Marked as Homeless: Reconciling with Ambiguities about Housing Status in Death Records

Abstract

Research on the process of classification has long revealed biases and limitations embedded within social data. Expanding or contracting definitions of homelessness and the perception of housing status as relevant information play a large part in data and knowledge about homelessness. The author explores how homelessness is conceptualized and documented (or not documented) within death records through the use of qualitative interviews with field investigators (n = 20) who attend and document circumstances related to suspicious deaths in New Mexico. The findings reveal uncertainty in marking a person as homeless that is newly described as labeling ambivalence, which can be resolved more systematically through a process of increasing social mattering in which the social determinants of health are accentuated as factors related to premature mortality. The findings suggest important considerations for more systematically describing housing status and other indicators of the social determinants of health within social records.

Throughout the process of classification or categorization, efforts to label, quantify, and respond to a given issue (e.g., poverty, mental illness) give way to the limitations and biases of the parties involved in the matter at hand (Hacking 1996; Whooley 2010). A unique application of issues related to classification can be found in the process of identifying people who were affected by homelessness within secondary data sets. Studying mortality trends among people affected by homelessness often involves reading through text descriptions or matching names and other identifiers with additional sources that can substantiate a previous experience with homelessness (Funk et al. 2022; NHCHC 2021). Accordingly, researchers and their decisions and judgments are part of the social construction of these data. Such work with death data informs mortality statistics and supports annual homelessness memorial vigils, in which the lives of people who died in the previous year are recognized across numerous communities in the United States. ¹ Given the potential of these data to influence public sentiment and inform social policy, it is important to understand the decisions and challenges associated with producing death data and labeling a person who has died as homeless.

In addition to having a lack of familiarity with definitions of homelessness, people who enter death data may be influenced by perceived stigma (Belcher and Deforge 2012; Goffman 1986; Rayburn and Guittar 2013) and may experience ambivalence in what documentation is relevant and needed (Alasuutari, Kelle, and Knauf 2020). As follows, a better understanding of how housing instability and homelessness is understood and constructed within social records is paramount. This study presents novel findings on the basis of interviews (n = 20) with field investigators across the state of New Mexico to better understand how homelessness is conceptualized and how indicators or signs of homelessness are reconciled with ambiguities, uncertainty, and tensions related to marking a person as homeless within narrative data.

Background

Categorizing People and Conceptualizing Homelessness

Categorizing groups of people is a form of scientific inquiry that has long been studied as subject to social construction (Hacking 1996; Pontikes and Kim 2017). The interplay between social construction and classification has been described in research on the collection of data on racial categories in the United States (Anderson and Fienberg 2000), the processes leading to the institutionalization of Hispanic ethnicity (Mora 2014), and understandings of citizenship (Kennelly and Dillabough 2008). Research on the classification of homeless decedents is a unique application to literature on classification and social construction, given that decisions to identify people as homeless are made in relation to death circumstances and are often decided from the perspectives of people who are not trained in homelessness or housing issues.

Defining homelessness for inclusion and exclusion has been a challenge for generations (Burt 1995; Cordray and Pion 1991; Gabbard et al. 2007; Hopper 1995; Shlay and Rossi 1992; Wright and Devine 1995; Yousey and Samudra 2018). This has been particularly pronounced within the U.S. Department of Housing and Urban Development Point-in-Time counts (Hopper et al. 2008; Smith and Castañeda-Tinoco 2019). Conceptualization for experts and nonexperts is challenging because of differing and nuanced definitions. For example, the Department of Housing and Urban Development defines homelessness as inclusive of the following abbreviated criteria: (1) people who are living in places not meant for human habitation; (2) people who are losing their primary nighttime residences within 14 days; (3) unaccompanied youth; and (4) people attempting to flee domestic violence. In comparison, the Health Resources and Services Administration defines homelessness as an individual who lacks housing (without regard to whether the individual is a member of a family), including an individual whose primary residence during the night is a supervised public or private facility that provides temporary living accommodations, and an individual who is a resident in transitional housing. The Health Resources and Services Administration definition can be viewed as broader because it is less restrictive on temporary living conditions and includes people living in transitional housing, a group who could be considered formerly homeless in some circumstances.

In clinical practice, definitions vary by programs and funding mechanisms. To reconcile these differences Amore, Baker, and Howden-Chapman (2011) conceptualized physical, social, and legal dimensions of homelessness and typologies on the basis of homelessness and housing exclusion. As the definitions have become more complicated and nuanced, it is likely that inconsistencies in data documentation have followed. This has implications for how data on homelessness and mortality are used by researchers and policy makers. Similar concerns may be raised across other activities in which homelessness or housing status is collected (e.g., sheltered and unsheltered prevalence counts, student homelessness, criminal justice data) (Lee, Shinn, and Culhane 2021; Lee, Tyler, and Wright 2010). It is also relevant to a growing movement to collect indicators of the social determinants of health in records more broadly (Stafford and Wood 2017; Weir et al. 2020).

The Social Construction of Data

Social notes are recorded within medical encounters and contained within electronic health records (Chen et al. 2011). Such documentation is often written as narrative text and is largely unstructured compared with structured medical codes used for diagnoses, billing purposes, or documentation of causes of death as in the case in death records (Gottlieb et al. 2015). Recent research has begun to analyze narrative text data within medical records to specifically study homelessness and housing (Bejan et al. 2018). Review teams have increasingly relied on narrative text to inform analyses on trends in death among several at-risk subgroups such as those affected by domestic violence (Wallace et al. 2020) and deaths among infants and young people (Bohn, Chiasson, and Huyer 2018), in addition to deaths of people affected by homelessness (NHCHC 2021).

Such existing data are part of the world of social booking (Merton 2000), or in other words, data (statistical and text) from agencies in society that collect the data as part of their work. Although existing data present some degree of opportunity for research without reactive bias, the process of constructing such data reveals important variation to be studied as well (Becker 2017). Indeed, there are differences in what is focused on as the foreground compared with what is background information or, rather, what is considered relevant compared with what is not (Brekhus 1998; Zerubavel 2015). Research on the construction of social notes used to study homelessness is essential to better understanding this form of data and how it can best be used in future research.

This study addresses the following questions:

1. Which indicators of homelessness are most frequently observed and described by field deputy medical investigators (FDMIs) in field notes in mortality data?

2. What barriers prohibit labeling decedents as homeless?

3. Which indicators of homelessness are observed in cases that are not ultimately marked as homeless, even when homelessness has not been ruled out?

Data and Methods

Qualitative Interview Data

This study makes use of original interview data with FDMIs, ² who write social notes recorded about death scenes, to explore how and why homelessness is revealed or concealed within death records. I formed relationships with the Office of the Medical Investigator through several years of community-engaged research that included collaborative work with homelessness services organizations (e.g., including work on several needs assessments, participation on a program review board, and formative work on a local community report on mortality trends among people affected by homelessness in New Mexico). Such work involved connecting with the New Mexico Office of the Medical Investigator and engaging with FDMIs about homelessness issues. These activities laid the groundwork for access to data and interviewees and helped formulate the study research questions. In addition, an early version of the study protocol and interview schedule was critically reviewed as part of a community-engaged research studio session that included people with lived experience of homelessness. Before research began, the study protocol was reviewed and approved by the University of New Mexico institutional review board.

Participants were recruited through an introductory e-mail and screening survey sent to all FDMIs in New Mexico by the field investigator training supervisor. The e-mail introduced the study and included a study consent form and a link to a brief survey using Opinio secure survey software. The screening survey included informed consent language and items to confirm a participant’s eligibility, such as ensuring that they had experience as field investigators in New Mexico, were older than 18 years, and were willing to be contacted to learn more about the research study. Participants were eligible for the study regardless of whether they had investigated cases that involved homelessness. After participants completed surveys, they were contacted by e-mail or phone to schedule an interview. All interviews were scheduled and completed within a period of three months in 2021 (May to August). Semistructured interviews were conducted and recorded using Zoom online meeting software. These files were then transcribed, formatted, and entered into NVivo version 12 for qualitative content analysis. Semistructured interviews lasted 45 to 60 minutes. They began with an introduction to the research study and questions about the process of doing field investigations, challenging cases, cases involving homelessness, and conversations with family members (and other witnesses). Questions specific to primary research questions included the following:

1. If you were to help a new field investigator recognize signs of experiencing homelessness, how would you explain certain clues that might suggest that a decedent was affected by homelessness or unstably housed?

2. If you were at a death scene and suspected that the decedent was affected by homelessness, what would your next steps be?

3. What kinds of conversations have you had with next of kin or witnesses about a decedent having been affected by homelessness?

Interviews concluded with demographic questions on age, gender, race and ethnicity, and the participant’s highest level of education. All participants were offered a $25 gift card that was sent by e-mail after the interview.

Data Analysis

Interviews (n = 20) were recorded and transcribed. Transcriptions were uploaded into NVivo version 12. Each interview was analyzed using an abductive approach to theory (Timmermans and Tavory 2012), in which responses were coded toward a priori research questions in addition to identifying emergent themes. The first coding tree included themes for barriers and facilitators to identifying people who were homeless. This expanded to include subthemes on the relevance of categorizing homelessness, problems with addresses, and the meaning of home. Coded data were used in several rounds of memo writing, which informed the organization of the article and application to theory and potential study implications. Member checking was also later completed through a presentation and discussion of findings and the overall conceptual framework with field investigators.

Summary Statistics

Participants had a mean age of 49 years, ranging from 22 to 79 years. Slightly more men participated (53 percent). Sixty-six percent identified as White non-Hispanic, and 33 percent identified as White Hispanic. Participants were skewed toward having more years of experience, with 60 percent reporting having more than 10 years of experience as field investigator, 10 percent reporting 5 to 10 years of experience, and 30 percent having fewer than 5 years of direct experience. All participants reported having greater than a high school education, with 10 percent having some college, 60 percent having bachelor’s degrees, and 25 percent having master’s degrees. Field investigators reported attending scenes in 13 of the 33 counties within New Mexico. Most field investigators conducted their work in a single county or represented several adjacent counties within one region of New Mexico.

Of note, during interviews, participants described having multiple related roles either before becoming a field investigator or concurrently acting in a related role. These roles included work as nurses, as paramedics, or within law enforcement. Accordingly, participants were able to draw on experience encountering people affected by homelessness from the perspective of a field investigator and other personal and professional perspectives. Nearly all participants had investigated scenes that involved decedents who were homeless. The few who had not reported having that experience had worked with people affected by homelessness in other roles.

Results

The findings reveal a number of key signals associated with interpreting a decedent as affected by homelessness. These include the place itself; the appearance of the decedent; descriptions of the decedent from others at the scene, including family, friends, and witnesses; engagement with homelessness services in some way; and related records from medical or criminal justice systems. In addition, cause of death, in particular deaths from exposure to the elements, may also indicate homelessness. Data organized below under key signals and indicators of homelessness serve to answer the first research question.

However, the presence of such indicators does not necessarily mean that homelessness or housing status will be documented in notes. Ambiguities and tensions exist and influence documentation as well. To this end, findings addressing the second research question are organized as ambiguities and tensions that lead to labeling ambivalence. Such factors include the presence of an identification card and an address, the presence of family with places for the decedent to stay, the amount of time a person has been without stable housing, and histories of homelessness. The contrast between these indicators and ambiguities creates tensions in which investigators may decide against documenting housing status to avoid making judgments and to attempt to maintain objectivity, thus avoiding the appearance of personal judgment on the living status or conditions they record. This may also coincide with thinking that housing status is not relevant to a report. Addressing the third research question, indicators and ambiguities are presented and analyzed with factors that help reconcile tensions in Figure 1.

Figure 1.

Reconciling signs of homelessness, ambiguities, and labeling ambivalence.

Key Signals (and Indicators)

Key signals were provided by field investigators when describing specific cases in which a decedent was affected by homelessness and further described while replying to follow-up questions about how they knew (or how they would try to determine) if a decedent was experiencing homelessness. Place and appearance were among the most common and earliest relied upon signals. As a field investigator working in the southwestern part of the state for almost 10 years reported,

When I am on scene the first one you look for is the location itself. Is this someplace where they have access to water and electricity, are they the legal occupant? Do they get their mail there, are they renting that place, are they squatting? A lot of it goes back to looking, and identifying where do they live, and what are the living conditions they are living in.

In this description, we see the process of examining a scene and its location and features. A place is a home if one is legally allowed to reside there. It is also the place where a person receives mail and a place to which one has access at all times. In contrast, a place is not a home if it is temporary or if there is not access to running water, heat, or electricity. In other cases, field investigators described the processes of corroborating their observations through interviews with witnesses such as family members or friends of the decedent. This comes through in a description provided by a field investigator from the southeastern part of the state with more than 10 years of experience:

Motels are another indicator, why are they in this motel . . . you’re going to wonder why are they here? . . . and when you talk to management, and see how long they’ve been there, or you contact family, that’s also something to ask, interviewing people, how long have they been here, that’s usually a good indicator. We like to know why that person was there at that time. . . . How did they end up in the river? Oh they were sleeping in the arroyo and then it rained, and it got washed out. Or why were they here?

Here we see how a housing status can be clarified by asking, “Why was the person at this particular place?” Additional locations and context described above include tents and motels. However, further substantiation comes from talking with witnesses: hotel or motel management, family, friends, and others. This is in relation to time as well. As such, being at a motel or having slept near an arroyo for more or less time appears to make a difference that comes through as part of follow-up interviews with people who can fill in missing context. Specifically asking where a person lives or usually stays can help corroborate housing status. A field investigator from the southwestern part of the state shared that sometimes friends and family will say that the decedent did not have a home:

How about when we look at it, there are the ones that stay with this person for some time and then, if you ask them if they live here they say no, well they don’t have a home, occasionally they will spend the night here, but they don’t have a home, they don’t have a place, everything they have, they have with them, and to me, that’s homelessness, if you don’t have a [place] to have your things, a place that you can go back to all the time.

The participant further describes the process of asking witnesses about a decedent and learning of homelessness by word of mouth. A person may be found in a home, but not actually live there, and then be described as not having a home or a place. Such cases might also involve a person keeping everything with them, without a consistent place to keep their things. Furthermore, the belongings a person does have may reveal details about their housing status through paperwork or documentation connecting them to shelter systems or to homelessness services. This scenario is further described by a field investigator with more than 10 years of experience working the central and northwestern parts of the state:

Sometimes they have documents from the homeless shelter on them, so we know that they are living off of a homeless shelter somewhere out there. Or maybe there is someone there like an acquaintance who says he doesn’t have a home, we live on the street or we live at this homeless shelter, and at that point you know you can kind of tell sometimes . . . but it’s not often the case that you walk up to a scene and say yeah he’s homeless, it takes some research to determine that.

As indicated, another corroborating indicator of homelessness is documentation of shelter or services. The participant notes how documents from a shelter can point toward a history of homelessness that would likely be observed and included in their report of the scene and death circumstances. In addition, the investigator notes how additional signals come from the place a person is found and then are corroborated by research such as interviews with witnesses, belongings, and documentation or paperwork found that helps describe the decedent’s life. In the next example from field investigator working in the central part of the state for less than a year, we see how not having an address may indicate homelessness as well:

In one case in particular, a person had committed suicide, and they were homeless. . . . So they were at one point living in a shed in the grandmother’s backyard before taking their own life. So I’ll look—do they have a residence that is established, have they been there a while, do they hop, and in this case, I was able to confirm because everyone else told me she’s effectively homeless. She doesn’t live there, she doesn’t claim an address, she can’t go back there anyway, she left, she went over here, she lived in a shed, that’s you know, that’s homeless, to me.

As described, multiple signals come together to substantiate marking a person as homeless in a report. The field investigator notes that homelessness was confirmed through sources including the grandmother of the decedent, who also stated that the cause of homelessness was related to an abusive relationship with their partner. Here we see the relationship between experiences of homelessness and domestic violence as well. We also see a high level of transience and housing instability. The decedent did not have an address and stayed at multiple places temporarily. Another important preliminary indicator appears to be cause of death as well, particularly if cause is related to place and exposure to the elements, as depicted by a field investigator with more than 10 years of experience working the northwestern part of the state:

I’ve had two or three exposure deaths and of my exposure deaths, most of them we would regard as homeless, I’d say people that have a home but prefer to live on the streets, I’m saying homeless if they tend to live out of the home.

As described, cause of death was another early indicator. Exposure-related deaths signal homelessness because people are staying outdoors at the time of their death and appear to be unsheltered and homeless. In addition, the participant notes that people who appear to live outdoors are categorized as homeless even if they have other options for living spaces. In summary, multiple signs of homelessness help field investigators to decide whether to indicate homelessness in their reports. Signs include, place, appearance, word of mouth from social relations, or documentation from social records such as homelessness services or health records. However, research on the part of the investigator is also needed to corroborate homelessness, and there are also ambiguities and tensions within the decision to include homelessness specifically in the text of a report.

Ambiguity (and Tensions)

Although some forms of homelessness were clearer to field investigators, many also approached housing status with hesitancy or uncertainty. Language included semi-, partially, or basically homeless. Uncertainty was related to having an address as well—though the person had an ID with an address—or an address to collect mail—they may have been staying somewhere else. This particular circumstance appears to be related to age, as younger people are more likely to have identification with parents’ or caregivers’ addresses, even if this is not where they are currently staying. A field investigator with 5 to 10 years of experience working in the central part of the state described why the presence of addresses can conceal living status.

Cases that you’re talking about like couch surfing, it’s usually younger people, I don’t think that they are listed anywhere as homeless. This is because often times they will keep their parents’ address as their residence and I think that something that might be difficult for you guys is that whenever we put a case in . . . part of our responsibility for the death certificate is a home address. If there is no home address it can delay the death certificate for the Bureau of Vital Statistics and so they push very hard that you put something in there, and that might skew what you guys are looking at. We might have put an address on a driver’s license, or a parent’s address or an aunt’s address or something where they’ve ever gotten mail or stayed for any period of time, but really . . . they are not, that’s not their residence. Like you said they are being turned away from that place, it’s not their home, it’s not where they are living.

As described, the presence of an identification and/or an address blurs the lines of housing status. Younger people may be experiencing homelessness but still have their parents’ addresses attached to them. However, pressures to enter an address further obscure housing status. In some cases, a person may have an address only for their mail. The presence of an address does not rule out homelessness or housing instability, but it does change the focus of what is relevant information. A person’s address is not necessarily where they are living. This can be extended even to addresses that are for hotels, motels, or places of business that are not true residences. As shown below, place, history of homelessness, and time can interfere with decisions about describing housing status in a report. Another field investigator from the southeastern part of the state reported,

I did have another lady. She was living in a motel, but she was staying somewhere, not a house, I wouldn’t consider her homeless, but I think she did at one point temporarily live on the streets. But at this particular time, she had been living in a motel and she had been there a few days.

In the preceding case, we see how a history of homelessness or housing instability can factor into a report. In this instance, the person was not identified as homeless in the report despite having a history of homelessness and living in a motel for only a few days. However, recent homelessness is different than current homelessness and creates another area of uncertainty in the documentation process. Time is an important variable for current and former histories of homelessness, in addition to how recent the experience of homelessness was. However, these details were not ultimately included in report. During interviews, cases that were not described as homeless were sometimes reflected on differently as more likely homeless than not. For example, a field investigator working in the central part of the state for several years stated,

I tracked down the equivalent of the stepmother, she never adopted [the decedent] so she wasn’t the next of kin, but it was at that address where she would occasionally get mail even though she didn’t live there, so I would call that temporarily homeless. I don’t think I described it in the case in that way, I think I included the almost stepmother’s address, but I did mention that she comes from one place to another so she had a history to the police in that way. I think that qualifies as homeless. She had a roof over her head most of the time, but it wasn’t her home.

Several key signals are once again corroborated above, including word of mouth, length of time, and in between categories of homelessness. In this particular case, the person was not described as homeless or temporarily homeless at the time but within our discussion was reconsidered as such. An address was present and mail was received at that address, but the decedent had not stayed there. Rather, she was described as having gone from place to place, usually having a roof over her head, but not a home. In another case, a person stayed at a hotel and was likely homeless though different definitions of homelessness are described as a barrier to documentation. Another field investigator working in the central part of the state for several years reported,

The central office deputies are the ones that generate cases originally, and the field deputy reports which we review, are the ones that we go over, but depending on whoever is on the case their definition of what’s homeless changes. Like, for example, I did a death the other day, where the guy had been in a hotel for three weeks because he didn’t have a home, like I know that some of my colleagues might mark that as not homeless because they had a place to stay and put that motel as their residence and just describe it as that.

Accordingly, the investigator acknowledges that documentation is not always consistent across investigations when there is an address or residence (the motel) that could be entered. Furthermore, working definitions of homelessness vary across investigators. It is important to note that not only do definitions vary, but the focus on and perceived relevance of housing status as an important factor to document varies as well. Interpretations of housing access and housing status from witness and family members vary as well. A newer field investigator from the central part of the state with less than a year of experience described the following example:

One person, she had a hardship. She was living in the basement of her brother’s home, but she got in her car, her car was her office, everything you could possibly imagine was in that car. She would go up to the hot springs and meditate. She primarily lived in her car. She was coming back from the hot springs one day and lost control of her car and hit a tree and destroyed the car. She wasn’t hurt when the ambulance crew arrived on scene, but she stuck a .45 caliber pistol under her chin. And when the state police arrived, as soon as he got out of his car to talk to the EMS crew, he turned to her and she shot herself in the head. But talking to her brother . . . she had her mobile home so to speak, the car, but she had a home in the basement with her brother in the basement. He said she very seldom stayed there, she mostly lived in her car.

In this scene, the decedent is described as having a place to stay in her brother’s basement but that she primarily lived in her car. The tension here is between what others, specifically family, might have considered the home versus what the decedent considered home and whether the decedent was homeless at the time of their death. Decisions to mark someone as homeless or to comment on one’s housing status are subjective, and there is not a category for indicating this status, so it is inconsistently noted even when indicators are present. For example a field investigator from the southwestern part of the state with several years of experience shared the following:

Here’s my boggle with this. Personally, I don’t like the term [homeless] so I don’t unless it’s category I need to mark, then I don’t. I’ve had decedents in homes with no glass, and no plumbing and no electricity, but it’s a structure, are they under-homed? . . . I don’t know what to call them.

As shown, the question of what is a home and individual judgment about that guides whether to mark a person as homeless. Also, the attention paid toward housing status and relevance of it varies. The distinction is based on varying judgement among field investigators and their investigative reports, in addition to varying definitions of home (and homeless) as well as ambiguities and tensions that arise when observing and documenting homelessness and housing instability.

In summary, field investigators observe place and appearance, in combination with additional information obtained from law enforcement, family, friends, and/or witnesses at death scenes, and additional records. Alongside these factors, investigators may begin to consider homelessness in relation to certain causes of death such as exposure. And yet, conflict—labeling ambivalence—about a person’s housing status and whether they were homeless is present. For example, a person may have an address, and have a place to receive mail, but they may not be staying there. In addition to not having a structured field for housing status, having an address can make clarification on housing status less pressing—the relevance of housing status decreases. In addition, family members may make it known that they had opened their doors and that their family member could have stayed with them. A history of homelessness and the time affected by homelessness may also affect perceived relevance of housing status. These considerations influence decisions on whether to denote homelessness in a field report.

Furthermore, there are tensions on whether homelessness is important to note—whether it involves making a judgement—perhaps making inferences beyond what is directly in front of them and therefore not being objective about the decedent’s life, living conditions, and contributing causes of death.

Importantly, certain elements appear to be more consistent between signs of homelessness and potential ambiguities. These include having or not having access to electricity, heat, or running water in the place where one resides. The level of transience and perceived safety also appear to matter. In relation to these factors are levels of privacy and security. Such factors appear to bridge signs and ambiguities and also connect to adverse outcomes associated with experiencing homelessness, among cases observed to involve homelessness and those that were considered differently in hindsight.

Bringing findings together, Figure 1 presents a model for noting signs of homelessness in relationship to ambiguities. Accordingly, as greater signals of homelessness are present, labeling ambivalence decreases. Conversely, as greater ambiguities present themselves, labeling ambivalence increases. Tensions can theoretically be resolved by linking housing instability and homelessness to health outcomes: better connecting social determinants of health to causes of death through housing status in a process I describe as social mattering. This is conceptualized as a process that makes social context more relevant to health and well-being, moving it from background to foreground.

Figure 1 reconciles ambiguities present in documentation when housing status is not specifically required information alongside a rationale to resolve tensions and make the case for the impact that unstable housing has on health and early mortality. This can be seen through the relationship between living conditions present in both cases that were described as homeless and those in which ambivalence was present. Examples include lack of access to heat, electricity, and running water as well as high levels of transience and low levels of privacy and safety. Practically, ambivalence could also be reconciled with a structured field that would indicate a level of housing stability or instability or where housing status was a concern related to the decedent’s health.

Discussion

This study connects scholarship on classification, operationalization of homelessness, and the social construction of data. This study describes some cases of homelessness that are more obvious, such as living in tents and connection to homelessness services. Signals of homelessness tend to include place, appearance, word of mouth from social relations, or documentation from social records such as homelessness services or health records. However, many cases involve more consideration. There are varying definitions and types of both homelessness (Amore et al. 2011; Cronley 2010; Greene and Centrone 2020) and home (Duyvendak 2011) and this precipitates uncertainty.

Signs of homelessness are in relation to ambiguities that result in labeling ambivalence. Such ambivalence is weighted by perception of relevance of housing in records, concerns about improperly making judgement about decedents, and factors such as family interpretations of housing options, history of homelessness (and time), and the presence of a viable address. In particular, it appears that having a parent’s address and pressure to enter an address for vital statistics and death records may obscure data about youth homelessness in death records. This is a critical finding because mortality studies have rarely included young people such as runaway youth. This study suggests that such individuals, particularly runaway youth or emancipated youth, may have been unintentionally but systematically excluded from previous studies.

Importantly, ambivalence may theoretically be resolved by better linking housing and homelessness as a social determinant of health related through observation of the conditions present in cases of homelessness and their relevance to public and social health and—a process described as social mattering. Through social mattering, the social determinants of health are accentuated as factors related to premature mortality and the humanity of people experiencing homelessness is further recognized. Examples include connecting homelessness or housing instability with lack of access to electricity, heat, and running water as well as high levels of transience and low levels of privacy, safety, and security (Mueller and Tighe 2007; Rolfe et al. 2020). Making this clearer suggests the value of making social determinants important within data collection in health records more broadly and points to how housing status is an important data point. These connections may help change how and to what extent housing and homelessness is focused on and how housing is connected to health. Indeed, scholars have begun to urge sociologists to study housing as a human right and not just a commodity (Pattillo 2013).

From a practical stance, this study contextualizes how issues related to the presence of an address may influence research on homelessness, particularly related to youth who often have a parent’s address for documentation even if they do not stay with their parents. This issue may likely influence related research on homelessness using addresses in electronic health records (Vickery et al. 2018). This also extends to debate on including housing or homelessness status within social records more systematically (Biederman et al. 2019). This study suggests that a structured marker for housing stability or instability, which does not currently exist, would be beneficial. Incorporating this field would validate the importance of collecting these data and make systematic collection of these data more likely. Furthermore, focusing on a continuum of housing stability or instability would be relevant to all cases and further the process of social mattering. Making the physical conditions more relevant may reduce both the uncertainty of whether documenting housing stability is relevant and stigma associated with discussing housing and homelessness history as part of death investigations and records.

Many have studied social determinants of health (Patra et al. 2021) and homelessness within health and social records (Bejan et al. 2018) but less—heretofore—has been known about how these notes are constructed. Indeed, researchers have argued that recording homelessness in social and health records is valuable (Bejan et al. 2018; Biederman et al. 2019). Acknowledging how such records are socially constructed and biased is important. Numbers are often used to suggest objectivity, but the numbers are both influenced by human actors and influential to others (Stone 2020).

Limitations

Although this study makes important theoretical and practical contributions to the social construction of homelessness and connections between experiences of housing instability and public and social health, several limitations should be noted. First, this study is specific to cases within New Mexico and is not representative of other areas of the United States or beyond. Second, interviews likely occurred with self-selection bias. Those with interest in this research and homelessness in local communities were likely more inclined to participate than those with less interest. This may influence findings on how homeless stereotypes and stigma might also influence whether housing status is incorporated into investigative reports which was largely not found in this study. Nonetheless, participants in this study constitute 20 percent of field investigators in the state and geographical representation of the varied regions within it.

Future Research

Future researchers should examine how similar processes of labeling ambivalence influence the collection of housing stability in other records such as department of education data, criminal justice data, and medical social records. Additional work is needed to incorporate findings from this study on documenting death to related work documenting counts, particularly when using existing data. Accordingly, future researchers should examine existing questionnaires and develop and test refined measures of histories of homelessness and housing stability. Such work could lead to better data available for research on the impact of housing instability and homelessness on death (and health) at larger scales, from community reports to national assessments. This would also lead to better research on housing as a social determinant of health and a right and would increase the possibility of studying relationships between national homelessness counts and death in state level analyses. Future research might also further our understanding of the construction of social notes on housing and homelessness status in medical records through interviews with administrators and professionals in other settings (e.g., schools, law enforcement, medical centers) to better learn about constraints and opportunities for similar improvements to data within electronic health records (Cantor and Thorpe 2018; Wark et al. 2021).

Policy Implications

This study supports improving systematic data collection on a decedent’s history of homelessness within death investigations. Incorporating a structured field for current or previous homelessness can help reduce labeling ambivalence about housing status. This would benefit homeless mortality research and annual homelessness memorial vigils. Improved numbers on homelessness, both deaths and prevalence, can influence investments in (and research on) structural changes like investment in affordable housing, universal housing vouchers, and policies to address income inequality such as basic income (Kerman 2021; Padgett and Henwood 2018).

Conclusions

As scholars have written, homelessness and housing can be viewed as a barometer of social justice (Aldridge 2020): greater homelessness prevalence suggests worsening social problems of many kinds. Indeed, homelessness is related to poverty, racism, and social exclusion (Jones 2016; Shinn and Khadduri 2020). Better linking homelessness and housing instability with health outcomes through social mattering can reconcile ambiguities and reduce labeling ambivalence that obscures existing knowledge and data. Such work can pave the way for more consistent documentation and better advocacy for housing rights and policy to prevent and end the experience of homelessness. However, we should always be concerned with how and why numbers are produced.