Disability in Costa Rica Operationalized as a Social Problem Points to Social Solutions
Erika Sanborne, University of Minnesota
I shared a preliminary iteration of this study at:
Population Association of America (PAA 2024) Annual Meeting,
Columbus, OH US
Session: Health, Health Behaviors, and Healthcare
If anything here is not accessible to you, please let me know how I can help you.
Abstract
This study employs two distinct analytical models—the social and medical models of disability—to explore well-being among Costa Rican women. By using ‘difficulty seeing’ as an indicator of disability in the medical model and ‘lack of eyeglasses’ in the social model, the research contrasts a functional impairment with an accessibility issue. Grounded in the social stress process model and the capability approach, this study aims to reveal how operationalizing disability in these two different ways can lead to nuanced understandings of well-being disparities. The findings are intended to enrich sociological perspectives on disability as a social issue, and to contribute to demographic scholarship on the interplay between disability and well-being.
Research Questions
- How do the results of analyses using a social model of disability differ from analyses using a medical model, when comparing national survey data on well-being?
- Are different well-being gaps apparent through the social model approach?
What is the Social Model of Disability?
The social model of disability contrasts with the medical model of disability.
social model of disability, noun.
definition of
The social model of disability views disability as a problem within society. A person’s medical problems are not what disable them, but when the person with impairments encounters the attitudinal and physical barriers in the social world and experiences exclusion, that is disability.
Disabled people’s well-being would be improved, according to the social model of disability, if their social milieu were made more accessible by the removal of physical and attitudinal barriers, and the addition of more safe pathways toward full and equal participation in the structures of society. According to the social model of disability, disability is a violation of human rights, as disabled people are deprived of equal access to the resources of the social world. Disability is a social issue.
The medical model of disability, on the other hand, locates the problem of disability within the bodies of disabled people. A person’s medical problems are what disable them, and these functional impairments would remain true in any social context. Disabled people’s well-being would be improved, according to the medical model of disability, if their bodies functioned better. According to the medical model of disability, disability is a personal health problem.
Sociology of Disability Historically
Earlier years in sociology framed disability within a medical model. The disabled person would be expected to take on the role of sick-person (Parsons 1951). The subjective experiences of disabled people were not relevant in Parsons’ model (Barnes and Oliver 1993), and what was important were the views of the responsible parties in society – i.e. the medical providers.
If not the sick role, then the rehabilitation role was called for (Safilios-Rothschild 1970), in which disabled people were expected to rehabilitate, to mightily fight against whatever ailment, injury or disease had taken hold of them. Professionals were to help the disabled people learn to cope through psychological stages of things such as shock, denial, anger, and depression (Flaig 1978).
These early sociology of disability theories are medical models of disability, and frame disability as functional impairment, a personal health problem. There is much to challenge the predictive validity of these theories in addition to what might seem obvious to the reader as a lack of face validity considering them from today.
They are also deterministic. They omit social factors. And they do not account for the subjective experiences of the disabled person. Historically this went to the extreme of sociological theory being used to teach providers how to use psychological tools of behavior modification to manipulate and control disabled people (Flaig 1978:654), thus threatening autonomy. In sum, early sociology of disability theory was medical model, “personal-problem” oriented, and can collectively be referred to as theory of disability as personal tragedy (Oliver 1986).
Why has it endured? In sociology, we know that our institutions are nothing if not durable, and the durability of the medical model is understandable perhaps because it inherently exonerates medical providers of responsibility. Medical providers determine treatment goals, yet if goals are not met that can be due to the shortcomings of their patient (Barnes and Oliver 1993:5). That is compelling for those working in medical fields. The medical model exonerates other parties too, everyone except for the disenfranchised disabled people themselves.
The sick role and its derivatives are also sometimes conceptualized more broadly as social deviance (Becker 1966; Lemert 1967). This linkage of disability with deviance contributes to a negative view of disabled people, particularly within capitalistic societies which value productivity, and within individualistic societies, which value independence. The deviance would be greater for a disabled person additionally marginalized by race, class, gender or other status. Disability as deviance became important in sociology of disability historically, for looking at how labeling produces stigma (Goffman 1963).
Intersectionality also plays a crucial role in this context. Individuals in the sick role are often marginalized to the extent that they are perceived as socially deviant, dependent, or unproductive. Additionally, members of otherwise marginalized social groups may struggle to even be recognized in the sick role, despite having similar impairments or ailments.
A classic example of the latter consideration is the categorization of “drug addicts” in the U.S., where perceptions often vary based on race. Individuals with the same addiction might be viewed differently: a “White addict” is often seen as needing medical care and thus placed in a sick role, whereas a “Black drug user” might be viewed as needing incarceration.
Goffman suggested that the terms used for disabled people (i.e. the blind man, the drug addict, the psych patient) involved an application of stigma which was a result of social interactions between so-called normal and so-called abnormal people. A noteworthy limitation here is that Goffman based his theorizing solely on the perceptions and experiences of non-disabled people (Barnes and Oliver 1993:6).
Another broad category of work in sociology of disability has involved understanding disability as incapacity, and framing the social processes involved as normalization of that incapacity in terms of a disabled person’s existing role relationships (Haber and Smith 1971). The premise here was that disabled people exhibit “inadequate role performance” but they were not rightly expected to perform the duties of the roles they would otherwise have had.
Thus, a legitimization process brought together some modified expectations into a new pattern. This normalization process became a new topic for sociological inquiry in this conceptualization of disability as “a form of adaptive behavior provided for by the norms of role relationships” (Haber and Smith 1971:87). In other words, while they also defined disability as deviant behavior, it was now specifically “the pattern of behavior emergent from incapacity” (Haber and Smith 1971:88).
Disabled People and Accessibility
As the late disability activist Stella Young once said in an interview with ABC News Australia, “No amount of standing in the middle of a bookshelf and radiating a positive attitude is going to turn all those books into braille” (Australian Broadcasting Corporation 2014). Her quote demonstrates the absurdity of attempting individualistic solutions to social, inaccessibility problems.
There is a similar anecdote common in disability justice spaces: If you cannot reach something on a very high shelf, just keep trying to become taller. This is especially true if you are going to be seeking reasonable accommodations. Before requesting them, make sure that you have first tried to become taller, as you will be asked to recount such efforts.
What about Medical Sociology?
In the 1980s, the field of Disability Studies was born in the United States, and it first grew from within Medical Sociology. Irving Zola was a disability activist and Medical Sociologist, who was disabled from a childhood polio infection, later earning his PhD in Sociology from Harvard. Zola was on faculty at Brandeis where we had a prestigious career, won many awards and held numerous leadership positions in ASA, as a consultant to the WHO, and worked as a member of President Clinton’s transition team on healthcare. He once said that “until we own our disability as an important part, though not necessarily all, of our identity, any attempt to create a meaningful pride, social movement, or culture is doomed” (Pace 1994).
The social model of disability can be linked to a number of sociological theories in certain ways, but it is most directly connected to materialistic perspectives (Watson and Vehmas 2019:20). Similar to how Karl Marx might consider how life is patterned by power relations and economic processes, the basis of the social model of disability is asking macro-level questions about how specific relations of power and economic structures have formed in a way that has disallowed disabled people access to participate in society.
Those barriers to equal participation, that lack of accessibility, is a lack of real capability, an impediment to well-being, and a human rights issue. It is also a social issue called disability within a social model framing.
These insights were largely ignored in sociology at-large until at least the turn of the 21st century, with sociology of disability still centering on medical model and deviance framing. Parsons’ influence is enduring: non-disabled is normal, disabled is studied as a sick-role, and one has a sort of relative legitimacy to get out of obligations in that sick-role, being deviant, but only to the extent that one is fighting to return to normal and become non-disabled again. At no point should the disabled person think of their status as anything but “abhorrent” (Parsons 1951:312-13).
This lingering influence can be seen in a burgeoning literature about how disabled people “adapt to the onset of chronic illness and impairment” (Watson and Vehmas 2019:20) and much sociology of disability literature centers on personal problems, an individualistic framing of disability as functional impairment, and of that impairment as social deviance.
Recent years have seen a promising shift in the landscape of sociological research where disability equality is concerned. This transformation is evident in social movement literature and in the growing recognition of disability as a politicized identity, which opens new avenues for scholarship. Additionally, sociologists are increasingly acknowledging the existence of disabled artists and a vibrant disability community, rich with its own culture and identity, which offers fertile ground for further study and shared knowledge. This emerging focus invites a more collaborative and inclusive approach to understanding these dynamic social elements.
References
Cartoon Explainer of Concepts
This cartoon explainer appeared on the poster I shared at PAA 2024. It was near the top left, under the heading of Disability Primer. On that day, this website served an an accessibility tool for my poster presentation, and this transcript was more prominently featured.
I’ve kept it here more as a demonstration of accessibility in data science for my readers, although a future web visitor to this page might also appreciate it. A descriptive transcriptive follows immediately below the graphic.

This cartoon introduces two relevant IPUMS MICS variables (diffsee and glasses) used in the construction of the two different disability indicators used in the two regression models in the present study (medical model and social model). It also alludes to the capability approach, central to the theoretical grounding, and it even touches on how both accessibility and functional disability fit within both the theory and the models.
It does a lot for a cartoon. Including it on my poster made my research more memorable (anecdotally, I was told), easier to engage with for a wider audience (for networking across disciplines), and was generally more accessible. I say all that only to acknowledge the obvious and overarching exception that all poster sessions privilege the participation and presence of non-disabled people, most notably sighted people. Although I am sighted, it took a small team of heroes to make it possible for me to attend PAA 2024 in spite of so many barriers, and I mostly could not participate.
The inclusion of this cartoon explainer was a topic of discussion for several people who visited me at PAA 2024. Some take-aways included how it helped bridge the space between the research silos of potentially distinct fields, thus potentially fostering interdisciplinary collaboration. I talked to someone whose research is in global policy, for example. She told me that she wouldn’t have stopped to talk if not for this cartoon.
This also lets anyone have a quick understanding of key concepts and, with interest, a way to ask questions. It’s also more accessible to researchers with ADHD heedless of their familiarity with my work or this topic in general.
Descriptive transcript
This is a comic-style cartoon explainer with seven frames. There is one main character, Maria. She’s a woman with medium brown complexion, and lots of hair styled neatly and high above a headband.
frame 1: Maria is sitting at a desk with a stack of books, one open book, and a steaming beverage. She has a frustrated or sad facial expression. Maria says to herself, “I have a LOT of difficulty seeing! This affects SO many things – work, family, school, friendships…”
frame 2: A black and white silhouette of a man is shown standing tall in front of the scales of justice. He replies, “That sounds like a YOU problem.”
frame 3: Maria is musing, thinking to herself in a thought bubble, “If only I had glasses.”
frame 4: The silhouette man replies to Maria’s introspection, saying, “Okay. Hmm. If you had glasses, could you drive, work, and do what you value?”
frame 5: Maria replies to the silhouette man, “Wow. If I had glasses, I would have no difficulty seeing. Then I could drive, do my job, and so much more.”
Frames 6 and 7 are the teaching frames in summary. In between them is a teacher, pointing to a chalkboard, emphasizing that these two frames are instructive.
frame 6: There is a small group of disabled people off to the side. One person is sitting in a wheelchair, another person is missing a leg, a third is wearing dark sunglasses suggestive of being blind. The text overlay for frame 6 reads: “In the MEDICAL MODEL, difficulty seeing is the disability, a personal problem.” Difficulty seeing is in italics.
frame 7: Decorative graphics frame the text overlay for the last frame which reads: “In the SOCIAL MODEL, lack of access to glasses is the disability, a social issue.” Lack of access to glasses is in italics.
Background
Please consider too the social model of disability section, which includes discussion of background related to sociology of disability.
Disability in Global Development
According to the United Nations, “Disability results from the interaction between persons with impairments and attitudinal and environmental barriers that hinder their full and effective participation in society on an equal basis with others” (United Nations 2006).
The World Health Organization defines disability similarly, as “the outcome of the interaction between individuals with a health condition (e.g. cerebral palsy, Down syndrome or depression) and personal and environmental factors (e.g. negative attitudes, inaccessible transportation and public buildings, and limited social supports)” (WHO 2021:10).
And there is an increasing trend of researchers disaggregating demographic data by disability, as recommended by many (i.e. United Nations (UN) Convention on the Rights of Persons with Disabilities (CRPD); UN Statistical Commission; World Health Organization (WHO); The World Bank; Organisation for Economic Co-operation and Development (OECD)).
What is the Washington Group Short Set?
The Washington Group Short Set (WG-SS) has become foundational for how we operationalize disability in global health research today, for the purpose of assessing a country’s progress towards sustainable development goals as well as their compliance with the UN CRPD (Madans, Loeb and Eide 2017; Washington Group 2013).
The WG-SS was developed by the Washington Group on Disability Statistics, a UN Statistics Commission working group formed after the 2001 International Seminar on the Measurement of Disability in New York to deal with conceptualization and measurement. Standardizing measurement is essential for comparability.
From the Washington Group on Disability Statistics website –
The WG Short Set of six questions on functioning for use on national censuses and surveys was developed, tested and adopted by the Washington Group on Disability Statistics (WG). The questions reflect advances in the conceptualization of disability and use the World Health Organization’s International Classification of Functioning, Disability, and Health (ICF) as a conceptual framework.
In a break from the past and the medicalization of disability that placed disability within the person and characterized it by impairments or deficits in bodily functions, the ICF presents a bio-psychosocial model that locates disability as at the interaction between a person’s capabilities (limitation in functioning) and environmental barriers (physical, social, cultural or legislative) that may limit their participation in society. The WG-SS used the ICF as a framework, focusing on the component of activity limitations.
The WG-SS has six basic questions, and they measure functional impairments, specifically related to seeing, hearing, walking, cognition, self-care, and communication. The rationale of how functional impairment measures are something other than medical model is that they can point to potential barriers to a person’s full participation in society.
Reflection on Study 1, Which Lead to Study 2
While analyzing evaluative well-being across all countries in the IPUMS MICS-6 dataset (n=513,744 women, in 29 national samples), Costa Rica emerged as an important case study for several reasons. Costa Rica had the highest life evaluation of all surveyed countries. Costa Rica also had the highest disability prevalence among women.
Importantly, in Study 1, I was using IPUMS MICS disability variables, which are based on the WG-SS as is industry standard, and which makes my Study 1 a medical model study.
I was led to more fully investigate how it could be that the women in Limón who were not disabled had among the highest evaluative well-being of women in Costa Rica, while the women in Limón who were disabled had the lowest evaluative well-being.
Social Stress Model
To better understand these disparities, I focused on the social stress process (Dohrenwend 1978; Pearlin 1989). What are the exogenous variables, and what might be some of the stressors forming that multidimensional array? In particular, status strains and ambient strains seemed probable, the latter given that women in Limón Province are geographically set apart as compared to women in any other province.
Limón is a distinct subnational region of Costa Rica in several other ways, including that the region has the fewest resources and is home to most of the country’s Black/Afro-Costa Rican population.
And even though the women with the lowest evaluative well-being were disabled, and those with the highest evaluative well-being were not disabled, they were all neighbors in Limón. To not consider a myriad assembling of factors as creating their current situation would have been to risk potentially accepting a mere “proxy indicator of chronic hardship” (Pearlin 1989:245), which remains one of my favorite phrases.
I like that phrase because it gets at intersectionality (Crenshaw 1989) in the stressors, and the reality of multiple status strains, also important to understanding a situation. This holds at the subnational level in terms of the history of cohorts, and the accumulation of culture.
I suspect therein lies some of the joy that buoys the non-disabled women in Limón, because I have come to appreciate that there’s no place like it. While the women of Limón have this obvious potential vulnerability to a constellation of stressors, they also have this obvious access to a constellation of coping resources, just not the kind that’s needed when it comes to disability. This is reflected in the development-disability gap suggested by Figure 1, Study 1. And this is what motivated Study 2.
Capability Approach
According to the capability approach, well-being is evident through people’s freedom to be and to do the things that they have reason to value (Sen 1999). Somehow that freedom accounts for how the non-disabled women in Limón, the subnational region with by far the fewest resources, are the happiest. They have something more valuable than the resources normally tallied.
The only thing they can’t necessarily overcome, apparently, is disability. That’s why the existing ways of doing demographic research on disability are not emancipatory: there is no pathway to making people become non-disabled.
Disability does something. And if development is to be sustainable, countries must leave no one behind. The United Nations (UN) has Stakeholder groups – non-state actors who regularly engage with the UN to influence and contribute to its initiatives, policies, and decision-making processes. One of them is the Stakeholder Group of Persons with Disabilities.
Reading their position papers as submitted to the UN High Level Political Forum on Sustainable Development (HLPF) over the years has been instructive. While they contributed a robust, 17-page paper in 2016, which had 26 footnotes and the endorsement of an impressive global array of 312 Disabled Persons’ Organizations (DPOs) drawing from every region of the world, most recently in 2023, they submitted just over one page of text, with no references or co-signors. It was primarily a plea for accessibility and inclusion:
“… we do not want to be left behind. Despite these (tasks and labors which this group has done towards sustainable development worldwide) persons with disabilities were largely left out of the national-level consultations. DPOs are looking for opportunities to work with governments, and many are being turned away. Public consultations often exclude persons with disabilities themselves and their representative organisations. Even when wider society is invited to participate, meetings and documents are not accessible for many persons with disabilities, thus excluding them from democratic processes…” (Stakeholder Group 2023:2).
Sustainable development must be disability inclusive development, and development is not sustainable if persons with disabilities are left behind.
And so it is that the theoretical framework undergirding the present study is where the capability approach meets sustainable development for disabled women in Costa Rica.
Hypotheses
Hypothesis 1: The results from the social model of disability will show significantly different well-being gaps compared to the medical model, due to the differing conceptualization of disability.
Hypothesis 2: Using the social model approach will reveal unique well-being gaps that are not observable when applying the medical model, particularly in the context of access to resources such as eyewear.
Contributions
My hope with this study is to advance demographic scholarship and theory around well-being and the operationalization of disability, especially for sociologists who do quantitative research in population health and demography. While there are limits to what I can do with the limited data that are available at the time of Study 2, I hope to articulate this theoretical argument more robustly in this chapter, using these suggestive results as preliminary and hopeful.
Sociologists are already doing such important demographic research, telling the data stories that can illustrate inequalities that inform policy agendas and potentially identify well-being gaps necessary for the world to be better than it is.
For that progress to include disabled people, sociologists must think about disability and disabled people sociologically. Then, ideally, our research can point to solutions that improve accessibility, human rights, and well-being for disabled people, especially for disabled Black and Indigenous women. If we do not begin conceptualizing sociologically, we may continue this practice of “counting crips” and there is no solution for that which is emancipatory.
References
Methods
Data Source
This research will analyze the IPUMS MICS Round 6 for Costa Rica. Please find a Methods discussion: About IPUMS MICS on the home page for this prospectus.
Dependent Variable: Life Evaluation Measure
The outcome measure of evaluative well-being in this study is the Cantril ladder item, from IPUMS MICS Round 6. See also: Methods discussion: About the Cantril ladder.
Key Independent Variables: Constructed Disability Indicators
The regression models in this study (medical model and social model) examine the associations between well-being and disability within the theoretical framework largely established by the Washington Group.
Medical Model: disabled
For the medical model, disability is functional impairment. The indicator of disabled in the present study is constructed from an IPUMS MICS survey item that asks whether respondents have difficulty seeing. If women reported at least “a lot of difficulty” seeing, they are disabled. This cutoff aligns with WG recommendations for comparability.
While there are additional survey items that measure functional impairment across more domains, this indicator is constrained to the domain of vision so that it is comparable to the social model which, due to data limitations, can only consider accessibility related to vision.
Social Model: lackAccess
For the social model, disability is lack of access. The indicator of lackAccess in the present study is constructed from two IPUMS MICS items. For women who reported specifically ‘a lot of difficulty’ for difficulty seeing, and who also reported that they did not wear eyeglasses, they lackAccess. Due to data limitations, this indicator does not span across other forms of access.
Key Independent Variable: ethnicity
ethnicity: constructed binary indicator; Costa Rica is about 2/3 White and due to data limitations, ethnicity of household head needed to be binned into White and People of Color (Non-White respondents).
Controls
wealth: binary indicator constructed from wealth index score such that 1 = women in households with a positive wealth index score and 0 = women in households with a negative wealth index score.
marital: binary indicator constructed using marital status; categories were combined so that women were 1=married or partnered, and 0=not married, with the latter including women who reported being divorced, widowed, separated or never married.
education: highest level of school attended by the woman, categorized into four ordered levels (less than primary, primary, secondary, or tertiary/higher/university).
region: reports the categorical, geographic subnational region according to which of the seven provinces the woman lives within in Costa Rica.
Regression Models
Both models in this study share an outcome measure (Cantril ladder), and are therefore modeled using ordered logistic regression. For a more detailed explanation of the rationale for this, please refer to the Ordinal Regression Model discussion, in the Methods section of the previous chapter.
Social Model
The social model predicts the probability of the outcome variable, a proxy for evaluative well-being, being less than or equal to its various cutpoints, conditioned on several predictors, which include ethnicity, wealth, education, marital status, perceived discrimination due to disability, log-transformed age, the interaction of ethnicity and lackAccess, and the social model disability indicator itself, lackAccess, as previously defined.
-
Where j indexes the cutpoints of the 7-level ordered outcome variable Y, which is the binned Cantril ladder, life evaluation measure,
logit(P(Y ≤ j)) = -
βj0 + β1 lackAccess + β2 ethnicity + β3 wealth
+ β4 edlevelwm + β5 married + β6 discriminated
+ β7 log(age) + βinteraction (lackAccess × ethnicity)
Medical Model
The medical model also predicts the probability of the same outcome variable being less than or equal to its various cutpoints conditioned on several predictors. The medical model differs from the social model in two ways: the medical model does not include discrimination as a factor, and it includes the medical model disability indicator, disabled, as previously defined, rather than the social model disability indicator.
-
Where j indexes the cutpoints of the 7-level ordered outcome variable Y, which is the binned Cantril ladder, life evaluation measure,
logit(P(Y ≤ j)) = -
βj0 + β1 disabled + β2 ethnicity + β3 wealth
+ β4 edlevelwm + β5 married + β6 log(age)
+ βinteraction (disabled × ethnicity)
Preliminary Results
- Non-disabled women in regions with fewer economic resources have a remarkably high predicted probability of attaining the highest levels of evaluative well-being.
- Significant interactions emerged between ethnicity and lack of access. And the well-being penalty associated with disability is greater for Black and Indigenous women in regions with more economic prosperity.
- This suggests a development-disability gap.
Province & Ethnicity1 | n (disabled)2 | Proportion ‘disabled’ (as %) |
---|---|---|
Heredia – White residents (n = 182) | 14 | 7.7% |
Heredia – People of Color (n = 232) | 10 | 4.4% |
Limón – White residents (n = 431) | 34 | 7.9% |
Limón – People of Color (n = 628) | 36 | 5.8% |
1 For brevity, Tables 1 through 5 summarize weighted means and frequencies for White residents and People of Color in Heredia and Limón, as the provinces with the most and the fewest resources respectively.
2 The ‘disabled’ indicator represents ‘a lot of difficulty’ seeing or ‘cannot see at all’. See also the complete Stata Code section for more.
Note. People of Color show a disproportionately lower rate of reportedly having ‘a lot of difficulty’ seeing or ‘cannot see at all’ as compared to White residents across provinces. White residents’ rate is similar in Heredia and Limón. These initial surprises may reflect measurement error, bias, differences in reporting, or something else. They are likely reflecting something other than an aspect of disability prevalence, because extant data affirm that disability prevalence is higher across the board for People of Color in Costa Rica. See also Table 6 for some granularity.
Province & Ethnicity1 | n (lackAccess)2 | Proportion ‘lackAccess’ (as %) |
---|---|---|
Heredia – White residents (n = 182) | 6 | 3.2% |
Heredia – People of Color (n = 232) | 6 | 2.5% |
Limón – White residents (n = 431) | 23 | 5.3% |
Limón – People of Color (n = 628) | 22 | 3.4% |
1 For brevity, Tables 1 through 5 summarize weighted means and frequencies for White residents and People of Color in Heredia and Limón, as the provinces with the most and the fewest resources respectively.
2 The ‘lackAccess’ indicator combines having exactly ‘a lot of difficulty’ seeing with having no eyeglasses. See also the complete Stata Code section for more.
Note. Data limitations are apparent here, yet there is some preliminary information of use. Limón residents clearly lack access on this measure more than residents in provinces with more resources. Despite obvious data limitations, need for further inquiry is suggested, with greater statistical power, to better understand the role of accessibility. See also Table 7.
Province & Ethnicity1 | n (‘discriminated’)2 | Proportion ‘discriminated’ (as %) |
---|---|---|
Heredia – White residents (n = 182) | 25 | 13.5% |
Heredia – People of Color (n = 232) | 35 | 15.1% |
Limón – White residents (n = 431) | 32 | 7.5% |
Limón – People of Color (n = 628) | 56 | 8.9% |
1 For brevity, Tables 1 through 5 summarize weighted means and frequencies for White residents and People of Color in Heredia and Limón, as the provinces with the most and the fewest resources respectively.
2 The ‘discriminated’ indicator includes perceived discrimination on the basis of disability, age, or other reason only. See also the complete Stata Code section for more.
Note. Ethnicity does not seem to be a big factor in who reports perceived discrimination on the basis of disability, age, or other reason. Province does seem to matter a lot, with residents in Heredia reporting discrimination at much higher rates. This could reflect a reporting bias, in which Heredia residents are possibly more educated on their rights or perhaps feeling more empowered to assert them. Differences between provinces could reflect local culture. These differences could also be a matter of measurement error related to question order or something else.
Province & Ethnicity1 | Less than primary (%) | Primary (%) | Secondary (%) | Tertiary / higher / university (%) |
---|---|---|---|---|
Heredia – White residents (n = 182) | 0.16% | 11.55% | 30.73% | 57.56% |
Heredia – People of Color (n = 232) | 0.57% | 10.09% | 46.81% | 42.53% |
Limón – White residents (n = 431) | 2.22% | 29.06% | 51.90% | 16.82% |
Limón – People of Color (n = 628) | 1.39% | 24.58% | 53.70% | 20.33% |
1 For brevity, Tables 1 through 5 summarize weighted means and frequencies for White residents and People of Color in Heredia and Limón, as the provinces with the most and the fewest resources respectively.
2 The ‘education’ variable from IPUMS MICS is a 5-level ordinal measure of highest level of school attended by women. The Costa Rica sample had only 4 levels in all but one province. See also the complete Stata Code section for more.
Note. Both ethnicity and province factor separately and together towards highest level of school attended by the woman. And this variable syncs with SDG 4 (Quality Education) and Targets 4.1, 4.3, and 4.5. If disabled women (by either operationalization) are left behind on this measure, that’s an important data story also.
Province & Ethnicity1 | n (‘wealth’)2 | Proportion ‘wealth’ (as %) |
---|---|---|
Heredia – White residents (n = 182) | 159 | 87.8% |
Heredia – People of Color (n = 232) | 194 | 83.4% |
Limón – White residents (n = 431) | 212 | 49.2% |
Limón – People of Color (n = 628) | 278 | 44.3% |
1 For brevity, Tables 1 through 5 summarize weighted means and frequencies for White residents and People of Color in Heredia and Limón, as the provinces with the most and the fewest resources respectively.
2 The ‘wealth’ indicator represents anyone whose wealth index score is above zero. The wealth index itself is a standardized distribution (with a mean ~ 0 and a standard deviation ~ 1) intended to reflect the effects of wealth without directly measuring it, computing a wealth index score mostly by tallying ownership of household goods and basic services.
This household wealth estimation practice is based on the popular Filmer and Pritchett method of doing so. In this study, ‘wealth’ is 1 for all positive wealth index scores, which is approximately the 50% above the median for the country, given the normalized index from which it is constructed. See also the complete Stata Code section for more.
Note. Wealth varies in big ways by province and in small ways by ethnicity, and an interaction between province and ethnicity is likely here. Since a chi square test for independence cannot be done with survey data in Stata, relationships between ethnicity, province and ‘wealth’ could be interrogated via logistic regression, i.e. (svy: logistic wealth i.geo1_cr##i.ethnicity). I could also compute a chi square statistic using a pencil and paper.
Table 6: Proportion of ‘disabled’ by Province and Ethnicity
Province | Ethnicity | ‘disabled’ Proportion (%) |
---|---|---|
San José | White | 36.95% |
POC | 30.11% | |
Alajuela | White | 20.05% |
POC | 18.26% | |
Cartago | White | 13.20% |
POC | 9.56% | |
Heredia | White | 10.65% |
POC | 11.11% | |
Guanacaste | White | 4.52% |
POC | 9.31% | |
Puntarenas | White | 7.16% |
POC | 11.68% | |
Limón | White | 7.47% |
POC | 9.96% |
See also the complete Stata Code section for more.
Note. This table depicts the proportions of ‘disabled’ women stratified by ethnicity and province.
Table 7: Proportions of ‘lackAccess’ and ‘discriminated’ by Province and Ethnicity
Province and Ethnicity – ‘discriminated’ | lackAccess – Proportion |
---|---|
San José White – No | 35.54% |
San José White – Yes | 46.58% |
San José POC – No | 29.28% |
San José POC – Yes | 35.59% |
Alajuela White – No | 20.41% |
Alajuela White – Yes | 17.57% |
Alajuela POC – No | 18.12% |
Alajuela POC – Yes | 19.18% |
Cartago White – No | 13.39% |
Cartago White – Yes | 11.93% |
Cartago POC – No | 9.74% |
Cartago POC – Yes | 8.42% |
Heredia White – No | 10.56% |
Heredia White – Yes | 11.26% |
Heredia POC – No | 10.88% |
Heredia POC – Yes | 12.67% |
Guanacaste White – No | 4.71% |
Guanacaste White – Yes | 3.19% |
Guanacaste POC – No | 9.44% |
Guanacaste POC – Yes | 8.45% |
Puntarenas White – No | 7.47% |
Puntarenas White – Yes | 5.09% |
Puntarenas POC – No | 12.09% |
Puntarenas POC – Yes | 8.96% |
Limón White – No | 7.93% |
Limón White – Yes | 4.38% |
Limón POC – No | 10.45% |
Limón POC – Yes | 6.73% |
See also the complete Stata Code section for more.
Note. This too-long table depicts the proportions of ‘lackAccess’ and ‘discriminated’ stratified by province. Overall, the regions with more resources report lackAccess at higher proportions. And in regions with fewer resources, groups of people who experienced discrimination had lower lackAccess. This could reflect measurement error, reporting bias, or something else.
Predictive Graphics

Note. Even with the data limitations, it does seem that the different ways of operationalizing disability may reveal different things in terms of who’s left behind.
Limitations
While there is a compelling case to be made for Costa Rica as a case study, and for prioritizing women’s well-being in order for development to be sustainable (leaving no one behind), limitations range from the obvious to the less-so.
Generalizability is limited. Without longitudinal data, I cannot investigate causal relationships or really interrogate changes in an aggregated outcome over time because these data are cross-sectional, and those conclusions would not be justified.
Another limitation is using life evaluation as a sole outcome measure. Much validation work has been done on this indicator of well-being, and yet perhaps other things have led to the outcome as measured – unnamed factors, unaccounted for response bias.
Also, this present study, with two starkly different disability indicators, risks losing nuance of what it means to be disabled. While ethnicity can be an important factor, I reluctantly had to bin small (not-White) categories due to a lack of statistical power. In several aspects of this study there are limitations confounded by the small subgroup sizes of disabled women, which can also lead to estimation issues.
The social model (the regression model in this study so named) does reach across more people by including perceived discrimination according to disability, age, or other reason, but it can only account for a lack of accessibility, the central point of this study, related to one domain of functional disability (vision). This limitation is inherent due to data limitations.
The hope is that upcoming surveys will continue to improve on their disability data practices, so as to begin narrowing the well-being and development gaps experienced by disabled women and especially by additionally minoritized women such as Black/Afro-Costa Rican and Indigenous women.
As the graphics on the PAA 2024 poster and elsewhere on this webpage depict, the well-being (Cantril ladder) gaps that are evident when using the social model versus when using a medical model are more of a minor translation than a major shift. This may be due to their shared, convergent validity and also the data limitations of only considering vision and glasses.
But when we are considering the lives of people who are already multiply disadvantaged, a model that somewhat captures the additional burden of inaccessibility may be a better fitted model to predict well-being.
Implications for subsequent research include exploring this matter of convergent validity more fully, and perhaps testing better disability survey questions that give disabled people a chance to report something of the accessibility of their social world rather than just their diagnoses and functional impairments, which are their personal problems.
Lastly, I have not been able to find accessibility information for many global health surveys, but presumably, or ironically, our global health data are largely comprised of the responses from people who are not very disabled.
I feel that this matter of leaving no one behind may remain a lingering wish if we don’t even have a way of obtaining information from disabled people in LMICs.
And if we all, as researchers, continue to accept that our survey data is comprised of the responses provided from the largely non-disabled populations, we are not helping that validity issue, that sustainability matter, or those people.
As accessibility is a social issue, its provision necessary for sustainable development, good answers to better questions could lead to more relevant data stories being shared in national and international policy briefs. If disabled people are to be not left behind, we’re going to have to ask them how they are doing, in order to find out what barriers they are experiencing.
Stata Code
Loading Data
/******************************************************
LOADING DATA
******************************************************/
clear all
use "disscleaned.dta" , clear
* this file was created using the steps shared in study 1 Stata Code tab
Cleaning Data and Creating Indicators
/**********************************************************
CLEANING AND CREATING INDICATORS
FOR STUDY 2
***********************************************************/
* already have 'disabled' (medical model disability indicator)
* create 'lackAccess' (social model disability indicator)
gen lackAccess = (glasses == 0 & diffsee == 3)
replace lackAccess = 0 if missing(lackAccess)
label variable lackAccess "Lacks Access - Vision (A lot)"
label define lackAccess 0 "No" 1 "Yes"
label values lackAccess lackAccess
tab lackAccess
/* Explainer for lackAccess:
This will create the lackAccess indicator, a proxy for lack of access,
representing people who have a lot of difficulty seeing, yet no glasses.
Everyone who has 'no difficulty' seeing, or only 'some' difficulty, as well
as everyone who cannot see at all, and all those who have glasses, are
zero on this indicator. This indicates only those with a lot of difficulty
seeing + no glasses.
*/
Descriptives
/*****************************************************
Let's-a-go!
*****************************************************/
* starting point with this study's survey design:
svydescribe
/*
From the above, I note there are 32 strata with 2-45
units in each. With 500 PSUs,
How many observations are there per cluster or PSU?
Minimum of 2, mean of 16.4, maximum of 33
Observations in this design are female respondents.
And the degrees of freedom for this design would thus be
500 - 32 = 468 design df (n-PSUs minus n-strata). This
basically represents the number of independent pieces
of information available for estimating variances.
*/
/*****************************************************
DESCRIPTIVES
******************************************************/
* here is the list of variables that should be in one's consideration
codebook disabled disabled_lot lackAccess ethnicity ///
wealth education marital agewm lnage discrimination ///
diffsee glasses round year CR* region ethnicity lsbin ///
weightwm cluster stratum psu, compact
* Who lacks access? Who is disabled?
* Who has a lot of difficulty seeing + no glasses?
* Who is disabled with 'a lot of difficulty' seeing?
svy, subpop(ethnicity): mean ///
lackAccess disabled ///
discriminated edlevelwm wealth, ///
over(region)
*interrogating
estat size
* Descriptives for - key variables - weighted
* frequencies across binary ethnicity in
* Heredia and Limón, the Provinces w/most
* and least wealth
* Table 1
//disabled - weighted frequencies and CIs -
* Heredia - White
svy, subpop(if ethnicity == 0 & region == 4): tabulate disabled, ci
* Heredia - POC
svy, subpop(if ethnicity == 1 & region == 4): tabulate disabled, ci
* Limón - White
svy, subpop(if ethnicity == 0 & region == 7): tabulate disabled, ci
* Limón - POC
svy, subpop(if ethnicity == 1 & region == 7): tabulate disabled, ci
* Table 2
//lackAccess - weighted frequencies and CIs
* Heredia - White
svy, subpop(if ethnicity == 0 & region == 4): tabulate lackAccess, ci
* Heredia - POC
svy, subpop(if ethnicity == 1 & region == 4): tabulate lackAccess, ci
* Limón - White
svy, subpop(if ethnicity == 0 & region == 7): tabulate lackAccess, ci
* Limón - POC
svy, subpop(if ethnicity == 1 & region == 7): tabulate lackAccess, ci
* Table 3
//discriminated - weighted frequencies and CIs
* Heredia - White
svy, subpop(if ethnicity == 0 & region == 4): tabulate discrimination, ci
* Heredia - POC
svy, subpop(if ethnicity == 1 & region == 4): tabulate discrimination, ci
* Limón - White
svy, subpop(if ethnicity == 0 & region == 7): tabulate discrimination, ci
* Limón - POC
svy, subpop(if ethnicity == 1 & region == 7): tabulate discrimination, ci
* Table 4
//edlevelwm - weighted frequencies and CIs
* Heredia - White
svy, subpop(if ethnicity == 0 & region == 4): tabulate education, ci
* Heredia - POC
svy, subpop(if ethnicity == 1 & region == 4): tabulate education, ci
* Limón - White
svy, subpop(if ethnicity == 0 & region == 7): tabulate education, ci
* Limón - POC
svy, subpop(if ethnicity == 1 & region == 7): tabulate education, ci
* Table 5
//wealth - weighted frequencies and CIs
* Heredia - White
svy, subpop(if ethnicity == 0 & region == 4): tabulate wealth, ci
* Heredia - POC
svy, subpop(if ethnicity == 1 & region == 4): tabulate wealth, ci
* Limón - White
svy, subpop(if ethnicity == 0 & region == 7): tabulate wealth, ci
* Limón - POC
svy, subpop(if ethnicity == 1 & region == 7): tabulate wealth, ci
*Table 6
// 'disabled' **************************************
svy: tab region disabled, ///
count cellwidth(12) format(%12.2g)
svy: proportion region disabled, ///
over(ethnicity)
Analysis
/******************************************************
ANALYSIS
******************************************************/
//