My career has focused on understanding the needs of people with disabilities, so I could help develop accessibility accommodations. Focusing on this scope often touched on the question, “What is a disability?”. The word “disability” itself implies that a person has reduced ability in some fashion.
The social model changes the focus, to say that disability comes from the intersection of person and their environment, thus shifting the definition and responsibility to the larger society. That leaves the question, “What intersections of human characteristics and environmental affordances can be disabling?”, which we need to know before we can figure out how to alter the environment positively.
Many of these intersections have been identified, and have formed the basis of accessibility work up to now. However, user groups that are smaller or are less able to self-advocate have often been overlooked. To provide a complete picture of the accessibility space, we need ways to expand our knowledge beyond what has already been accumulated.
Human characteristics and the environment
A systematic top-down approach might help to identify accessibility needs for groups that do not have well documented needs. To start building a model, I suggest that human characteristics come in a continuum, with some people having a given characteristic more or less intensely than others. For instance, with height, some people are short, some are average height, some are tall. Or with eyesight, some have lower vision, some have good vision, and some have great vision.
Those are super-broad examples just to illustrate the idea. The characteristics we examine can be more narrow, to the level of specific human abilities or characteristics. For each of those abilities or characteristics, there will be this variation. People with a less intense level than the average may experience a disability in some circumstances.
Experiencing disability still depends on the environment, which can change. For example, brain research suggests that the neurological difference behind dyslexia has existed in a certain proportion of the population for millenia. It was not disabling, however, until the widespread use of written text happened only a few hundred years ago. Because of this change in the environment, this relatively common variation in the human brain became associated with a disability.
New technologies that create new forms of human interaction with the environment also risk creating new disabilities, or making known ones more pronounced. Personal computing and then the Web really made this apparent, leading to the evolution of the modern accessibility industry. The virtual and augmented reality space promises to have an even broader impact.
Often, the accessibility impact is recognized only after the technology has become well established, and then it can be more difficult to build solutions. To get ahead of that, we need some way of predicting future accessibility needs while technologies are still being developed. We need a way to answer the question, “What areas of human variability might intersect with future technologies in disabling ways?”.
The bell curve of variability
If we could take a given human characteristic, rate its level, and plot that on a graph against population size having that strength, we would probably observe a distribution like the normal curve from statistics. In principle, any given human characteristic would probably show such a curve.
Interactive accessible version
Figure 1 shows the “normal curve” with lines showing the position of integral standard deviations (1, 2, 3) to the left and right of the middle, with a rainbow spectrum under the curve representing the full set of values.
The normal curve is the familiar “bell curve” that shows that measurements of a property cluster around an average value at the middle of the curve, and tail off on both sides away from that value. Statisticians express the distance from the middle with the concept of “standard deviations”, which are statistical distances from the middle of the curve.
In figure 1, there are vertical lines at the position of each integral standard deviation, from minus 3 to plus 3. There is nothing special about those particular standard deviation values, they simply form convenient boundaries to group the data.
The centre two groups, from minus 1 to 0 and from 0 to plus 1, encompasses a majority whose measurements are all statistically fairly close. From the definition of a normal distribution, 68 percent of the population falls into this central region. Thus this central part of the graph represents a moderate range of values within this majority.
Both to the left and right of that region, there are bands for plus or minus 1 to 2, and for plus or minus 2 to 3. In these regions, there are both fewer measurements, and the values are farther from the average. The plus or minus 1 to 2 regions encompass 27 percent of the population, and the plus or minus 2 to 3 regions only 4.5 percent.
Within these regions, though, are a large set of diverse values that are important to recognize. Consider the range of measurements for a human characteristic, say adult height. The mean height might be around 170 cm (5' 7"), with a standard deviation of 8 cm (3"). Therefore, heights from 162 to 178 cm (5' 4" to 5' 10") fall within the central area, representing a majority with a relatively close to average height. In the region of minus 1 to minus 2 standard deviations, heights are 154 cm to 162 cm (5' 1" to 5' 4"). These values are more noticeably different from the average, while the minus 2 to minus 3 range goes even further with heights between 146 and 154 cm (4' 9" and 5' 1").
Applying conceptual labels to these regions helps to clarify this. From the height examples above, it makes sense to label the middle region as “average”, the region of lower height as “low” and heights below that as “minimal”. The same can be done on the other side of the curve, with above average as “high” and beyond that, “extreme”. These labels may sound awkward when paired with specific terms like height, but they are intended for general consideration of a wide range of human characteristics and need generic forms.
Having identified these bands, the population density curve is no longer of interest. It is these value groups that we want to use. Figure 4 collapses the curve into a band with these 5 regions labeled. To emphasize that this is an abstraction, the full spectrum is shown in a band below, and the 5 regions are coloured to an average of the spectrum beneath them: orange for “minimal”, yellow for “low”, green for “average”, blue for “high”, purple for “extreme”.
Spectra of human characteristics
In principle, just about any human characteristic can be rated on some form of strength scale that shows the range of that characteristic in the population, and that range will probably show a distribution like this. The 5 groups for the characteristic simplifies examination of these characteristics.
The centre band, labeled “average”, is the region of moderate variability and most numbers. I would say this corresponds to the field of “usability”, which has generally addressed that average majority of people.
The “low” and “minimal” regions correspond to the field of accessibility. We even use comparable terms, like “low vision” and “no vision”, or “limited arm mobility” and “no arm mobility”. Therefore, these are areas that accessibility specialists want to examine, to consider if there are disabling interactions with the environment, and if so what accommodations might be appropriate.
The “high” and “extreme” regions have not typically been recognized as related to accessibility in a general way. One might expect that having a higher than average strength is an advantage in the environment, and sometimes that is true. But the environment is often equally poorly designed for people in this range, and I believe disabling conditions can emerge. For instance, being extra tall is a considerable disadvantage in an airplane seat designed for an average person. Another example I came across is “hyperacusis”, which is an above average sensitivity to sound, but it can mean that a person finds many environments to be painfully loud.
Identifying potential accessibility needs
How do we apply this model to the accessibility field? In principle, every separate human characteristic or ability could be measured in a way that we can identify these 5 regions of values. Doing this for every human characteristic would yield a (potentially huge) set of spectra. Every ability or disability that we know about could be described in part by its appropriate placement in the continuum of one of these characteristics.
In the real world, we don’t know nearly enough to populate such a model completely. But we do know enough to get started, identifying many abilities and disabilities and how they relate to each other. Because our knowledge is limited, there would be many spaces where nothing shows up, and this is where we start to explore. We can “fill in the blanks” by looking at what we know about the characteristic, and what we might extrapolate about the other bands. This makes it possible to hypothesize populations that exist, and the needs they may have.
Ideally, all of this would be based on scientific research. Where research is known, it should inform the model. When research is not able to do that, the model becomes a way to suggest research. As research comes available, it could reinforce a part of the model, or it could lead to changes in the model. However, aspects of the model not yet backed by research are still important to include.
Discussion
Maybe this seems like an abstract way of identifying accessibility needs. As the environment evolves and new technology is created, new accessibility intersections emerge, and the field needs to learn about new types of accessibility needs and accommodations. A model like this can help to predict these intersections in time to address them.
The model is intentionally a simplification, and as such has some key limitations and considerations:
- This is purely a model to aid exploration of accessibility, and is not meant to be a scientific description of human characteristics.
- The normal curve is an abstraction, and real-world measurements won’t match it exactly, but they often are close enough for the model to be valid.
- Dividing the regions by bands defined by integral standard deviations is purely a rule of thumb. When working with real data, the appropriate dividing lines for a given characteristic may be at different parts of the standard deviation line.
- This model simplifies measurements into 5 groups, which inherently means diversity within the groups is overlooked. It is important to understand that diversity while maintaining that simplification.
- How a characteristic is explored can affect the model. For example looking at vision as a single characteristic will likely yield different results than looking at color perception, depth perception, visual acuity, and field of vision as separate characteristics.
- Measurements beyond 3 standard deviations are not included in the model. In some cases, people in this range need highly personalized accommodations, and an abstract model may not be able to provide useful predictions about values in this range.
Recognizing all this, I think using a model like this can help expand the understanding of the accessibility space and help predict emerging accessibility needs. I offer this idea for consideration, and hope to engage with people on further exploration of this idea.
Acknowledgements
Thank you to Rachael Bradley Montgomery and Bern Jordan for their review and assistance with this post.
Comments