation, disability studies, cognitive ergonomics, psychophysics, linguistic research, human–computer interaction, and sociotechnical analysis. The World Health Organization (2001) establishes a useful ecological premise: functioning and disability are understood in relation to contextual factors. The corresponding engineering question concerns how environmental constraints, interface requirements, training conditions, social arrangements, and available aids change what a person can actually do.
This dossier specifically addresses system-wide accommodation and its functional verification. General methods of human-centered development are the primary subject of Human-Centered Design, Usability, and Iterative Evaluation; sensory thresholds belong to Applied Psychophysics and Sensory Accessibility; motor integration belongs to the broader sensorimotor research tradition; spatial access and community participation belong to Accessibility, Inclusive Environments, and Environmental Participation. The present monograph integrates their relevant mechanisms around the distinct outcome of human-system task accessibility. It examines the complete path from a person's intention to a reliable outcome, including preparation, perception, input, interpretation, response, correction, and consequential completion.
Historical Formation of Inclusive Interaction Research
Early ergonomic research sought to match physical equipment, information displays, and controls to human performance. Experimental analyses of reaching, selection, signal identification, and workload eventually informed technologies used at work and at home. Fitts, P. M (1954) in 1954 offered a quantitative account of the relation between movement demands and performance under specified laboratory conditions. Its subsequent adoption in interface engineering made target dimensions and distance explicit design variables. Yet performance functions based on particular participants and apparatus do not define the full human capability distribution.
As interactive computing expanded, design for a hypothetical typical operator proved insufficient. Disabilities, age-related changes, temporary impairments, differing levels of literacy, and variable physical situations exposed the consequences of assumptions embedded in controls and software. Accessible computing evolved alongside disability-rights movements and demands for direct participation by people who experienced exclusion. In Newell, A. F., & Gregor, P (2000), Alan Newell and Peter Gregor argued for collaborative methods that include people with disabilities within the design process rather than treating disability as an afterthought.
The next phase connected inclusive design to systematic engineering and empirical validation. Richards, J. T., & Hanson, V. L (2004) proposed a broader understanding of web accessibility and the populations who benefit from it. Gajos, K. Z., Wobbrock, J. O., & Weld, D. S (2007) demonstrated that interface generation could incorporate individual motor performance and visual needs. Wobbrock, J. O., Kane, S. K., Gajos, K. Z., Harada, S., & Fr (2011) then articulated ability-based design as a research program. Parallel advances in web accessibility standards provided shared technical specifications, while user evaluations exposed important differences between coded compliance and successful use.
The intellectual history is therefore cumulative. Psychophysical measurement contributes testable limits; ergonomics describes demand and fit; human-centered design supplies an iterative development process; disability-informed inquiry contributes lived experience and rights-based constraints; and accessible computing tests whether software and devices can reorganize themselves around people. Each tradition addresses a different level of the same human-system relation.
Functional Accessibility as a Scientific Construct
Functional accessibility can be defined operationally as the degree to which specified people, with their actual capabilities and chosen supports, can achieve specified goals through a system within a specified context. The outcome depends on task requirements, interaction pathways, technical compatibility, contextual resources, and the person's preferences. The relevant measurement unit is not an isolated person or feature but an encounter between people, technology, tasks, and circumstances.
Three features are essential. First, accessibility is conditional: the same individual may experience markedly different results using a large screen, a phone in bright light, a voice interface in a noisy room, or a switch-access device during fatigue. Second, accessibility is relational: an interaction barrier arises where task demands and available means fail to fit, even when both the person's abilities and the component's engineering specifications remain unchanged. Third, accessibility is consequential: a nominally accessible control has limited practical significance when subsequent identification checks, payment screens, timed sessions, or support channels prevent completion of the actual goal.
The International Organization for Standardization (2018) defines usability with reference to specified users, goals, effectiveness, efficiency, satisfaction, and context of use. Accessibility research shares this emphasis on contextual outcomes and extends scrutiny to people who are frequently excluded by default interfaces. The resulting construct is multidimensional. Goal success, time, effort, error consequences, reliance on assistance, subjective experience, control over disclosure, and the ability to recover from breakdowns can vary independently. A system can therefore improve average response time while becoming inaccessible to a minority of users, or improve formal keyboard coverage while retaining a confusing workflow.
A rigorous research claim must identify its level: accessible information, operable interaction, end-to-end task success, sustained independent use, or equitable opportunity across contexts. Calling all these outcomes simply accessibility without specifying the estimand invites misleading comparisons and weak measurement.
The Person–Task–System–Context Configuration
A useful analytic model treats human-system access as a configuration containing at least five interacting elements: the person's available capabilities, the demands of the goal, the implemented interface and service, external supports and assistive technologies, and the environment in which the activity occurs. Institutional policies and change over time modify every element. The model is descriptive and testable rather than a claim of a newly discovered psychological mechanism.
Capabilities may include visual acuity, contrast sensitivity, auditory discrimination, motor accuracy, response speed, working-memory resources, linguistic comprehension, learned expertise, and fatigue tolerance. A task may require reading instructions, distinguishing alternatives, entering a code, manipulating a small target, sustaining attention, or explaining an error. A system imposes input devices, timing rules, terminology, authentication sequences, navigation structures, error messages, and feedback conventions. Context supplies lighting, noise, network stability, social privacy, financial resources, and access to support.
This configuration generates concrete hypotheses. Increasing touch-target size should reduce errors disproportionately among people who experience high motor-control costs under the smaller-target condition. Providing an alternative to a memory-intensive verification sequence should benefit users whose task performance is constrained by retention demands. Adding captions to spoken information should improve completion when auditory access is limited, provided that captions remain accurate and comprehensible. These are domain-specific predictions requiring controlled comparison; their effect sizes and generality must be established empirically.
The configuration also protects against misplaced causal attribution. A user who repeatedly fails an online form may be facing invalid labels, unpredictable focus movement, timeouts, confusing instructions, and a changing keyboard. Assigning the entire failure to visual impairment loses the properties that design can change. Proper analysis estimates both person-level variation and modifiable system-level contributors.
Disability, Functioning, and Environmental Demands
The World Health Organization (2001), endorsed by the World Health Assembly in 2001, describes functioning through health-related domains while explicitly incorporating environmental factors. In human factors, that framework encourages investigation of how a condition, an assistive device, and a particular design collectively shape performance. Disability status can be important for recruitment, access needs, and equity analysis, yet a diagnostic category alone rarely supplies a precise prediction of interaction capability.
Two people with the same broad impairment classification may prefer different modalities, possess different expertise, and experience different contextual barriers. A person with limited hand movement may use voice, eye gaze, switch scanning, keyboard shortcuts, or a personalized pointing interface depending on the setting. A person with low vision may use enlargement, speech output, contrast adjustment, or combinations. Disability-inclusive measurement therefore needs directly observed access requirements and user-defined goals rather than an assumed single profile.
The social model of disability directs attention to built, digital, institutional, and attitudinal obstacles. The biomedical and functional traditions investigate individual capacities and health conditions. A scientifically adequate human-system account can incorporate their evidence without reducing the phenomenon to either a diagnosis or a slogan. It asks which interactions are constrained, how design contributes, how support changes outcomes, and whether people can determine the form and degree of accommodation themselves.
This analytic stance includes temporary and fluctuating conditions. Pain, injury, medication effects, pregnancy, recovery, sleep loss, environmental stress, or sudden interruption may change the interaction resources available at a given moment. Age, developmental transitions, and experience produce further heterogeneity. Functioning remains the empirical object, while clinical classification supplies relevant context when warranted.
Universal Design, Inclusive Design, and Ability-Based Design
Universal design expresses an ambition to make environments and products broadly usable without unnecessary special adaptation. Inclusive design emphasizes the breadth and diversity of actual human requirements, recognizes exclusion, and involves affected users in developing and evaluating solutions. Ability-based design more specifically reorganizes interface design around users' demonstrated capacities and potential, using the system as the object of adaptation. These frameworks overlap in goals but differ in methods and the scientific claims they support.
The contribution of Newell, A. F., & Gregor, P (2000) was methodological: designers pursuing universal usability should engage people with disabilities directly and collaboratively. Wobbrock, J. O., Kane, S. K., Gajos, K. Z., Harada, S., & Fr (2011) developed seven ability-based design principles and illustrated their application through accessible-computing projects. Their proposal makes individual capabilities a first-order design input and treats the interactive system as adjustable. The concept is a theoretical and engineering framework, not evidence that every adaptive solution will outperform a well-designed universally accessible interface.
An important implication follows. A single fixed interface may provide excellent access to many tasks through standards-based semantics, keyboard functionality, flexible text, and multiple modes. Adaptive alternatives become especially valuable when capability differences require conflicting spatial layouts, input mechanisms, or temporal demands. The correct choice depends on observed task performance, users' preferences, cost of adaptation, and stability over time.
Effective inclusive design therefore combines shared baseline access with optional personalization. A common structure supports predictability and interoperability; meaningful alternatives accommodate variation. Neither uniformity nor personalization is a universal scientific optimum. Both must be evaluated through the outcomes they enable and the costs they introduce.
Usability, Accessibility, Agency, and Participation
Usability is conventionally assessed through effectiveness, efficiency, and satisfaction in a defined context, as summarized in International Organization for Standardization (2018). Accessibility asks which people can enter and complete the interaction at all, and under which conditions. Agency concerns the person's ability to select goals, understand consequences, control action, refuse assistance, and revise choices. Participation describes the wider activities and opportunities made available through successful interactions. These dimensions overlap but have different scales of analysis.
Consider a transport booking system. The interface may display readable options and accept keyboard input, satisfying important design requirements. A participant using speech navigation might nevertheless encounter a payment confirmation whose security code expires before it can be entered. The relevant failure occurs at the complete-service level, not solely at the screen-reader compatibility level. Conversely, a person might finish a booking with substantial unsolicited help; counting that outcome as simple success would obscure independence and privacy costs.
The relationship between efficiency and access is also conditional. A slower route chosen freely because it provides better orientation, greater confidence, or an alternative input method can be a superior functional solution. Evaluators should not turn a single speed metric into a universal value judgment. They can instead record completion, duration, effort, frustration, perceived control, and assistance separately, while making tradeoffs explicit.
Research should locate the unit at which participation is claimed. An accessible instructional page is one component; accessible enrollment, assessment, support, and certification are a larger educational process. Such distinctions permit more meaningful linkage to rehabilitation psychology and environmental participation while preserving the specific engineering question at the center of this dossier.
Sensory Perception and Redundant Information Channels
Systems transmit goals, states, hazards, and instructions through visual, auditory, tactile, and linguistic signals. Sensory accessibility concerns whether information can be detected, discriminated, interpreted, and used under realistic conditions. Psychophysical threshold research supplies indispensable knowledge about visibility, sound, timing, and discrimination, while inclusive design organizes the information into alternate pathways for accomplishing tasks.
A visual alert may be readable to users with typical contrast sensitivity yet hard to recognize under glare, color-vision variation, low vision, or visual fatigue. An audible notification may be effective in quiet surroundings while failing in noise or when hearing access differs. Redundant presentation through text, speech, vibration, and visual changes can increase the number of accessible routes; however, concurrent signals can also create distraction or overload. Engineers must test information equivalence, timing, and interference for the tasks at hand.
The WCAG 2.2 Recommendation describes requirements for digital content organized around perceivability, operability, understandability, and robustness. This standard supplies operational design constraints such as text alternatives and accessibility of controls. It does not supply a complete theory of the perceptual psychology involved in every task. A compliant message may remain too complex, too late, or too poorly coordinated with the user's activity to support actual decisions.
Evaluation should therefore include recognition accuracy, comprehension, response appropriateness, delay, and availability of alternatives, not only a technical inventory of output modes. When high-stakes or time-dependent action is required, alert information must be assessed in context, with realistic competing demands and consequences of missed or misunderstood signals.
Motor Interaction and the Distribution of Control Demands
Pointing, dragging, tapping, typing, gesture recognition, switch activation, and speech-triggered action require different motor capabilities and impose different error costs. Human motor control research, including Fitts, P. M (1954), explains why movement time depends in part on target amplitude and accuracy demands. The inference relevant to inclusive design is that control geometry, input modality, feedback, and task structure can change the amount of performance support a system provides.
Motor accommodation can involve larger targets, alternative commands, adjustable dwell periods, configurable gesture thresholds, simplified sequences, reduced need for simultaneous key presses, and resilience to inadvertent activation. The target should remain functional within the whole task. An enlarged button offers limited benefit when subsequent workflow steps still require precise dragging or a rapid response under rigid time limits.
Ability-based interface research has tested a more extensive approach. Gajos, K. Z., Wobbrock, J. O., & Weld, D. S (2007) developed interfaces generated from observations of individual motor performance, also considering visual needs. Their pilot evidence suggested that model-guided adaptations could change task feasibility and speed; its exploratory nature limits population-level claims. Their Gajos, K. Z., Wobbrock, J. O., & Weld, D. S (2008) provided stronger task-specific experimental support for automatically generated motor-accessible interfaces.
Motor accessibility further depends on fatigue, pain, posture, positioning, and device mounting. In a laboratory, pointing performance may be measured across repeated tasks using a stable workstation. In everyday life, the same action may occur while standing, using public transit, or coordinating assistive devices. Representative sampling and setting-specific tests are essential when converting psychomotor findings into general accessibility recommendations.
Cognitive Accessibility and Executive Demands
Cognitive accessibility concerns the match between task demands and the resources required for comprehension, memory, attention, planning, error detection, and decision-making. People with learning disabilities, acquired cognitive changes, intellectual disabilities, neurodivergent profiles, age-related variation, and situational cognitive load may experience substantially different difficulties with the same interface. Operational requirements should be grounded in the actual activities users undertake.
Interfaces commonly require users to remember instructions from earlier screens, infer hidden rules, interpret unfamiliar jargon, retain codes, switch tasks, and recover from ambiguous failures. These demands create measurable opportunities for exclusion. Progressive disclosure, stable navigation, visible instructions, explicit confirmation, meaningful feedback, and forgiving recovery can reduce unnecessary demands. Each intervention must still be evaluated: collapsing information may conceal context required by experts, while additional prompts may slow routine use.
The W3C Working Group Note World Wide Web Consortium (2021) supplies user needs and supplemental design patterns beyond the mandatory success criteria of WCAG. Its status matters: the document is guidance and should not be described as a separate required conformance level. This distinction illustrates how a technical standard and a broader empirical account of human comprehension can complement one another.
A rigorous evaluation distinguishes recognition of an interface element from comprehension of its implications. For a medication portal, understanding the label of a button is different from correctly interpreting an instruction; for an education service, locating an assignment is different from understanding what constitutes successful submission. Tasks should be designed around consequential comprehension, with ethical safeguards when health or safety decisions are involved.
Linguistic Access, Literacy, and Communication
Language is part of the interaction mechanism. A person may have adequate visual and motor access to every control yet be excluded by syntactic complexity, unexplained technical terms, inconsistent terminology, culturally unfamiliar examples, or instructions that rely on implicit institutional knowledge. Functional accessibility requires comprehension of actionable content at the point where decisions are made.
Research designs should include reading and comprehension tasks appropriate to the user population, rather than treating language preference as a proxy for proficiency. Plain language can improve directness, but a superficial readability score does not establish conceptual understanding. Relevant measures include accurate paraphrase, recognition of consequences, successful completion without prompting, and ability to locate clarification. Human translation and interpretation quality are distinct variables when multilingual operation is offered.
Multimodal communication adds further opportunities and risks. Speech output, captions, symbols, text, signed-language services, and haptic alerts can support different capabilities, but each pathway requires fidelity testing. Automatic captions that omit a negation can change the instruction's meaning; simplified content can omit a critical condition; machine translation can misrepresent eligibility or risk. Accessibility is therefore also an information-quality problem.
The interface should make language selection and explanation available without stigmatizing the user, and it should preserve equivalent essential information across available modes. An empirical study can compare task completion and comprehension by condition and by participant characteristics while recording the costs of switching representation.
Aging, Development, and Changing Access Needs
Functioning changes with development, experience, illness, environment, and aging. These changes rarely proceed uniformly across capabilities. Designers should therefore avoid treating age as a substitute for detailed assessment, while recognizing that age-related variation in sensory, motor, and cognitive performance can influence system fit.
Developmental accessibility raises questions about the match between task requirements and children's evolving comprehension, reading ability, fine-motor control, and reliance on caregivers. Older adults may bring considerable domain expertise while encountering visual or motor barriers created by small targets, low-contrast information, or inflexible authentication. Both populations are heterogeneous; meaningful evaluation requires sampling within them rather than making assumptions from broad categories.
The World Health Organization & United Nations Children's Fund (2022) documents the importance of assistive products across the life course and highlights resource barriers. Its estimates concern population-level need for assistive technology; they are not evidence that a particular interface intervention is effective. Developmental research and long-term field observation are necessary to understand how technology needs and preferred accommodations evolve.
A system designed for longitudinal use should accommodate changes in input devices, visual presentation, cognitive support, expertise, and account management without forcing disruptive loss of learned navigation. The relevant engineering question extends beyond initial usability to continuity of access across transitions.
Intersectional and Situational Variation
Functional access depends on combinations of characteristics and conditions. Sensory limitation may coincide with motor impairment, low bandwidth, economic barriers, limited privacy, language differences, or the need to use a shared device. Studying each factor separately can overlook interactions that explain why a system works in one setting and fails in another.
Intersectional analysis requires thoughtful measurement and sufficient sampling for the comparisons being claimed. Researchers should not convert every identity category into a fixed capability assumption. Instead, they can describe the circumstances, device configurations, support needs, and task requirements that plausibly alter outcomes, then test relevant interactions when sample sizes and study design permit.
Situational constraints can imitate or intensify disability-related access barriers. Glare reduces visibility; noise undermines speech recognition; a temporary hand injury limits precise input; emotional stress degrades memory and attention; an unstable network interrupts time-sensitive forms. A flexible interface may support both enduring accommodations and ordinary fluctuations in capability.
The scientific implication is that generalizability depends on the tested distribution of people and contexts. An average usability statistic from volunteers using ideal equipment provides little evidence about users of low-cost devices in crowded, poorly connected, or highly interrupted environments. Representative conditions are therefore part of the construct, not merely logistical background.
Assistive Technology as Part of the Interaction System
Assistive technology encompasses devices and services that support communication, mobility, cognition, hearing, vision, manipulation, and other functional activities. In digital environments this includes screen readers, speech recognition, magnification, switch and eye-gaze input, alternative keyboards, captions, text prediction, and augmentative communication systems. Its effectiveness depends on compatibility with the application and the surrounding workflow.
A screen reader interacts with semantic structure, labels, focus order, and application state. A keyboard user depends on predictable movement among controls and an operable escape from complex components. Switch scanning requires temporal properties compatible with the user's input capability. Thus an accessibility failure may arise at an interface boundary even when the application or assistive product appears technically adequate in isolation.
The World Health Organization & United Nations Children's Fund (2022) estimates that more than 2.5 billion people globally require at least one assistive product, while nearly one billion lack access. These are worldwide, report-specific estimates as of 2022, covering a broad spectrum of products; they should not be equated with the number of people who need digital screen readers. Distribution, procurement, affordability, local support, training, repairs, and interoperability materially affect realized access.
Field studies should record device type and version, user expertise, setup, compatibility problems, costs, assistance, and changes over time. Excluding these factors from the analysis produces unrealistic confidence in the transferability of an interface-level intervention.
Personalization, Adaptation, and User Authority
Adaptation adjusts interface presentation, input methods, control arrangement, or workflow to better match a user's capabilities and preferences. Explicit settings permit predictable user-directed changes. Automatically generated interfaces can use performance measures, inferred preferences, or learned models. The central scientific question is whether adaptation improves usable goal achievement without creating new uncertainty, surveillance costs, or loss of control.
Gajos, K. Z., Weld, D. S., & Wobbrock, J. O (2010) described the Supple system for generating personalized interfaces and formalizing tradeoffs. Their program illustrates how optimization can use information about individual users to select different interface configurations. Yet a model trained on one session may perform poorly when fatigue, device positioning, medication, or environmental demands change.
Personalization can also introduce errors in users' mental models of the interface. If controls move or terminology changes unexpectedly, learned navigation may fail. A system should consequently support stable alternatives, preview or explanation of changes, easy restoration of prior settings, and recovery when adaptation proves unsuitable. These features have to be evaluated through longitudinal tasks, not inferred from the presence of a personalization option.
User authority is a measurable design property. A participant should be able to choose among accommodations, reverse automated changes, understand what data are used to customize the interface, and preserve essential functionality without disclosing more information than necessary. Adaptation constitutes a benefit when it expands effective choices while maintaining understandable control.
Agency, Privacy, Stigma, and the Social Experience of Access
An accommodation can produce instrumental benefit and social cost at the same time. Assistive devices may make a task possible while attracting unwanted attention, creating privacy concerns, or triggering misconceptions about competence. Accessibility research must therefore examine the person's broader experience of using the system in social and institutional settings.
Shinohara, K., & Wobbrock, J. O (2011) conducted qualitative interviews with 20 people about assistive technology and social interaction. Their analysis identified social misconceptions surrounding device use, alongside the practical value of the technology. This qualitative evidence reveals mechanisms and experiences that a simple task-time experiment would miss. It does not establish their population prevalence or the effect size of any proposed remedy.
Design can reduce imposed disclosure by making accessible modes ordinary and available to everyone, avoiding unnecessary special registration, and supporting private interaction. Organizations should also allow accommodation without presuming dependence or incapacity. A system that works only when the user must publicly announce a condition may impose a participation cost even where technical functionality exists.
Measures of agency include whether participants can choose the pathway, revoke permission, correct system inferences, finish independently when they wish, and decline intrusive assistance. Self-report can capture dignity and perceived control; observation can capture unsolicited intervention, delays, and disclosure events. Combining methods clarifies how access is experienced as well as how it is technically executed.
Error Prevention, Feedback, and Recoverability
Human-system interactions produce errors through slips, misunderstandings, inaccessible controls, timing demands, inconsistent information, and changing contexts. Functional accessibility includes the capacity to detect errors, understand their consequences, and recover without disproportionate cost. Preventing all errors is seldom achievable; recoverability is a distinct engineering outcome.
Accessible feedback should state what occurred, where correction is possible, and what action remains available. A form that merely announces an error without identifying the field imposes a search burden. A transaction that times out without saving progress disproportionately affects people who need additional time, use sequential input, or must switch assistive modalities. The severity of a barrier depends on its place in the task chain.
Empirical evaluation can distinguish first-attempt success from success after correction, recording total attempts, repeated failures, lost work, assistance needed, and emotional responses. Severity should consider the real consequence of failure: a missed appointment, financial loss, inaccessible public service, or an irreversible action. Time alone is insufficient.
Recovery design also supports learning. Clear feedback can help users form accurate expectations about system behavior, while unpredictable focus shifts and invisible status changes undermine the ability to plan subsequent actions. The goal is a system whose state and recovery options remain legible across the supported sensory and input pathways.
Standards-Based Accessibility and the Meaning of Conformance
Standards establish a common language for requirements, testing, procurement, and accountability. The Web Content Accessibility Guidelines 2.2 specify web-content success criteria organized into four principles. The W3C first published WCAG 2.2 as a Recommendation on October 5, 2023, followed by an updated Recommendation on December 12, 2024. The International Organization for Standardization & Internation (2019) code of practice addresses accessible ICT development across organizational and system life cycles, including procurement and maintenance. International Organization for Standardization (2019) addresses human-centered design activities across interactive-system development.
These documents have different scopes. WCAG specifies testable criteria for web content and related accessibility practices; ISO 9241-210 describes lifecycle processes; ISO/IEC 30071-1 addresses organizational accessibility governance. Compliance with one does not automatically satisfy the others, and none independently measures every user's successful completion of an end-to-end task.
A credible evaluation records both conformance and observed function. Automated scanners can detect some missing semantics or contrast failures, but they do not substitute for assessing comprehension, task sequencing, recovery, contextual compatibility, or social consequences. Manual expert evaluation and user testing complement automation. World Wide Web Consortium (n.d.) reflects this methodological distinction.
Standards are necessary for reproducible requirements and consistent implementation. Their empirical validity must be assessed against actual barriers. The task is to connect requirements to outcomes, identify where requirements fail to predict outcomes, and revise design or assessment accordingly.
Empirical Evidence on Guideline Coverage and Blind Users' Tasks
The study by Power, C., Freire, A. P., Petrie, H., & Swallow, D (2012) is central to distinguishing technical criteria from observed accessibility. Thirty-two blind participants performed tasks across sixteen websites. The researchers documented 1,383 instances of user problems and reported that only 50.4 percent of those instances were covered by WCAG 2.0 success criteria. The work directly evaluated tasks instead of relying solely on automated or expert judgments.
Its central conclusion is a measurement distinction. A standards checklist describes selected properties of content; task-based observations can reveal disorientation, interaction breakdowns, recovery difficulties, and other failures beyond that checklist's coverage. The result provides evidence for combining guideline assessment with participant-centered evaluation. It does not establish that all later versions of WCAG, including 2.2, cover precisely the same proportion of problems, nor that the study sites represent all websites or all blind users.
The investigators also reported cases in which sites implemented techniques recommended by the guidelines while user problems remained. This observation cautions against interpreting the presence of a recommended technique as proof of effective interaction. Differences among screen readers, site structure, task goals, and users' strategies contribute to the relationship between design and observed performance.
The practical methodological consequence is to define concrete tasks, sample participants with relevant access modes, record failures systematically, map each failure to available criteria, and separately analyze uncovered barriers. Standards and observed interaction are complementary sources of evidence, with different units of analysis and limitations.
Screen-Reader Frustration and Ecological Experience
Lazar, J., Allen, A., Kleinman, J., & Malarkey, C (2007) studied web frustration using time diaries from 100 blind screen-reader users. The reported sources of frustration included confusing page layouts, conflicts between screen readers and applications, problematic or unlabeled forms, absent image alternatives, misleading links, inaccessible documents, and software failures. Unlike a narrowly controlled laboratory task, the diary method captured difficulties as users experienced them in everyday contexts.
This design offers ecological information that typical conformance audits may miss. The participants could report actual breakdowns across different sites and tasks, and the research could reveal recurring categories of problems. Its constraints include self-report and diary compliance, variation in user devices and skills, and an inability to isolate causal effects of particular design changes. A frequent diary complaint is a signal for further research, not automatically a quantified effect of a specific interface feature.
Read together with Power, C., Freire, A. P., Petrie, H., & Swallow, D (2012), the diary evidence shows why broad problems and controlled task failures should be studied together. Diaries reveal problems that matter to users over time; task studies show how problems unfold on specific workflows; technical audits help identify implementation conditions. Triangulation permits more defensible inference than any one method alone.
The studies also have historical scope. Their technologies, browser environments, and standards reflect the periods in which they were conducted. Their lasting scientific contribution is the method of examining real users' problems, while contemporary estimates require renewed testing with current interfaces and assistive technologies.
Experimental Evidence for Ability-Based Interfaces
Gajos, K. Z., Wobbrock, J. O., & Weld, D. S (2008) compared automatically generated interfaces with baseline interfaces for users with motor impairments. One approach personalized from elicited preferences, while another used measured motor capabilities. In the study's reported comparisons, participants with motor impairments completed tasks 26.4 percent faster with the capability-modeling system and made 73 percent fewer errors than with the manufacturer's default interfaces. They also expressed stronger preferences and reported reduced physical burden.
These findings establish a task-specific empirical demonstration that system-generated accommodation can improve performance and experience for sampled users. The experimental comparison is stronger causal evidence for those conditions than a design rationale or a satisfaction-only survey. It also directs attention to the kind of personalization performed: measures of motor performance were used to inform the system's layout decisions.
Interpretation must remain bounded. The results concern participants, tasks, devices, and alternative conditions examined in that study. They do not establish a population-wide effect size for all interface technologies, or a guaranteed improvement for people with cognitive, sensory, or multiple disabilities. The difference between preferences and measured motor ability is important but does not imply that measured ability should always supersede user choice.
The preceding Gajos, K. Z., Wobbrock, J. O., & Weld, D. S (2007) provided earlier, exploratory evidence. The subsequent Gajos, K. Z., Weld, D. S., & Wobbrock, J. O (2010) developed the engineering methodology further. Together, these studies support the feasibility of systems that adapt to users, while leaving open questions about longitudinal adaptation, privacy, model drift, maintenance, and transfer to different classes of tasks.
Qualitative Evidence and the Limits of Performance-Only Evaluation
A narrow experimental outcome can underrepresent the full consequences of a technical intervention. Faster task completion may coincide with higher stigma, forced disclosure, greater setup time, or reliance on specialized equipment. Likewise, a device that attracts attention may still be a participant's preferred solution because it supports effective independent action. Research needs methods that can represent these tradeoffs.
The Shinohara, K., & Wobbrock, J. O (2011) investigated social and professional contexts of assistive technology. Its contribution is explanatory depth: interviews made visible how technology use interacts with others' perceptions, expectations of competence, and interpersonal arrangements. This evidence adds a dimension to functional accessibility that is often missing from interface performance studies.
Qualitative samples are typically selected for analytic relevance rather than population prevalence estimation. Their value lies in revealing processes, categories, contradictions, and overlooked variables. A robust mixed-method program can use interviews to identify barriers, task experiments to test whether design changes affect performance, and longitudinal fieldwork to examine the stability of gains.
Researchers should therefore predefine the type of claim each method can support. A narrative about stigma does not by itself establish how widespread it is. A numerical speed improvement does not prove that users will accept an interface. Integrating these findings is a matter of matching evidence to questions, not reducing different outcomes to a single score.
Task Analysis and the Structure of Evaluation
Evaluation begins with a well-defined user goal and a representative task pathway. Researchers should map prerequisites, action sequences, decisions, feedback, possible errors, and dependencies on external systems. This makes it possible to identify where participation breaks down and which barriers are produced by design rather than by the user's selected strategy.
Task decomposition should include authentic setup and recovery costs. For example, account creation, authentication, permissions, input-device configuration, help retrieval, and confirmation can impose more access burden than the central content screen. End-to-end assessment must include these stages whenever they are necessary for the intended outcome.
A credible protocol specifies who the participant is, what assistance is permitted, which assistive tools are used, what counts as success, how errors are classified, and how interruptions are handled. It also records whether the activity is novel or familiar. Expertise can modify effects markedly: a keyboard shortcut may help practiced users but impose learning demands on first-time participants.
Field observation adds institutional and contextual detail to a controlled protocol. Actual tasks may be interrupted, delegated, completed collaboratively, or transformed by rules not visible to laboratory designers. A rigorous research program therefore moves iteratively between controlled experiments and situated observation, measuring both mechanism and representativeness.
Measuring Effectiveness, Effort, and Equity
The primary functional outcome should be linked to the goal: successful submission, information retrieval, decision comprehension, independent completion, accurate response, or safe task execution. Secondary measures can include time, error type, number of attempted pathways, correction cost, physical effort, cognitive workload, perceived control, satisfaction, and need for assistance. The International Organization for Standardization (2018) distinction among effectiveness, efficiency, and satisfaction provides a recognized starting framework.
Group-level averages can conceal exclusion. Suppose a new interface saves time for most participants yet prevents a small subgroup from completing its tasks. The mean may improve while access becomes less equitable. Analyses should report disaggregated success and failure where ethically and statistically possible, and examine the tails of performance distributions rather than relying solely on average duration.
It is also important to distinguish absolute and conditional success. A participant who abandons the task should not disappear from time analysis without explanation. An interface that makes a task impossible cannot be adequately evaluated by calculating speed only among those who succeed. Alternative handling of abandoned attempts, censoring, and missing data should be specified in advance.
For an intervention study, useful estimands may include the difference in completion probability between interface conditions, the change in errors among successful attempts, and the interaction between the intervention and a measured capability variable. Confidence intervals and uncertainty estimates should accompany numerical findings when original data permit. The present monograph does not pool study-level effect sizes, because the available designs and outcomes do not constitute a preregistered systematic meta-analysis.
Sampling, Recruitment, and Disability-Inclusive Research Ethics
Sampling determines which claims an accessibility study can sustain. Recruiting a small convenience sample of expert assistive-technology users can be valuable for discovering complex barriers, but it cannot establish the access experience of novice users or people without reliable devices. Conversely, excluding expert users eliminates valuable evidence about advanced interaction pathways and workarounds.
Inclusive recruitment requires explicit consideration of motor, perceptual, cognitive, linguistic, developmental, and multiple-access profiles relevant to the target task. It also requires accommodations within the research procedure itself: accessible consent, communication alternatives, scheduling flexibility, appropriate compensation, and usable reporting channels. Researchers must avoid imposing a single testing modality that excludes the population whose needs they intend to study.
Informed consent and data minimization are especially important where participants share diagnostic information, accessibility preferences, biometric performance, or detailed assistive-device configurations. Researchers should distinguish needed capability data from unrelated medical history, provide appropriate controls over recordings, and explain how performance data will be used for interface adaptation.
Participation in study governance can improve the relevance of research questions and the interpretation of results. The World Wide Web Consortium (n.d.) is a practice-oriented complement to experimental design. Co-design and consultation should also be evaluated critically: nominal participation that lacks influence over decisions provides weaker assurance than sustained collaboration with documented consequences for the design.
Experimental Causality and Alternative Explanations
An accessible interface may outperform a comparator because its targets are larger, its workflows are shorter, its language is clearer, its feedback is more visible, or its participants received additional training. When these design elements vary together, the experiment estimates the effect of the complete intervention, not the effect of every component separately. Causal analysis must match the actual manipulation.
Randomized or counterbalanced comparisons can help separate interface effects from practice and order, provided that carryover is managed. Within-participant designs reduce some between-person variation but may be complicated by fatigue, adaptation, and changes in strategy. Between-participant designs avoid certain carryover effects but often require larger and carefully characterized samples.
Assistive technologies introduce additional technical confounds. Screen-reader version, browser, keyboard layout, input latency, display size, and operating-system accessibility services can interact with the experimental condition. Researchers should document configurations, test relevant combinations, and avoid treating software compatibility as a fixed property independent of updates.
Construct validity is equally important. A timed button-selection task provides evidence about a motor interaction, but it only approximates the broader construct of functional accessibility. Tests must link local task results to meaningful goals; claims about participation require additional contextual evidence. Distinguishing proximal performance from distal consequences preserves scientific precision.
Evidence Strength, Replication, and Transferability
The evidence base contains multiple levels of support. The Gajos, K. Z., Wobbrock, J. O., & Weld, D. S (2008) experiment offers intervention-specific evidence of improved motor-interaction performance for the studied users and tasks. Power, C., Freire, A. P., Petrie, H., & Swallow, D (2012) provide task-based observational evidence that WCAG 2.0 criteria did not cover all difficulties encountered in their study. Lazar, J., Allen, A., Kleinman, J., & Malarkey, C (2007) provide ecological diary evidence about sources of screen-reader frustration. Shinohara, K., & Wobbrock, J. O (2011) provide qualitative evidence on social processes around assistive technology.
These results are informative because they examine distinct aspects of the problem, and they should not be presented as a homogeneous body of trials. None is a universal effect estimate for inclusive design as a whole. Their samples, tasks, technologies, and outcomes differ. A pooled percentage of improvement would be scientifically misleading without an appropriate common outcome and systematic synthesis protocol.
Replication requires more than repeating a demonstration under the same conditions. External validity can be tested by different teams, more varied capability profiles, new devices, contemporary interfaces, longitudinal follow-up, and settings with different resources. Researchers should report null and adverse results as carefully as positive ones. Publication selection and proprietary system constraints may otherwise create an overly favorable picture of interventions.
The central empirical pattern is plausible and supported across methods: changing system requirements and accommodating capability diversity can affect task feasibility, performance, and experience. The magnitude and durability of specific benefits must be established separately for each intervention and population.
Organizational Systems, Procurement, and Maintenance
Accessibility is produced across a product or service life cycle. Requirements are established in procurement; design and development determine structural affordances; content teams create information; support staff mediate recovery; maintenance changes compatibility; and organizational policies control time limits, identity checks, and available accommodation. A technically accessible screen can be defeated by an inaccessible support process.
International Organization for Standardization & Internation (2019) addresses accessible ICT at both organizational and system-development levels. This broader scope reflects a fundamental systems principle: functional access is affected by decisions made outside the interface itself. Vendor contracts, update testing, accessibility statements, staff training, and complaint resolution contribute to sustained outcomes.
A useful accountability model records which tasks must remain accessible, what alternatives are supported, who validates changes, how users can report barriers, and how unresolved failures are escalated. It also distinguishes laboratory acceptance from maintenance over time. Updates to operating systems or assistive technologies can change compatibility without any intentional redesign of the page.
Procurement evaluation should require demonstrations with representative tasks and supported assistive tools rather than relying solely on supplier declarations. Independent testing and user feedback can identify failure modes that formal self-assessments omit. The scientific question remains the same across organizational levels: does the system enable its intended population to achieve defined goals with acceptable costs and control?
Artificial Interfaces and Accessibility in the Artificial Era
The Artificial Era introduces systems that generate explanations, interpret speech and images, recommend actions, personalize workflows, and sometimes execute tasks through tools. These capabilities can extend accessibility when users gain alternative ways to express intentions, receive information, or coordinate complicated actions. They also introduce uncertainty about output accuracy, timing, privacy, and system authority.
Conversational interfaces may assist people who have difficulty with conventional navigation, but speech input is not universally accessible. It may fail amid noise, be unavailable for people with speech-related differences, compromise privacy in shared spaces, or produce incorrect interpretations. Image description can improve access to visual material while misidentifying details. An agent capable of submitting a form may increase task reach while creating new hazards if the user cannot inspect or reverse its actions.
Research should operationalize artificial assistance as a change in the functional workflow. Relevant outcomes include the rate of correct task completion with and without assistance, accuracy of user understanding, recoverability after wrong suggestions, accessibility of refusal and override, privacy costs, and how errors propagate when the system acts on external services. Human authority and control must be designed into the interaction and measured directly.
The conceptual horizon From Homo to Artificial places new nonbiological systems within the history of human interaction. The psychological evidence discussed here concerns observable human-system behavior and outcomes. Philosophical models of Artificial Sapiens may generate questions about agency and intersystem coordination, while claims about actual performance require experiments and accountable evaluation.
Adaptive Systems, Explanations, and the Risk of Invisible Barriers
Artificially adaptive interfaces may personalize information, timing, or control pathways based on estimates of a person's needs. Their performance depends on the accuracy of those estimates and on whether the person can understand, correct, or reject them. A mistaken inference about cognitive capability may oversimplify information; a mistaken inference about mobility may conceal useful controls; a failed speech transcription may trigger an inappropriate action.
Explanation should be tailored to the decision at stake. A user needs to know what the system plans to do, which information it used, what uncertainty remains, and how to intervene when necessary. The explanation itself must be accessible across supported modes. Long, generic descriptions may impose cognitive burden and obscure the decision point.
A robust design permits stable access without compulsory profiling, visible switching among modalities, correction of mistaken personalization, and a recoverable fallback if adaptive components fail. Audit logs can support accountable investigation, provided retention and access controls protect sensitive user information.
Experimental research should compare adaptive and stable alternatives under both normal operation and deliberately introduced errors. Outcomes should capture whether assistance improves users' goal success, whether it changes their awareness of consequential decisions, and how quickly they can regain control. This shifts evaluation from interface novelty to measured capability and agency.
Critical Scientific Discussion
The principal scientific challenge is scale. A technical criterion describes a component; a psychophysical threshold describes a sensory or motor relation; a task experiment measures performance under controlled conditions; a diary records experience; and a longitudinal study examines sustained opportunity. None alone represents the whole phenomenon of functional accessibility. A defensible synthesis must preserve those levels and connect them through explicit assumptions.
A second challenge concerns tradeoffs. Personalization can improve motor task performance while increasing setup cost; multimodal cues can broaden access while generating attentional interference; simplified language can reduce comprehension burdens while removing important distinctions; human assistance can permit success while diminishing privacy or autonomy. Research should specify which outcomes matter to the participants and institution, and examine conflicts rather than replacing them with a single accessibility score.
A third challenge is the relationship between rights and evidence. Accessibility obligations concern fair access and legitimate participation. Experimental results establish whether a particular intervention achieves particular outcomes. Ethical or legal requirements are not contingent on a laboratory showing that every accommodation raises average efficiency; similarly, compliance alone does not prove that an experience is usable. Clear differentiation of normative requirements and empirical findings strengthens both.
Finally, the research corpus is uneven. Well-documented demonstrations exist for some motor adaptations and task-based screen-reader problems. Evidence on longitudinal adoption, multi-impairment configurations, low-resource settings, AI-mediated interfaces, and organizational effects remains less integrated. Reported benefits should be carried forward as testable claims rather than treated as universal settled effects.
An Integrated Research Framework for Functional Accessibility
A complete research program can organize evidence into six interdependent layers. The first defines capability profiles and contextual constraints without reducing users to diagnostic categories. The second identifies task structure, goals, success conditions, and consequences. The third measures technical implementation and compatibility with available supports. The fourth records observable task outcomes: completion, accuracy, time, recoverability, effort, and assistance. The fifth measures agency, privacy, dignity, and acceptance. The sixth examines continuity of access and participation across time and institutional change.
Each layer supports distinct methods. Psychophysical experiments characterize performance limits; usability trials test interaction alternatives; qualitative interviews reveal neglected meanings and constraints; field studies document actual work; longitudinal designs estimate persistence and adaptation; organizational audits examine procurement and maintenance. The layers can be linked through triangulation without pretending they share a single measurement scale.
A practical evaluation might randomize a design alternative while stratifying by motor capability, measure end-to-end completion and errors, interview participants about autonomy and fatigue, and repeat selected tasks after weeks of use. Such a design could distinguish initial laboratory benefit from learned, sustained functional gain. It could also identify whether an apparently successful adaptation burdens a subgroup or creates an unacceptable privacy requirement.
This framework is an analytic synthesis of the reviewed literature, not a newly validated psychometric instrument. Its scientific utility depends on whether researchers can state each variable, choose appropriate measures, and subject its predicted relationships to independent testing.
Unresolved Questions and Research Directions
Several research questions should guide future work. How can accessibility outcomes be compared across different assistive modalities when tasks and costs differ? Which personalization strategies preserve predictable control over months or years? What sampling procedures adequately represent people with multiple or fluctuating access needs? How should end-to-end service failures be attributed when an inaccessible step occurs in a different organization? These questions require designs that link specific mechanisms to consequential outcomes.
The relationship between standards and user-observed problems deserves continued direct study. Power, C., Freire, A. P., Petrie, H., & Swallow, D (2012) concerned WCAG 2.0, particular websites, and a particular group of participants. Contemporary research should repeat analogous evaluations with WCAG 2.2, current browsers and assistive technologies, diverse tasks, and transparent sampling. It should analyze which failures remain outside criteria and which are implementation failures despite nominal coverage.
Adaptive and agentic technologies introduce additional questions: when does system inference help and when does it obscure user choice? Can multiagent workflows preserve accessible confirmation, override, and error recovery? Which explanations meaningfully support people with different cognitive and language needs? How does a person's reliance on automated assistance affect retained skill and perceived authority? These are empirical issues, not outcomes to be presumed from system architecture.
Progress requires public reporting of unsuccessful as well as successful designs, accessible study protocols, stronger longitudinal evidence, and collaboration with users whose functional needs are often overlooked. Scientific advance should be measured by improved explanation and demonstrable opportunity to accomplish real goals.
Conclusion
Inclusive human-system design is a scientific inquiry into how engineered and institutional arrangements support meaningful human action across capability diversity. Functional accessibility emerges from the alignment of people, goals, interfaces, assistive technologies, contexts, and processes. It is best evaluated through observed task success and its accompanying costs, choices, and consequences.
The historical transition from ergonomics and universal usability to user-sensitive and ability-based design demonstrates an expanding research capacity to change the system around its users. Original studies have shown that motor-performance-based personalization can improve performance in defined experiments, that web accessibility criteria do not exhaust the problems observed in task-based evaluations, and that assistive technology use has consequential social dimensions. These findings support an evidence architecture integrating standards, experiments, user experience, field observation, and organizational maintenance.
The central research commitment is methodological: specify the functional goal, observe the whole task, include diverse participants, identify the actual design variable, and evaluate both immediate and enduring outcomes. Within the Artificial Era, the same commitment applies to adaptive and agentic interfaces. Their value for inclusion depends on reliable access, accountable action, and the continuing ability of humans to understand and direct the systems through which their goals are pursued.
References
- Fitts, P. M. (1954). The information capacity of the human motor system in controlling the amplitude of movement. Journal of Experimental Psychology, 47(6), 381–391. https://doi.org/10.1037/h0055392
- Gajos, K. Z., Weld, D. S., & Wobbrock, J. O. (2010). Automatically generating personalized user interfaces with Supple. Artificial Intelligence, 174(12–13), 910–950. https://doi.org/10.1016/j.artint.2010.05.005
- Gajos, K. Z., Wobbrock, J. O., & Weld, D. S. (2007). Automatically generating user interfaces adapted to users' motor and vision capabilities. Proceedings of the 20th Annual ACM Symposium on User Interface Software and Technology (pp. 231–240). ACM. https://doi.org/10.1145/1294211.1294253
- Gajos, K. Z., Wobbrock, J. O., & Weld, D. S. (2008). Improving the performance of motor-impaired users with automatically-generated, ability-based interfaces. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 1257–1266). ACM. https://doi.org/10.1145/1357054.1357250
- International Ergonomics Association. (n.d.). What is ergonomics (HFE)? https://iea.cc/about/what-is-ergonomics/
- International Organization for Standardization & International Electrotechnical Commission. (2019). ISO/IEC 30071-1:2019: Information technology—Development of user interface accessibility—Part 1: Code of practice for creating accessible ICT products and services. https://www.iso.org/standard/70913.html
- International Organization for Standardization. (2018). ISO 9241-11:2018: Ergonomics of human-system interaction—Part 11: Usability: Definitions and concepts. https://www.iso.org/standard/63500.html
- International Organization for Standardization. (2019). ISO 9241-210:2019: Ergonomics of human-system interaction—Part 210: Human-centred design for interactive systems. https://www.iso.org/standard/77520.html
- Lazar, J., Allen, A., Kleinman, J., & Malarkey, C. (2007). What frustrates screen reader users on the web: A study of 100 blind users. International Journal of Human–Computer Interaction, 22(3), 247–269. https://doi.org/10.1080/10447310709336964
- Newell, A. F., & Gregor, P. (2000). "User sensitive inclusive design"—In search of a new paradigm. Proceedings of the 2000 Conference on Universal Usability (pp. 39–44). ACM. https://doi.org/10.1145/355460.355470
- Power, C., Freire, A. P., Petrie, H., & Swallow, D. (2012). Guidelines are only half of the story: Accessibility problems encountered by blind users on the web. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 433–442). ACM. https://doi.org/10.1145/2207676.2207736
- Richards, J. T., & Hanson, V. L. (2004). Web accessibility: A broader view. Proceedings of the 13th International Conference on World Wide Web (pp. 72–79). ACM. https://doi.org/10.1145/988672.988683
- Shinohara, K., & Wobbrock, J. O. (2011). In the shadow of misperception: Assistive technology use and social interactions. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 705–714). ACM. https://doi.org/10.1145/1978942.1979044
- United Nations. (2006). Convention on the Rights of Persons with Disabilities, Article 9: Accessibility. https://www.un.org/development/desa/disabilities/convention-on-the-rights-of-persons-with-disabilities/article-9-%20accessibility.html
- Wobbrock, J. O., Kane, S. K., Gajos, K. Z., Harada, S., & Froehlich, J. (2011). Ability-based design: Concept, principles and examples. ACM Transactions on Accessible Computing, 3(3), Article 9, 1–27. https://doi.org/10.1145/1952383.1952384
- World Health Organization & United Nations Children's Fund. (2022). Global report on assistive technology. World Health Organization. https://www.who.int/publications/i/item/9789240049451
- World Health Organization. (2001). International Classification of Functioning, Disability and Health (ICF). https://www.who.int/classifications/international-classification-of-functioning-disability-and-health
- World Wide Web Consortium. (2021). Making content usable for people with cognitive and learning disabilities (W3C Working Group Note). https://www.w3.org/TR/coga-usable/
- World Wide Web Consortium. (2024, December 12). Web Content Accessibility Guidelines (WCAG) 2.2 (updated W3C Recommendation). https://www.w3.org/TR/2024/REC-WCAG22-20241212/
- World Wide Web Consortium. (n.d.). Involving users in evaluating web accessibility. https://www.w3.org/WAI/test-evaluate/involving-users/