This is being provided in a rough-draft format. Communication Access Realtime Translation (CART) is provided in order to facilitate communication accessibility and may not be a totally verbatim record of the proceedings.
MEHGAN SIDHU: So good morning, everyone. Thank you so much for being here. It's always a pleasure to be in front of this audience.
My name is Mehgan Sidhu.
I am a doctoral student and lecturer at the University of Pennsylvania.
SILVIA YEE: My name is Silvia Yee. I’m the policy director at Disability Rights Education and Defense or DREDF as I tend to say because it's a lot easier.
MEHGAN SIDHU: So we're going to have a conversation this morning in a conversational style of talking about how to leverage data and research in litigation and advocacy, which in the present time I think is especially imperative for us all to be thinking about.
I'm just going to kick us off by asking the first question, which to Silvia and Stephanie, what do you see as the relationship between data and litigation or advocacy?
STEPHANIE ENYART: So I see the relationship as something that research and data is really -- it's the precursor or it is the subtext of where we end up then developing policy or law or regulation. It's the underpinning. And even just earlier this morning I was talking through one of the issues that I have related to disability, which is lighting sensitivity.
We do understand, from a research perspective, that the type of indoor lighting that is prevalent worldwide now and is often energy serving, has tremendous detrimental effects on a multitude of disability types from people with attention deficit issues, people with visibility issues like myself, people who have epilepsy and are migraine sufferers.
The list could go on.
But ultimately with lighting having the challenges that it creates, we know that it is challenging and detrimental, but also we don't have the research and we need it to understand what is that lighting type or frequency that won't trigger a bunch of detrimental effects for disability.
If we had that then worldwide, we could be in those conversations about trying to advance something into the built environment that would be far more palatable and accommodating for a big cross section of the community. So I see this as like it's that kind of prelegislation or that validation for what the regulatory standards should be, and it's tied and underneath so much of what we need to accomplish.
SILVIA YEE: Thank you, Stephanie. That's really insightful and thoughtful. I think of data as a form of information.
Information is power.
I think we see that now more than ever as we see data sources disappear or appear in modified form.
We see who can exercise that power and how those who use that data, their power can be established by their access to data. Discovery is one key litigation tool as we use as lawyers to obtain specific data, and the data that we obtain from public and other sources, such as potential plaintiffs or declare ants informs our understanding of the problem that we want to try and solve through litigation, introducing a bill or changing policy.
The individuals' stories that we get are powerful and necessary.
But quantitative and qualitative data helps establish a systemic problem and guides our intended solutions, whether we seek those solutions through administrative complaint, civil action, a bill, or policy advocacy.
I just briefly I'm going to speak about a case that we brought in California with Ernie Galvan who's here as well, and it's a case that argues that the little or no wheelchair coverage in California's essentially health plan benchmark, that's the standard for health plans sold through the marketplace, is discriminatory against people who have mobility disabilities. It's a benefit design that discourages people with mobility disabilities from purchasing marketplace insurance and singles them out for coverage of a medical need.
We had some pretty compelling stories and sued the managed care plan that was a model for California benchmark for failing to recognize discriminatory design in choosing the benchmark. While it helps to make the case for systemic action and change is data.
There are not just a few individual compelling stories about the impact of not getting a medically needed chair, but identification of who is affected. The numbers of people affected.
Not many.
But enough. The cost to the insurer of wheelchair per member per month, literally a couple of dollars, and what has happened to other plans in other states that do cover wheelchairs.
They have not become bankrupt; they do not lack for profits.
These are just data sources that help provide a fuller picture for court, for an arbitrator, for any forum that you're in.
MEHGAN SIDHU: Thanks, fully agree. Those are great points. I'll just add on to that, how vital data is to all the stages of litigation and advocacy.
There’s so much more data and research that I think we could be getting to, but also there are so many gaps that we need to address as well.
My sort of entry into this issue comes from starting off as a disability rights litigation policy advocate and then moving into an academic setting and a social science setting really learning about all these issues with data and how to look at research and data and recognizing that there are a lot of potential -- there's a lot of potential to be using data and leveraging different types of research within the advocacy and litigation spheres I think that really bolsters or would bolster the work that we seek to do.
So opening -- becoming more aware of where those data sources are and how they could be used who is really motivated me to be thinking more about those questions and recognizing the vitality and potential of indicate in research in many forms.
SILVIA YEE: So moving along to our second question, I'll ask my fellow panelists: Why is it important for people with disabilities to be demographically recognized? Isn't it enough to have disability specific civil rights laws and have people with disabilities as a group included in policies. Mehgan?
MEHGAN SIDHU: Sure I'll start us off. So as I mentioned, I think data and demographics are so critical for all stages of advocacy connected to civil rights laws and independently from them. So identifying that a problem exists, that the scope of the problem, disparities, the patterns, particularly in advocacy for impact litigation or large scale policy really relies a lot on data and data recognition.
What is the scope of the problem?
Who is affected by the problem, all of this really requires demographic data about disability. A second reason is funding. Funding for programs, funding to implement programs about problems that are recognized also depends on population data and demographics at looking at who are the people with disabilities who are affected, what is the trajectory over time, and all of that is also related integrally with data and data collection.
I think a third area is looking beyond civil rights compliance. And using data to expand narratives and arguments.
For example, building economic data and economic arguments is really central and that's something where we don't have a lot of data, but where we can really bolster a lot of advocacy particularly in a climate where civil rights compliance or pushing for compliance isn't always going to get us as far as we would like to be. Being able to make those economic arguments relies on having good quality data.
The same with building intersectional arguments. The availability of data and being able to use multiple kinds of data sources, really gives the opportunity to look at disparities based on race, gender, and disability, but disability is often left out in a lot of the data that we have, or is problematic in terms of being able to operationalize.
I'll just give some examples of demographic data from the American community survey or the using the census data really gives the opportunity to drill down into specific issues, specific populations from the local level to the national level in drawing comparisons across disparities and looking at types of disabilities that are affected.
Similarly in education spaces the IDEA data from the department of education has been able to identify disparities in school to prison pipeline issues, looking at disability in race, looking at the legal level, to the state level to the federal level.
STEPHANIE ENYART: I'll share a little about the kind of work that I do to help contextualize some of these examples, some tangible examples in this space. So I manage the public policy and research institute at AFB so I have a couple of kinds of different parts to my team and two of them include a team of Ph.D. level researchers who have lived experience of disability, and then we have a policy apparatus that basically uses that research to kind of distill and run our policy work and campaigns.
So from our vantage point within AFB, we're often trying to kind of get ahead of the questions that we think we need answered in our research, or kind of exploring things from a research perspective that's then going to enrich our advocacy. So a couple of places where we've been effective in using our research and data to kind of help drive towards something in a policy realm include in the Title II NPRM part of our comments as we filed them, public comments, included a lot of the research that we had done in the education space. We did a trilogy of studies during the pandemic time and looked at different aspects of education of K-12 students.
You could see from a systemic place that there were a lot of things that were happening related to technology access.
If you look at the rationale for that now rule related to web and app accessibility, you'll see that the Department of Justice cited as a part of their rationale that they could see that these -- this study really helped them understand the systemic impact. So it's just a moment where we could see that there was something that helped resonate, and potentially, I believe, it helped make it a far more inclusive rule.
So from that standpoint, I think that research can certainly play a role in rule making at different phases, and in did in that particular rule.
Another place, as Mehgan indicated, is really looking at appropriations issues.
I think as we are in a moment where a lot of focus will go, and a lot of pressure will go on the shoulders of states, I know that our education research has also helped people advocate for schools for the blind to get appropriate funding.
So when you're looking towards appropriations and state issues, I think that this is another really important opportunity to be able to leverage data in research.
The third and my final example, it's really related to this dialogue that we had related to website and software applications accessibility act. When we were trying to get something off the ground and we're talking to members who often are more advanced in age and maybe not the most savvy users of technology, and then we're talking to their legislative assistants that are generally millennials, they all ended up having the same question: How often do these accessibility barriers really happen? Is this all the time? Once a month?
We didn't have a real answer.
We had a lot of stories we could tell, but we couldn't really answer those questions.
So it was -- we did some research and come back around with some of the answers.
We did a diary study. And this was a very diverse sample population that was participating.
They were all quite savvy users of assistive technology.
A bunch of blind participants walked through this diary study, and they were able to help us uncover that in a week's team they experienced an average of 12.2 barriers a week. That made them spend 2.4 hours on all of these barriers in a given week.
That digital activities took twice as long as those that didn't have barriers.
Then it also helped us look towards thoughts related to other aspects of research.
So we also drilled in another study into looking at some of kind of the impacts in an economic sense, something that could speak to kind of the business community.
We began to look at percentages in a different study, but a related study on website and app accessibility.
From that participant pool, which included hundreds -- I believe it was just shy of 400 participants; we could quickly see that when people experienced barriers with websites and apps, 41% of them opted to do business elsewhere after experiencing the barriers. And just think about that. If you were in the business community, that's a huge portion of people being turned away. So this was something that definitely became the kind of thing people in the policy setting were writing down when we would go and meet with them.
Then more than 60% couldn't complete their transactions in websites and apps with accessibility barriers at all. And 65% had to wait for the assistance of a sighted individual to actually complete whatever that task was. So when we were able to speak in numbers, it does tend to be the thing that is the moment people begin scribbling on their pads or typing in their computers.
But it's something that I feel can really be that moment of a ha and clarity and people begin to get their head around, okay. I can understand this problem now.
Maybe I should reframe my thinking on it.
SILVIA YEE: Thanks, Stephanie. And those are great examples of really concrete delving in and getting information from people with disabilities and numbers that legislators, and lawmakers often want. We'll use the phrase demographic disability data; I mean the collection of information.
Like the federal government, the state governments, many entities get information all the time.
They ask questions of people. And some of that information, when started asking about race, ethnicity, gender, has provided us with key indicators of just how much difference it can make to be a person of color. To be a woman in employment, et cetera.
But we don't generally have that information when it comes to disability groups. That kind of self reported information about disability that helps us to look at general information sources and get an idea of what is the experience, what are the numbers of people with disabilities that are experiencing barriers.
I think -- so there have been health equity bills in California over the last few years, and many of those have included people with disabilities as a demographic group.
Because California has learned that, well, if you're going to have a list of protected characteristics, disability needs to be included. So disability was included among characteristics such as race, ethnicity, gender, age, sexual orientation, gender identity, on a health disparities or health equity bill that the department of managed healthcare put out in California.
But while it was there, and there was an advisory group that I sat on to help get that information, disability was not actually included because of a lack of validated specific tools for identifying and collecting information about people with disabilities in the health sphere. Federally there are six questions from federal surveys such as American communities survey.
There was one state, Oregon, that has incorporated and expanded on those questions in its state Medicaid data collection. So those questions have to be included.
Services don't depend on them. People with disabilities can choose to answer or not answer.
But the questions have to be asked in the Medicaid space.
Good data that is collected from self identification by people with disabilities requires developing and testing good tools, and that information needs to be collected and analyzed by trained front line staff and researchers. And especially important, that has to happen with intentional action and the active involvement and direction of people with disabilities and disability advocates.
I think Kate raised this yesterday in one of the plenaries yesterday, about the need for this -- that people with disabilities have to be at the center of how information about us is collected, and how it's analyzed and what it's done. It has to be compared against lived experience to be meaningful. So I think we're going to move on to the next question, which is: What do we mean when we say "data" and "research"? And how might the current federal administration's interpretation of these terms impact disability litigation and advocacy? Mehgan?
MEHGAN SIDHU: Thanks, Silvia. So the question of what is data, what is research, we sort of talk about these terms and they're very broad and vague and so just to drill down a little bit more. I think a lot of times what we're talking about with data is the quantitative data from data collection that come from these data sets or surveys or like Silvia said from the American communities survey that really just gather this information, numerical, and administrative.
That's really helpful to get a sense of the scope of a problem, who's affected, and really having the numbers that, as Stephanie said are -- really speak to the importance of an issue in policy and advocacy.
But in addition to that, there are a lot of research studies, policy analyses, and journal studies that I think are also really useful. Some of those are drawing on quantitative data and data sets to really ask a research question and develop more statistical analyses or draw conclusion, as well as taking qualitative data from different ways of asking people about their experiences and compiling them.
I think there is a lot of research in that space that can be useful in different policy and advocacy efforts. So one example of that in the area of restraint and seclusion, I was preparing a training for state administrators to look at a particular model of an alternative to restraint and seclusion that had been successful in talking about why moving away from restraint and seclusion is important.
When I looked into some of the data and research and research studies on a restraint and seclusion and particularly on this model, I came across a study from 2018 that was a longitudinal study that looked at the implementation of this alternative to restraint and seclusion in a particular residential treatment facility.
The researchers had tracked the number of incidents over a 12 -- or 13 year period from the time that the program was instituted at the RTC to the time 10 years after. And what was really interesting in this research study was not only the finding that there had been a 99% decrease in restraint frequency, and a dramatic drop enter injuries to the young people at the residential treatment center, but also that the staff had a far less injury rate.
And what it actually prompted the interest or one of the primary reasons why this entity was looking for an alternative to restraint and seclusion came because there had been so many staff injuries over a period of time from staff who had been using restraint and seclusion techniques and then were themselves getting injured in the process. So what had happened was there had been you know a loss of -- time and expenses from staff being out, huge turnover costs and really expensive workers' compensation policy costs. So the trigger -- largely the trigger for this RTC, trying to find an alternative, was really about staff and costs.
Also looking for trauma informed alternatives.
But the study, because it was really trying to show in part the impact on the staff and the finances, showed how over this period of time, since instituting these alternatives, it had saved the organization over 16 million dollars in lost time expenses, turnover costs and workers' compensation policy costs. So I think having -- finding the data and the information that's already out there, that presents these, you know, firm, concrete numbers that aren't just based on the argument that, you know, this is the right thing to do. Civil rights laws say that you have to do this.
We need to protect the kids, which is also entirely true and valid and you know.
But also that there are these other arguments that can appeal to folks who maybe are not focused on the -- you know, that's not their incentive is looking at civil rights obligations, but rather the bottom line or you know. What's going to help in staff and functioning and being able to make these arguments and look at the data that's already out there and the research that's happening and who's doing this research can I think be really helpful in shifting the narratives and being able to provide concrete information to bolster advocacy beyond just the law and policy reasons. Oh, and just one more thing, because I was talking about different types of data.
The other thing sort of beyond this quantitative, the data collection that's available from surveys and so forth, and then these research studies, I think another thing to keep a close watch on is sort of big data and administrative data.
This is something that I'm trying to learn more about, but a colleague of mine is -- looks at administrative data systems.
So at the local level, for example, in Philadelphia's child welfare service. And with the advances in digitization and availability of information, there's now a trove of data that can be accessed and collected en mass from these agencies.
He looks at how do you kind of harvest this administrative data that's out there.
What can we learn through that process about how we're measuring what kinds of questions are being asked, what is useful from that.
Also how can we then improve the systems so that these administrative agencies at the local level, the state level, are actually bringing -- are changing their collection processes and questions so that they're bringing in data that is really going to be more useful in highlighting disparities and problems.
I think disability as is really often left out particularly in those areas, so that's another space to really look beyond just kind of the larger federal sort of national data sets to the more localized data that's available.
STEPHANIE ENYART: I think one thing to kind of keep track of, in a time where we have a lot of time spent looking at news developments, is really what's happening to the data collection sources that are kind of that foundation level, that many other aspects of our research landscape essentially build on top of. So any number of different changes could happen that could essentially affect one or more of these kind of baseline spaces, because funding landscapes are rapidly shifting. So I feel like it is something that's worth everyone collectively looking at.
Because the downstream effect of not being able to build certain kinds of research on top of data collection efforts will definitely impact us in a really wide array of things.
I think at this moment where information is being fed in teaching large learning models as well, we actually do want to have a diversity of very accurate information about disability available about us. Because without it, there are going to be other kind of holes in the way that AI will develop. So for a variety of reasons, that kind of foundation level of data collection sources is something -- and their funding, is something to pay attention to.
SILVIA YEE: I just have a small thing to say here. We think of those terms, "data" and "research" and I think as litigators we're often used to dealing with experts and experts are, we pay them and we tell them what we want and they provide it for us in reports and they may be witnesses as well or depositions are taken. I think this kind of -- what we're really talking about here is more of a collaborative, maybe equal relationship with researchers.
The and really listening to what their -- to what they're interested in and how they do things, which can be very different from what a lawyer wants.
If you've ever been -- I remember having a conversation years and years ago with someone who was a good collaborator, someone who has a Ph.D. in social services, an excellent researchers and speaking together about it was a particular question and we just seemed to be talking round and round. Because I was kind of like yes, but I want to know this.
She's oh, but I think the research -- and then she started formulating like how she might -- the kind of research she might do and the kinds of parameters she needed to figure out before she could do effective research.
But I don't -- the point I want is this.
I want to be able to bring this action and what I need is this.
I need X.
She's like, but you can't just get X.
It was a very instructive conversation for me.
Because I really respected her.
I knew she had points I needed to figure out.
So when you collaborate with researchers and you figure out what they need and why they need it, why they're after a certain asking questions a certain way and the data that will come from it. It may help you figure out a different way, maybe, of what you want to litigate.
Maybe you want to go further back in time the stream of when things happened.
But it is just a different kind of collaboration.
One that I think is instructive for both sides, but that both sides need to be humble about.
MEHGAN SIDHU: What are possible data sources that you have used and how can they be used by other attorneys and advocates moving forward? Silvia?
SILVIA YEE: Okay. I do have, there are just two slides. That I've brought with me.
For those of you who are visual learners, here's something to look at.
There are slides of research that we've had, we were fortunate to have Dr. Steven Kay, literally a former rocket scientist but he does a lot of analysis on disability research and he did some research for us using data from the National Health Interview Survey, the American Community Survey, and the Medical Expenditure Panel survey. So that provided the raw data.
We're fortunate to still have this. We, as I mentioned earlier, we're seeing formally public data sources taken down from government websites and sometimes they're put up again without note of the changes.
We're seeing a severe culling of federal agency staff that would typically be updating federal data sources.
This is happening in what you might call pure data sources, at say the DCD and at HHS.
There are further repercussions for privately funded data sources. Robert Wood Johnson has for the past decade had county health rankings and road maps data set which provides county by county data.
Robert Wood Johnson has pulled funding from that because it needs to prioritize different things. There are big cuts at civil rights agencies at Ed and at HHS and the pulling of grants from entities that enforce the fair housing rule means that it will be hard to get data on whether and how disability federal civil rights laws are being implemented and enforced across the country.
But data still provides us with -- still meets important needs. We had -- Dr. Kay did this information to help in the fight for Medicaid. And tried to put together some -- he has a 37 slide presentation that is on the DREDF website in an accessible PDF format. You're welcome to look through that. I only pulled two of those slides.
The first slide looks at how Medicaid covers working aged disability adults and the second slide we'll be looking at the policy of work requirements and why it's not a good policy going back to the first slide.
On the slide I have two pie charts in different colors. The pie charts, the first one looks at the primary payer among disabled adults aged 19 64 in 2023.
The second pie chart zeros in on who gets Medicare.
The Medicare type and supplemental coverage from Medicare. You'll see the pie chart shows the primary healthcare source, it shows that private employment based group insurance that is obtained from a currents or former employer or union is the most common insurance amongst working aged disabled adults.
More than one third have such coverage.
But Medicaid is next, and provides coverage for one quarter of disabled working age adults.
17% have Medicare coverage, and individual private insurance policies cover only about 7%. Five of those are purchased from the exchange, 5% of those.
Then 2% are obtained directly from an insurer. So only about 10% of the disabled working age adults are uninsured, which is down from about 20% before the Affordable Care Act took effect. So zeroing in a little bit on Medicare, so 18% of working age disabled adults have Medicare. 3% of this group has original Medicare, 4% has Medicare Advantage plans. About 1% has coverage from the VA or military. 3% have private coverage.
7% also have Medicaid. So they're dually eligible individuals. So this is just important for showing that how critical Medicaid is to many people with disabilities, that it's -- these are people who won't easily have another source of insurance. If Medicaid is lost, this is not a group that afford to pay for private insurance.
This is a group that relies for their healthcare on Medicaid.
Even amongst people who have Medicare, there is a significant group that also relies on Medicaid to help them pay for Medicare premiums.
Looking at the next group -- but also I wanted to note just before we leave that first slide, that this very important one third of employment -- one third of people with disabilities are getting employment based coverage. I don't know if that's a commonly known figure.
I think there still often is the perception, well, if you have a significant disability, you're obviously on Medicaid and that isn't necessarily the fact at all. So data, it's important.
The next slide is looking at work requirements, and this is an important slide for us. We really want to get this out there.
Because there is a very significant risk with the coming budget cuts that work requirements will be imposed upon Medicaid, whether by individual states or as a federal requirement. So this also has a pie chart, as well as a written side chart that looks -- gives more detail about the labor force. So this was an analysis looking at labor force and disability classification of Medicaid beneficiaries ages 19 64.
Basically working age.
Looking through the pie chart, it's divided into mutually exclusive categories, according to labor force status, school attendance and broadly defined disability status.
I have not changed it. So I will thank you for waving your hand at me. So 55.5% are in the labor force. Already in the labor force. More than half of Medicaid beneficiaries are working or actively looking for work. 3.7% are not in the labor force, but they are attending school. About 29% are not in neither of those categories. They're not in the labor force. They're not looking for work.
They're not attending school, but they can be considered disabled.
If you look at that chart on the right, the written one, for these reasons.
Of this group, about 37% receive supplemental security income. They became eligible for Medicaid benefits in the first place because they got SSI and they came in through the disabled pathway.
Very low income, and have a disability. Of those who are not on SSI, 36% receive social security disability insurance benefits, and therefore also meet the Social Security Administration definition of work disability. They have a disability that means that they can't work. 84.4% -- these are not exclusive categories, mutually exclusive categories. 84.4% self report a limitation in their ability to work due to a physical, mental, or emotional problem. That's how the question is phrased. 79% are classified as disabled, according to the standard six question disability measures used in the American communities survey. Those are questions that say you have a significant disability mobility or significant disability in seeing and so forth.
So Medicaid beneficiaries in this disabled category have high healthcare needs. 25% report poor health, 61% report fair or poor health. 40% are frequent healthcare users. They see their health providers at least 10 times a year.
24% have been hospitalized in the prior year.
The median healthcare expenditures from all of these payers, from all sources, public sources, private insurance, out of pocket, are close to 7500 a year. That's a lot to spend on healthcare. So this group is on Medicaid as well.
That leaves only 11.5%. And almost 73% of this group, this remaining group have kids under 18 living at home. So they're taking care of their kids. That leaves only about 3%. This is a lot of bother, a lot of changes in policy, a lot of paperwork to be throwing at all Medicaid beneficiaries to be trying to get at that 3%. So all the stories you hear about those who are abusing Medicaid, those who are not truly disabled. Sure you might get the occasional anecdote that seems striking, but what the data shows is that Medicaid is lean. It's reaching a group of people that needs it.
It's irreplaceable.
It's not being abused. So this is something that data helps provide, and I wanted to spend some time going into this, because yes, we got this data for a particular purpose. We got it for a particular advocacy purpose, to be used in Medicaid defense.
We also wanted -- I think it also illustrates some of the meta issues around data sources that we use.
How do we identify disability? What are the advantages and shortcomings of the tools we have for disability collection; what do we need those tools to do?. So we have the ACS survey, we have the surveys that Steve Kay has used, and again the survey -- the entire presentation is available on the DREDF site and in accessible PDF. This last, as recently as December, November December, hard to believe, we had a meeting with the census bureau that looked at instruments, these tools, how they collected information of people with disabilities, the kinds of questions they asked. There was a lot of discussion about how do we -- there were people who identify as having a disability but who don't identify with the six ACS questions.
How do we capture them?
What's the extent of the community? When we're talking about allocating money for programs, how do we get good numbers for who needs those services and supports?
Those are all really important questions that people with disabilities need to be at the heart of.
It seems like an entire age ago, because we're not asking those questions right now. We just seem to be cutting many of those same services and supports.
But it does really raise this question of, and makes us think the tools we use to collect information about disability, the information we use about disability is going to be used for many sources and for many reasons.
The information we collect for social security disability income, that serves a particular purpose and a lot of it is gatekeeping. That's not the tool we will want to use to identify the parameters of disability community for other programs, or other services, that maybe are not connected at all with income.
I wish we could get back to that point, which we almost seem to be on the cusp on at the end of last year.
We can't.
We have a new reality right now.
But hopefully we can still be thinking about these questions, getting ready for the next time when we'll be on the cusp of making changes, so that we are again in the heart of the discussion, directing the discussion about where we go next. So we can be subjects determining our fate, rather than objects for others to research about.
I would like to turn it over to Stephanie to give us lots more about other sources of information.
STEPHANIE ENYART: Sure, sure. So in a moment where I think maybe the federal footprint is changing rapidly, and much more attention is going to be going to the states, I would imagine a lot of people in this audience could be thinking about their strategies related to what those services and programs heavily relied on in that particular state may look like. Whether we're talking about appropriations, whether we're talking about the way the services should run in the communities they're serving, or whether you're going to be suing the state at some point for not doing what they really need to do on any number of fronts.
From that standpoint I think that one data source that I would think about and turn to.
This is updated annually, usually in the spring, usually in the month of March and has just come out earlier this week with its latest round of data.
Next March already another round or so.
Disabilities statistics compendium.
What this offers is a state by state look across a number of different domains related to the disability population. It's obviously going to look at the prevalence of disability.
But it also looks at information related to employment data, SSI and SSDI.
Then maybe this says more about me.
Me of these categories become kind of interesting to just look at for fun. I mean there's earnings disparities and poverty disparities which could be very useful.
But there's also housing types, like how many individuals are living in places with full kitchens v. overcrowded spaces, health indicators, which also include content related to obesity and binge drinking and smoking. The rural/urban divide, educational issues, voc rehab and something that may be very, very important for our community, which is voting and registration information.
So I would imagine any kind of state based advocacy can be really enriched by looking at the statistics in your particular state or in comparing some states. That's where they get amused in looking at the differences and why there are the differences there.
I feel like that could also really bring about a lot of interesting investigations.
There is a section on blindness and low vision that the American Foundation for the Blind contributes to the creation of, and so we've recently helped issue that portion of the disability statistics compendium and in our own work at AFB we create what's called the statistical snapshots that looks at a handful of different sources to get a lot of different kinds of demographics information about the blind community, the low vision community in the United States. It also comes out in the spring, so ours comes out around the same time as the compendium does. Other things that we do at AFB that are in the research and data space include our own original research.
We have a series of fact sheets and summaries that are really intended for consumption by a broader audience.
That distills information on a bunch of different topics whether we're talking about transportation or other topical issues. So you can find a summary of research on themes.
We also have individual studies themselves. In the COVID era we had a very, very large survey that 1,921 participants in it, which, for a low incidence population, just so you know in case that's not your wheelhouse, that's a really large sample size.
So there's still some data in there that has been a useful benchmark, as we build towards other kinds of collection efforts, because we have that great view of a huge sample of the community. We have an education series that has three parts to it called Access and Engagement, and then we have two different studies related to website and app accessibility, and I spoke to that a bit earlier today. We also have one that's on the workplace technology used, as well as a lot of the social and emotional aspects of the barriers that people experience in the workplace with technology barriers.
I honestly feel like if litigators were interested, giving that a read could spawn a lot of different interesting investigations, I feel, about things related to the employment landscape and the ADA.
We have also just embarked on artificial intelligence related research. I'll probably speak a little more about that later but we have released one study and a second one is in the works.
We're going to have a webinar next week related to the use of research in more blindness related advocacy, but there is some related to the disability community as well. So if this is something that you would like to join us to nerd out about, that's available for free next week. I think with that I can also just mention two other data sources.
The vision alliance, it was a snapshot in time, it's that one moment, it's not as I've described some of these are recurring annually.
With disabilities statistics compendium and our statistical work at AFB.
But what it did do is it offered county by county information related to blindness. It was two populations, it was working adults, working age, as well as older adults.
The older adults had a large percentage of people with multiple disability types.
If you were in a place where you were doing some county based or state based litigation, I feel like it's worth looking to see if this particular data source may have interesting data about prevalence and a variety of issues. I also want to put a plug in, although I don't have a lot of extensive notes, the National Disability Institute has definitely done some interesting research on a couple of different topics and they often take more of an economic or financial lens on things.
So that's another place to look for research.
MEHGAN SIDHU: I'll just add on to what Stephanie said about the annual statistics compendium, which I think is such a great resource, and also as Stephanie mentioned a great place to nerd out into looking at some of this data. I think that the website -- so if you go to researchondisability.org, it will take you to the center for research on disability, which is run through the University of New Hampshire. And one of the things I really like on this website, in addition to the compendium, it has this whole research database where you can find a lot of the different studies, different types of research, different data that pertains to all sorts of different disability related issues and questions.
There is a build your own statistics tab on there that allows you to really kind of get into the data and sort of select a particular state or a particular location, a type of disability, a population really kind of look deeply into that. And finally I think there is a training at the top menu of this website, there is a link to training, where they have these modules for disability statistics training that I recommend for anyone who's just interested in sort of delving into data and disability statistics a little bit more.
Particularly the first module, which really just kind of generally talks about data sources, how to access them, and then the -- I think the fifth module really dives into how to use the data.
That's a good source, if you want to learn more. AI is already impacted modern fast et cetera of life and it relies on heavily training data. What are the potential risks and benefits for people with disabilities?
STEPHANIE ENYART: All right. This is a whopper of a question if you ask me. I guess I'll just say that at AFB, one of the things that we wanted to do was we wanted to be really equilibrium rate, because you could really research AI and the disability community in a thousand different directions.
We wanted a real sense from experts where should we go and dive deep.
So the first study that we did, which just came out earlier this year, is a consensus study. It's study with 32 participants. The way a Dell if I study works is you ask a really broad question to experts and you have them give feedback about their opinions and you anonymize the way that their feedback came in.
You can't tell oh, well, we all have to agree with that because we know who that is.
Then it goes through more and more realms until essentially, after multiple kind of iterations of looking at these statements and asking people if they agree or if they disagree or why they disagree, you get to a place where experts actually have consensus around several different things they believe will happen in the future.
We took 32 individuals that were heavily in innovation spaces in AI related to disability, as well as in academia looking at AI and disability, and people that are working on policy and advocacy and disability.
We asked this panel of experts what do we need to dig into in terms of research, and what are the themes, the issues that are going to be kind of the juiciest to unpack with future research.
So that particular study is available now and we have some principles that also kind of ladder down from that. I can share just a little bit related to it, but it's better to look at it yourself because there's a lot of different things to think about. So obviously in the benefits category with assistive AI on the horizon, there's a lot of different benefits that will really potentially enrich the lives of many people with disabilities if we can get access to them.
So whether we're talking about really readily accessible and affordable fully autonomous vehicles being we'll available or way finding, object recognition. And what that could mean for people, also many different kinds of tools that could be used in education and employment related to grammar and writing support, as well as monitoring any number of health issues. All of those spaces, and many more, have wild benefits available to our community.
There are many areas that we had expert consensus around as areas of concern or areas where they felt that there would be particular areas for litigation in the future, potentially.
Some of the areas -- and there's many more in this study. AI should not replace human educator.
They felt that AI would potentially be going in that direction.
Me of the other things attached to that, there were concerns about chat bots as a tutor, and that this won't be fully kind of accessible either.
Because a chatbot will potentially never understand the diversity of experiences and nuances of disability.
I don't know if any of you have had experiences with chatbots but they truly don't understand what you, as an individual, can and can't do, since there is such variation in our experiences and disability. The methods that people will upskill and learn AI to be able to maintain and kind of gain foothold in society may be massively inaccessible.
This would be something that I hope the litigator crowd is taking note of. Because this will be the employment revolution coming.
So, please.
Please.
If you dig into nothing else, think about that. Because our employment statistics will plummet if you don't.
Then there is, of course, the healthcare bias. We've probably all digested some related to algorithms denying people, especially those with complex health needs, automatically denying them for different coverage within the healthcare system.
They felt this would also be a prolific problem.
Then in the employment setting, two areas of consensus were around candidate screening. So the algorithms that are literally selecting candidates to move forward v. to not move forward, there is a lot of concern about AI and its role there, as well as employment automated tests.
I hadn't really seeing this illustrated except for the fact that my husband recently walked through the employment process where he was looking for a job.
Although he's in finance, which I still totally don't understand, he had to go through these tests to -- with AI, determining what kind of manager he was.
He would sit in front of a screen where it was tracking his eye movements and making sure that he was remaining there and in camera, not looking at other things.
It was giving him a standardized test that was fully automated.
Then it literally spit out immediately a report telling him the type of manager he was and sent that back to the recruiter who had ordered it.
I'm sure you can imagine what that will look like in terms of the accommodations and equity landscape for our community. So that was something that they flagged and is already happening.
In terms of education, there will probably be a lot of need for advocates to really dig in and clarify a thorny landscape of using assistive AI in the classroom setting. Because it can be such a boon and truly, this will be the way that many people that are in education today will be employed later is really leveraging the tools that are available.
To what extent we can use assistive AI in the education setting will probably be extensively litigated. Other issues included the use of AI to develop IEPs, which of course I don't know how individualized they could be if we have AI kind of coming up with what it recommends. Perhaps it's a jumping off place.
But in so many of these things you could see the need for humans to be very much in the loop and very involved.
Some of these areas in the landscape may be areas to think about and to plan for, and perhaps fodder future tenBroek symposiums, I'm not sure.
But I also wanted to just plug we are now embarking on a second study, and that is centered on the lived experiences of people with disabilities, and their use of AI, their experience with AI.
We're going to have a nondisabled control group.
We will really be recruiting from all kinds of people to participate in this study, and we're also working with an advisory panel of many different organizations, some of them are here today, to help shape the lines of inquiry.
But we are at that key stage every figuring out what kinds of things we want to zero in on. So if you have brilliant ideas, please approach me over the next couple of weeks. Because this would be an opportunity to build in that line of inquiry into something that may spawn your next juicy investigation.
It could be fodder for any number of things on the other side.
So there's a place for you to lean in and get the research that maybe you want.
SILVIA YEE: I'm going to speak very quickly, hopefully not too quickly, about a project that DREDF has embarked on looking at state legislation of AI. And this is something that's happening right now. The federal government has really backed off on this because they want technological innovation and we don't currently have a regulation heavy federal government.
States are really going out, realizing oh, let's do something.
Colorado is one of the first out of the gate and passed a general bill governing cons consequential decisions in 2024. And many states are copying the contours of that bill.
It covers AI developers and deployers, those who use AI as covered entities.
The safeguards are key to pro active against algorithmic bias and that depends on knowing what bias against a particular group entails. So these bills rarely -- there's not a developed interaction with existing civil rights laws.
They tend to have one entity; one state entity enforce them.
Even within a state there cannot be -- there's not really an understanding necessarily of how the general bill will interact with specific AI bills that let's say cover prior authorization or insurance or other things. So there's a lot of sort of gaps to be aware of within the state and as other states copy.
I personally there are groups that are working on this AAPD is working on this, center for democracy and technology is working on this very actively. I tend to be kind of anxious.
I think there is a fundamental problem with how AI works when it comes to the inclusion of people with disabilities. AI is trained on huge swaths on data, and moves toward a norm.
It's almost -- I've been thinking of it as a modern form of industrialization. It doesn't work to standardizing assembly line of work to standard bodies and capacities, but instead it absorbs standardized ways of being.
How do most people react to a treatment?
What do most people do when they perform a task?
What are the credit characteristics of most people who pay their debts?
But people with disabilities don't always have the characteristics or thinking patterns or healthcare needs of most people.
Our disability rights laws recognize that with policy accommodations or policy modifications, with reasonable accommodations or policy modifications, people with disabilities can and do meet education or school or wellness standards.
But how we can be sure that the data used to train AI, and that it increasingly uses to train itself, is data that sufficiently includes results derived after accommodations and policy mods have been played.
This is my overarching fear about AI. When we were discussing this, the three of us, we were very excited because we're kind of data geeks.
Then we'll have so much time at the end for questions, because we want to engage with the people. And here we are.
A few minutes fast, a couple minutes past and we didn't even get applause.
Oh, sorry.
(Applause) -- We're happy to stay here and engage with any of you. So sorry that, again, I ran out of time. We'll have cards, happy to take your cards, and really hope to continue this conversation.
MEHGAN SIDHU: I think we have 13 minutes left.
SILVIA YEE: Oh, we do. I'm so confused. You see my computers is still on California time, and an apparently I can't translate three hours' difference.
That's fantastic!
Yay!
I'm really happy.
Let's have some questions.
Question & Answer Session
ATTENDEE: Thanks so much for the panel. My question is more towards the evaluation of -- hi. I'm Sam, AARP Foundation. So with generative AI I share similar concerns from the models learning from themselves and what that means when, often in the models people with disabilities end up being statistical outliers and thrown out in the equation.
I guess I'm curious on y'all's thoughts about if there is systemic fixes working within the idea of what AI is that can address those issues or if it is really going to turn into we need enough transparency and accountant to make sure it's not going to discriminate against people with disabilities.
STEPHANIE ENYART: I think you answered your own question. My understanding is we need humans at least at this point. As of the moment we're in, that's my answer for today.
Because I think that's where we are with the age of AI.
It's still really a toddler in terms of generative AI, so think of it that way.
SILVIA YEE: This is Silvia, and I -- yes. Stephanie makes really good points and I agree with that and I think you have been thinking this through. I think that to -- I think about generative AI and I also think about decision making AI.
If there are ways to build in rules that you can't get around, rules about you have to include and think about accommodations and how that would work, that requires a training, a level of training that I don't think we're at.
When I'm looking at the state laws, they rely upon entities that will come in and assess an AI tool, many of them.
And a lot of these entities actually are aware of, let's say, racial bias.
Because we've had some pretty good research into that and how it works.
Me of it in the healthcare field. Some groundbreaking research.
We don't have that equivalent when it comes to disability.
While these testers, they're aware of the problems they know and they don't know what they don't know.
I'm not sure who's going to be training the testers and how they're going to ensure that disability is considered in the right way. I don't think we're there, and I'm not sure who's going to -- as the legislation is building now, I'm not sure who's going to be able to force that or require that.
I think it's going to impact our lives more and more.
Because if you're an outlier now and you're not included in AI, as time passes you'll just be more and more of an outlier.
It's much like building the building. Who's going to fix the building 10 years after it's built? So that's a really big concern.
STEPHANIE ENYART: I agree with the elements of what you just spelled out. Because in the AI landscape in healthcare, there are some on the industry side, there are people that are trying to really get at that race and gender related bias.
So they have, you know, good companies out there trying to, you know, address the disparity gaps there.
I was recently in a place where I was on a panel and I was the advocacy person.
There were some feds on the panel. There was one person from industry.
He was the good apple, the one that was doing the anti bias work in race and gender and against all my legal training I asked a question I didn't know the answer to on a stage, mic’ed.
It worked because I said to him, look. You're doing all this great stuff on race and gender, but we just did a whole panel, Buddy, on the fact that disability is going to throw a loop through all of this, because the data isn't embedded in places that it needs to be. So what are you doing on disability?.
He had nothing to say. Because even the good actors aren't out there being able to crack the nut on healthcare and AI and the biases.
It's because it's a data issue.
Back to the beginning of what we were here to talk about.
ATTENDEE: Larry Berger. First, a question that I think goes to both where you can find it and also somehow to AI is electronic health records certainly in healthcare institutions are now pretty universal and are also in many institutions for people with disabilities, including group home administrators and the like. They're not all compatible with one another, but I think that's probably solvable at some point, so I think that's just a possible big area for the future.
Then I have another completely different comment, which is that so often in litigation we are trying to get a preliminary injunction.
Very often sole practitioners and small law firms and a great tension between needing information and needing immediate relief in the testing area, high stakes testing area, which I've done a lot of in the last few years, the National Board of Medical Examiners, for example, not just them, has lots of data.
Very often in these testing cases you go in without any discovery at all.
Because the time restraints are such.
Even if you have some discovery, it's very limited.
So that -- in big dollar litigation, which I used to do, securities cases and antitrust cases and things like that. There are law firms and litigation support companies and so forth that can get at the data they need more quickly, but in this area, because it's such a big area and so different in its nature.
I don't think that's always available.
MEHGAN SIDHU: Right.
SILVIA YEE: I wonder if there are pro bono researchers in.
STEPHANIE ENYART: They're grad students.
(Laughter)
SILVIA YEE: Yeah. Yeah. I mean sometimes you can make some of those connections.
Recently I was introducing some people from the National Immigration Law Center who are really interested in the vulnerability of the TM system.
I know it as a shortcut, but it stands for something I cannot remember. Transformed Medicaid statistical information system. It's a huge Medicaid federal database with a lot of information on Medicaid beneficiaries.
The lawyers from milk were worried about this being a potential weak data spot, an unprotected data spot or less protected data spot when it comes to immigration information about Medicaid, who's using Medicaid. So I put them in touch with some researchers who work with TM from a major university and that has a great connection to make.
Hopefully they'll learn from one another.
Yeah, I've been at my desk with something due tomorrow thinking that I do not have time to go through a tutorial on how the database works.
I want my information now. And you know, I wish I had a better solution for that. I think it's only in building longer term relationships, like learning, finding the grad student.
We've tried to have policy and graduate researchers working in the summer interns at DREDF for a few years now.
It does give you an in with different universities to sort of figure that out.
MEHGAN SIDHU: Yeah. I agree, Silvia.
I think those collaborations are really important to try to build with this community and with folks who are doing research, folks who want to be involved in doing more research in disability.
One of the big fears I have is, you know, we're looking at now a federal government that's losing a lot of funding, that goes into NIH grants and other funding for data.
The federal government and agencies that are being dismantled are the ones that are doing a lot of the data collection analysis and storing the data. And at the same time disabilities facing this particular uncertainty, as, you know, part of DEI or not part of -- you know, this sort of vague idea of DEI and these -- there's a grant application advisor or something that had sent out some blog or newsletter earlier this week that was noting that many applicants for NIH grants are trying to avoid entirely words that were listed in this New York Times list of DEI flagged words that should be left out of these grant applications or that people are trying to leave them out of grant applications and those words are things like disability, accessibility, discrimination. Accessible, disparity, equality, equal opportunity, mental health.
All of this is really, really concerning, because I think that even talking about the data structures that we have and trying to tap into them more, we're also looking at what's happening that is just crumbling and falling apart. So while I think that it's really vital to kind of keep together what we have, what's intact, I think it's also an opportunity for us to really be collaborating because we can't rely potentially on these federal systems that have been maintaining this information in data.
You know, and potentially people losing their jobs who are doing data collection or have a lot of experience in these areas. And my thought is how can we sort of build those -- this is the time to build these collaborations, if there ever was one.
This is the time to bring together what are the -- what is the research that we've saved, what is the data that we have, what do we have on sort of the litigation on the policy end, what do we have on the data end?
How can we bring that together and the communities together so we can be, you know, rebuilding and rethinking together about some of these questions at such a perilous time. Hopefully things get better, but maybe they don't.
We need to also be thinking about where are we all headed collectively in this heavy, risky, and also, you know, the possibilities that are within that.
You know that piece of collaboration, I think is really important to be able to draw those connections, because I think there are people who want to do the work and there are a lot of conversations that we could be having together.
STEPHANIE ENYART: Absolutely. I also think it makes sense to put a plug in for a nonresident fellowship program that we are expanding right now. It's stipend so it has some money attached to it.
It offers usually building towards publication opportunity as well, which is something that most researchers are really looking for, especially early career which is this is for.
Just planting the seed in case you know people in this other domain that would like to do this type of work.
(Applause)