Making BIPOC Lives Matter: A Qualitative Analysis of Managerial Resistance to Racial Exclusions in US Homeless Systems

ABSTRACT

The way racial disparities get re/produced or challenged by homeless systems is an underexplored subject in housing studies. This paper advances scholarship on that topic by examining how homeless system managers in the USA are responding to demands for colour-conscious vulnerability screenings. The VI-SPDAT is a popular tool for measuring the vulnerability of homeless service recipients and making resource allocation decisions. Recent studies suggest the VI-SPDAT is racially biased and thwarts Black Indigenous People of Colour (BIPOC) from getting rehoused. This paper presents interview-based research conducted with 35 participants from 28 homeless systems that answers questions about how managerial staff perceive the VI-SPDAT and steps they are (not) taking to address racial disparities associated with the tool. Our findings delineate various ways system managers indirectly support service recipients by resisting institutional practices that marginalize BIPOCs. This extends homeless scholarship by showing how race relations shape supportive interventions in the USA and the way system managers try to allay racial disparities.

KEYWORDS:

• Homelessness
• VI-SPDAT
• governance
• United States
• a.nti-racism

Introduction

Race scholars have recently critiqued the “datafication” of homelessness (Cronley 2020). Datafication is “the transformation of social life into online quantified data, thus allowing for real-time tracking and predictive analysis” (van Dijck 2014). Producing homeless data requires conceptual categories whose meaning is shaped by politics (Bevan 2021; Bullen 2015; Dyb 2021; Mostowska 2020). Datafication has increased as cities seek efficient solutions to social problems. A small body of research analyses the datafication of US homeless systems. Willse (2015) shows how the Department of Housing and Urban Development (HUD) requires grantees to produce a “data double” of service recipients. A data double is a quantitative signifier of an individual that is produced by institutional authorities for surveillance and resource allocation (Haggerty and Ericson 2000). Bureaucrats use data doubles to monitor homeless system performance and efficiently allocate housing resources (Community Solutions 2018). Recent studies suggest an assessment tool – The Vulnerability Index – Service Prioritization Decision Tool (VI-SPDAT) – that US homeless systems have commonly used to datafy clients and make programme referrals is racially biased (Cronley 2020; Kithulgoda et al., 2022; Wilkey et al. 2019).

We extend that research by analysing how managerial staff use those studies to evaluate/change the way their system datafies clients and allocates resources. It presents interview-based research conducted with 28 homeless systems that answers the following questions: How do homeless system managers view their assessment tool in light of recent racial critiques? How do these stakeholders understand the mechanisms of racial exclusion that have been associated with their assessment tool? What strategies are they using to address these racial disparities? The paper advances homeless scholarship in three ways. First, it builds on small case studies (see Fritsch et al. 2017; Slota et al. 2021) by showing managerial staff throughout the USA have accepted the aforementioned critiques and are looking for alternative ways to assess clients. Second, it analyses the way managerial staff understand how racism affects their homeless system. This is an under-studied component of homeless governance (DeVerteuil and Wilton 2009; Stuart 2016) and datafication (Andrew, Camron, and Lata 2021; Grainger 2022; Mostowska 2020) that we address. Third, we show anti-racist resistance is an unacknowledged form of indirect support that managerial staff give clients. This advances homeless scholarship by showing how resistance to racial exclusion has become integral to supportive interventions that managerial staff moderate (see Clarke and Parsell 2019; Dyb 2021; Lancione 2016; Mostowska 2020; Pauly et al. 2019).

Datafying Homelessness in the USA

The Homeless Emergency Assistance and Rapid Transition to Housing Act of 2009 (HEARTH) helped datafy US homeless systems (Berg 2013). A large portion of federal spending on homeless services is provided through the Continuum of Care block grant (US Department of Housing and Urban Development 2022a). In 2022, the Department of Housing and Urban Development (HUD) recognized 387 grantees known as “Continua of Care” (CoC). A CoC is a federally funded/regulated public-private partnership that delivers most homeless services in US communities and constitutes a significant portion of local homeless systems (US Department of Housing and Urban Development 2022b). HUD gives each CoC some discretion to make tailored policies that meet local needs. HEARTH established the CoC block grant and imposed datafication requirements (Leopold 2019). This requires CoCs to maintain a Homeless Management Information System (HMIS) where a data double of homeless service recipients is created/stored (Willse 2015).

Caseworkers produce a data double in HMIS whilst moving homeless service recipients through coordinated entry. Coordinated entry is a centralized system that allocates permanent housing to homeless individuals (US Department of Housing and Urban Development 2016). HUD requires caseworkers assess people at coordinated entry access points with a standardized tool that prioritizes clients for permanent housing (Federal 2015). The VI-SPDAT is a tool that service providers have often used to assess clients in the USA (Leopold and Helen 2015), Canada (OrgCode Consultancy, Inc 2016), and Australia (Andrew, Camron, and Lata 2021). It is a 34-item survey that asks about the respondents’ homeless history, household income, emergency service utilization, and mental/physical health (Grainger 2022). The tool connects risk assessment to resource allocation through a scoring system that ranges 0–17 (Shinn and Richard 2022a). Participants who score 0–3 are ineligible for any housing assistance (low vulnerability), 4–8 are eligible for temporary housing assistance (moderate vulnerability), and 9–17 are eligible for permanent (supportive) housing assistance (high vulnerability) (Petry et al. 2021). Roughly 65% of a VI-SPDAT score is derived from health-related questionnaire items (Grainger 2022). The VI-SPDAT thus largely measures morbidity. Community Solutions and OrgCode Consultancy Inc. created the VI-SPDAT in 2013 as a standardized way to triage homeless individuals (OrgCode Consultancy, Inc 2022). By spring 2020, OrgCode had updated the VI-SPDAT three times and created youth, singles, and family versions. Although OrgCode intended the VI-SPDAT to be supplemented with other information, many homeless systems used it as the sole vulnerability indicator (Leopold and Helen 2015).¹

Scholars have paid little attention to VI-SPDAT implementation. Previous studies say it lacks construct validity as well as test-retest and inter-rater reliability (Molly et al. 2018); produces biased scores that advantage White over Black Indigenous People of Colour (BIPOC) clients (Cronley 2020); and has raised concerns about accessibility, accuracy, and misclassification among frontline staff (Fritsch et al. 2017; Slota et al. 2021). Some scholars have theorized mechanisms that produce these outcomes. Molly et al. (2018) attributes poor reliability to deceitful or untrained caseworkers. Grainger (2022) shows how the location of VI-SPDAT screenings can underscore disabled Black men. Both Cronley (2020) and Kithulgoda et al. (2022) say the questionnaire design advantages White respondents. That said, some research indicates VI-SPDAT scores predict returns to homelessness if clients are not referred to programmes recommended by the tool (Petry et al. 2021) and some frontline workers see the tool as a useful way to allocate resources (Slota et al. 2021). Although these studies identify problems with the VI-SPDAT, they neglect how service providers view the tool in light of critical research and steps they are (not) taking to address these issues.

The Biopolitics of Homeless Datafication

We use Foucault’s (2004) biopolitics framework to do our analysis. Biopolitics manages life in modern society by manipulating fertility, morbidity, and mortality (Foucault 2004). Unlike feudal lords, who used sovereign power to kill political opponents and let allies live, nation-states rule via biopower that re/produces life for some and lets others die. Nation-states use biopolitical technologies – categories, databases, statistics, graphs, and charts – to “see” and control population dynamics (Scott 1998). This does not mean sovereign power has been renounced. Sovereign power must now be viewed as a legitimate violation of personal liberty. Racism is an ideological tool that nation-states use to exercise sovereign power. Foucault (2004, 254) defines racism as, “The break point between what must live and what must die.” According to him, racism divides a human population into “superior” and “inferior” segments, not necessarily according to phenotype. The superior race is made to live through biopolitical interventions whilst inferior ones are made to die through sovereign power (e.g. executions) or allowed to perish by getting excluded from biopolitical interventions. In other words, the nation-state “kills” those it deems inferior through assertive and passive means. Biopolitics is thus the nation-states’ use of racism to “defend” itself by culling “inferior” lives and nurturing “superior” ones.

Whilst Foucault (2004) theorizes the function of racism, sociology provides a conceptual framework to analyse how it impacts morbidity and mortality in American society. We define race as a categorical system that signifies phenotypic, ethnic, and/or geographic difference (Omi and Winant 2015). A shortcoming of Foucault’s (2004) theory of racism is it only accounts for intentional discrimination. Contemporary sociology offers a more nuanced interpretation that has implications for how we think about biopolitics. Murji (2007) differentiates individual from institutional racism. Individual racism is interpersonal discrimination that results from racial prejudice. Institutional racism is discrimination rooted in organizational rules and practices. Both kinds of racism can be intentional or unintentional. Intentional racism consciously excludes a racial group. Unintentional racism excludes people through unconscious discriminatory acts. We use these insights to extend Foucault’s (2004) biopolitical framework to incorporate intentional and unintentional forms of racism.

Racial projects re/produce race classifications that motivate and legitimize different kinds of racism (Omi and Winant 2015). Racial projects can harm or help BIPOCs. In the USA, a racist project stigmatizes BIPOCs as inferior to Whites and upholds institutions that harm BIPOCs. An anti-racist project helps BIPOCs by challenging racial stigma and discrimination. The USA was established as a racial state for White Europeans (Goldberg 2002). Early racist projects like colonialism used natural rights philosophy to bolster slavery (Levesque and Baumgarten 2008). This cultural framework was inherited by American descendents who use it to resist anti-racism. Although anti-racists challenge racist projects, White backlashes have weakened their achievements. Just prior to this study, anti-racists contested “colourblind” public policies (Fisher and Rouse 2022). Colourblindness is a racist project that bolsters White dominance by denying institutional racism exists and defending policies/practices that perpetuate racial inequality (Bonilla-Silva 2003). Because they view institutional racism as anachronistic, advocates of colourblindness attribute durable inequities to BIPOC “deficiencies.” By denying institutional racism, well-intentioned people, who view colourblindness as a way to promote racial integration, can facilitate institutional racism. Colourblindness erodes a shared understanding of institutional racism by reducing it to overt interpersonal discrimination. This makes it hard for people to identify and challenge institutional racism (Haney-López 2014). Anti-racist projects challenge colourblindness with “colour-consciousness” (Bonilla-Silva 2003). Colour-consciousness assumes institutional racism is endemic to American society and must be purposefully challenged. Housing agencies in some federal (US Department of Housing and Urban Development 2022c), state (WHEDA 2020), and local governments (Community Development Alliance 2021) have started using colour-consciousness to make policies. Racial equity has thus become a performance measure that some governments are using to evaluate policies.

Applying colour-consciousness to social policy can be an act of biopolitical resistance. According to Foucault (2004), social policies that increase morbidity and/or mortality for one group whilst reducing it for another are biopolitical interventions. Biopolitical resistance challenges in/formal policies and practices that systematically increase morbidity or mortality for a minority group through intentional or unintentional means (see Selmeczi 2009). If BIPOCs are unintentionally marginalized by a policy in this way, then it is a colourblind form of institutional racism. Anti-racists resist biopolitical exclusion in several ways. During the COVID-19 pandemic, academics advocated BIPOCs get prioritized for COVID-19 vaccines to reduce health disparities (Schmidt, Gostin, and Williams 2020). This motivated Minnesotan bureaucrats to prioritize COVID-19 vaccines for zip codes with large BIPOC populations (Minnesota Department of Health 2022). This illustrates two salient dimensions of biopolitical resistance. First, it is a collective process that involves cross-sector partnerships. Second, antiracist projects can function as biopolitical resistance when they challenge policies that un/intentionally expose BIPOCs to morbidity or morality risks. Because the VI-SPDAT primarily measures morbidity, we use biopolitical resistance to conceptualize anti-racist mobilization against it.

The VI-SPDAT as a Biopolitical Technology

We define the VI-SPDAT as a biopolitical technology that is used to ration permanent housing assistance. Willse (2015) shows how the US Government differentiates chronic from nonchronic homelessness. This decision was informed by research that shows chronically homeless individuals consume 50% of emergency services despite constituting 10% of homeless populations (ibid, 2015). Policymakers use these categories to prioritize chronically homeless individuals for permanent housing assistance. This manages exposure to morbidity and mortality risks according to the burden somebody’s homelessness puts on tax levies. The VI-SPDAT is similarly designed to enhance systemic efficiency by dividing homeless populations into subcategories for allocation purposes. Since homelessness exposes people morbidity and mortality risks (National Healthcare for the Homeless Council 2022), the VI-SPDAT allows some people to escape harm by helping them get permanent housing and perpetuates the exposure of others to danger by leaving them on the street. Because many US homeless systems use the VI-SPDAT (Leopold and Helen 2015), it is a crucial mechanism that determines whose morbidity and mortality risks are mitigated by coordinated entry.

Recent studies critique the VI-SPDAT. Fritsch et al. (2017) show service providers in Minnesota worried the wording of questionnaire items discourages honest responses and emphasized the importance of client-assessor rapport. Slota et al. (2021) extends that study with a qualitative analysis of caseworkers in Texas. Although participants reported racial disparities, they still believed the VI-SPDAT effectively allocated resources. Quantitative studies bolster concerns raised by qualitative research. Wilkey et al. (2019) presents a multivariate analysis of VI-SPDAT scoring across 3 CoCs. It found White singles got higher scores and were more likely to be prioritized for permanent housing. Cronley (2020) presents an intersectional analysis of VI-SPDAT scores in one CoC. She shows Black men score lower than White wo/men (excluding history of homelessness) despite being overrepresented in the US homeless population and White women scored higher in vulnerability (excluding history of homelessness) than Black women even though past research indicates no racial differences in risk exposure amongst homeless women. These studies suggest the VI-SPDAT helps Whites get rehoused faster than BIPOCs and perpetuates BIPOC exposure to morbidity and/or mortality risks (Minnesota Department of Health 2023; National Healthcare for the Homeless Council 2022).

OrgCode has publicly responded to these concerns. It says the VI-SPDAT was designed as a triage rather than assessment tool to facilitate need-based allocations and combat “creaming” by caseworkers who avoid hard-to-serve clients (OrgCode Consultancy, Inc 2022). As a triage tool, OrgCode says the VI-SPDAT was never intended to be the only factor caseworkers consider whilst making allocation decisions. However, this is exactly how many CoCs used it (Leopold and Helen 2015). OrgCode has therefore retorted that recent criticisms misrepresent the tool’s purpose and ignore misuse by providers. Whilst misuse of VI-SPDAT scores is relevant, it does not address concerns about survey items that advantage White people (Cronley 2020; Kithulgoda, Vaithianathan, and Parsell 2022; Wilkey et al. 2019). Given OrgCode’s mission to promote racial justice (OrgCode 2022), we define racial disparities associated with the VI-SPDAT as unintentional colourblind racism.

Supportive Governance as Biopolitical Resistance

The VI-SPDAT’s role in delivering homeless services makes it relevant to scholarship on supportive governance. Supportive governance is the collaborative management of entries into and exits from homeless services by public, private, and non-profit organizations. Research on supportive governance shows frontline workers deliver voluntary care in “spaces of care” by providing essential resources so homeless individuals can survive (DeVerteuil and Wilton 2009). Frontline workers also practise involuntary care by compelling homeless service utilization (see Stuart 2016). Service providers thus exercise supportive governance in various ways. Resource scarcity requires service providers to ration assistance (see Lipsky 2010). Such rationing is a biopolitical intervention when caseworkers influence the exposure of clients to morbidity and mortality risks.

The formal criteria used to allocate essential resources is usually not determined by caseworkers. Although caseworkers exercise discretion (Lipsky 2010), policymakers impose constraints that limit who can get what, when, how, and why. Dependency of non-profit agencies on federal grants has led some scholars to deem them a “shadow state.” Wolch (1990, 201) argues, “The voluntary sector has in effect become a shadow state: that is, a para-state apparatus with collective service responsibilities previously shouldered by the public sector, administered outside traditional democratic politics, but yet controlled in both formal and informal ways by the state.” Wolch warns changing voluntary organizations into a government surrogate risks neutralizing a locus of social change. Whilst some research supports Wolch’s claim (Mitchell 2001), others challenge it by illustrating ways grantees maintain semi-autonomy (Trudeau 2008). With respect to homeless policy, federal authorities devolve limited flexibility to grantees (Willse 2015). Because a large portion of US homeless services are funded by the federal government’s Continuum of Care (CoC) block grant (US Department of Housing and Urban Development 2022a), we argue CoC grantees are shadow state organizations. Dependency on federal grants bounds these agencies to the political interests and values of the US Government’s executive branch because each President appoints a Secretary of HUD. The President’s ideology are filtered into HUD’s regulations through political appointees (Hays 1995). In the neoliberal era, this has resulted in rationing criteria that minimizes public expenditures (Willse 2015).

The US shadow state is part of a political system that un/intentionally advantages White people. Nonprofit organizations can play a critical role in supporting or challenging racist projects (see Ray 2019). Reliance on government funding directs service delivery in accordance with the racial project of lawmakers and bureaucrats. Previous studies indicate some of organizations reinforce White dominance through service delivery (Trudeau 2012); however, this is not always the case (see DeVerteuil, Power, and Trudeau 2020). At the local level, caseworkers resist unfavourable policies by meeting needs they know exist, but policymakers ignore (Lipsky 2010). Recent critiques of the VI-SPDAT are biopolitical resistance to colourblind homeless policy by (non-)academic researchers (Cronley 2020; Wilkey et al. 2019). The authors of those studies promote colour-conscious allocation schemes that account for racial disparities. It is currently unknown how local bureaucrats perceive/apply those findings.

We advance homeless scholarship with interview-based research that examines how system managers in the USA resist biopolitical exclusions created by the VI-SPDAT. This paper provides evidence that managerial staff use colour-consciousness to support clients by challenging biopolitical interventions that deepen racial disparities. Building on Foucault’s (1978, 95) insight, “Where there is power, there is resistance,” we analyse resistance to biopolitical interventions like the VI-SPDAT that racially bias access to permanent housing. The few studies on race relations in homeless systems indicate caseworkers facilitate and discourage interpersonal racism (Lancione 2016), reintegrate BIPOCs into their communities (Pauly et al. 2019), and use CCTV to connect BIPOCs with supportive services (Clarke and Parsell 2019). Research on homeless datafication has either neglected the way colourblind forms of institutional racism affect resource allocation (Batterham 2019; Dyb 2021; Mostowska 2020) or the strategies managerial staff use to challenge institutional racism (Fritsch et al. 2017; Slota et al. 2021). We advance that research by analysing anti-racist strategies that managerial staff are using to challenge racial exclusions from the VI-SPDAT. Biopolitical resistance in this context is a supportive form of homeless governance that resists institutional racism by reforming policies and practices to enhance BIPOC access to permanent housing through coordinated entry.

Methods

This paper is from research on Built for Zero implementation. Built for Zero (BFZ) is a method for allocating permanent housing assistance through coordinated entry with real-time data (Community Solutions 2018). At the time of this study, Community Solutions, the organization that created/promotes BFZ, had recruited 98 out of 387 CoCs into its campaign. BFZ communities participate in workshops and access web documents where advisors give implementation guidance. They also participate in meetings where they discuss problems that their homeless system is confronting and solutions to fix these issues. Community Solutions is therefore a central node that connects and diffuses information to homeless systems.

The University of Wisconsin-Madison’s IRB Office gave us permission to conduct interviews with human subjects. We sent a recruitment email to coordinated entry directors from 74 CoCs in the BFZ campaign.² A coordinated entry director manages the way allocations are made through their system. Directors use HMIS data to determine eligibility for permanent housing. The VI-SPDAT score is an indicator that coordinated entry directors consider when prioritizing clients. Several contacts immediately responded to our request. We sent follow-up emails to other contacts who did not reply to our initial recruitment inquiry. This recruited 28 CoCs into our sample. Some respondents invited colleagues from their CoC like HMIS Analysts who they thought could speak on the topic. Thirty-five people in total contributed to this study. Table 1 lists the demographic characteristics of participants.³ We recruited participants from a diversity of regions and places (urban and rural) for sample variation. This was important because CoC’s serve communities with different racial dynamics. Different experiences of race could vary perspectives of the VI-SPDAT.

Table 1. Demographic Characteristics of Study Participants.

		Frequency	Percent
Race	Black	3	8.57
	Hispanic	2	5.71
	Native	2	5.71
	White	26	74.3
	N/A	2	5.71
Gender	Female	24	68.5
	Male	9	25.7
	N/A	2	5.71
Age	(18–35)	13	37.1
	(36–55)	15	42.9
	(56–99)	1	2.90
	N/A	6	17.1
Education	High School	2	5.70
	Associates	1	2.90
	Bachelorette	12	34.3
	Post-graduate	15	42.9
	N/A	5	14.2
Years of experience	(0–5)	4	11.4
	(6–10)	14	4.0
	(11–15)	3	8.6
	(16–20)	6	17.1
	N/A	5	14.3
Years in role	(0–5	21	6.0
	(6–10)	6	17.1
	(11–20)	2	5.71
	N/A	6	17.1
Total		35	100%

Interviews were conducted on Zoom in spring 2022 with an audio recording device. We used a semi-structured interview guide that included the following questions: How do people access coordinated entry in your community? What tool does your CoC use to do assessments? How well do you think that tool assesses clients? How does your CoC make programme referrals? Questions were broadly framed to not bias responses. Only a few participants did not immediately discuss racial equity when we asked these questions. In those cases, we probed on whether they had heard about racial critiques of the VI-SPDAT and if they thought those criticisms applied to their homeless system. We included data from all interviews to demonstrate various orientations expressed by respondents in the data analysis. Interviews lasted 30–90 minutes, but most were approximately one-hour. We sent recordings to a private transcription service. Returned transcripts were anonymized and then analysed using coding strategies from abductive methodology (see Timmermans and Tavory 2022). We first open coded the transcripts independently to identify the range of topics that we could engage with our data. After selecting the biopolitical framework, we jointly used focussed coding to conceptualize the practices that constitute the major sections of our data analysis. We grouped the open codes into superordinate themes. This allowed us to delineate variation and identify emergent findings.

Findings

Our data analysis proceeds by first analysing how participants problematized the VI-SPDAT. Next, we examine how respondents understood the way VI-SPDAT screenings in/equitably prioritized homeless individuals. Lastly, we analyse strategies participants said they were using to resist the biopolitical exclusion of BIPOCs.

Problematisation

Many participants said the VI-SPDAT had lost legitimacy in their homeless system. Although the VI-SPDAT was promoted a few years ago, it had since become viewed as problematic and needed to be replaced: “The VI-SPDAT … Going back to 2015, that was the tool to use … It’s not perfect and we know the fault … Right now, it’s fading out … ” [Sophia]. Pressure to replace the VI-SPDAT came from key stakeholders. When asked why her CoC decided to use the VI-SPDAT, Rachel answered, “HUD chose it … That’s changed now … HUD is saying that it’s an imperfect tool and that CoCs should be looking at certainly adjusting their assessments … ” Another participant described ways OrgCode is now discouraging VI-SPDAT usage, “I think it’s about two years old now, from OrgCode … Acknowledged at the rollout of the version three of the VI-SPDAT, that there were some equity concerns … And just recommending that the interviews start to kind of move away from the use of the SPDAT tool” [Rene]. Most participants understood this advice to reflect tool deficiencies, “I don’t think it was properly assessing people … It’s a cookie cutter model … Every community is different … It wasn’t meeting the needs of every community” [Sarah]. Local homeless systems are in/directly connected to one another. Government agencies and nongovernmental organizations (NGOs) disseminate implementation advice to local homeless systems. Participants said these organizations diffused critical research on the VI-SPDAT. This motivated some respondents to problematize the VI-SPDAT.

Participants varied in how they problematized the VI-SPDAT. Some criticized “traumatizing” questions, “The questions can be pretty traumatic … A lot of them aren’t even necessary … ” [Lauren]. Whilst others criticized it for not matching clients to appropriate housing programmes, “When people get referred to our [veteran’s] programme, just because they had a three on their VI-SPDAT doesn’t mean that clinically, they need case management … It’s a good starting point, but definitely not a great indicator of appropriate fit” [Cassie]. Most critiques pertained to race, “There’s been a lot in terms of racial disparities … And we’ve seen it in our county … People who identify as Black are scoring about one point lower than people who are identifying as White … ” [Marie]. Jason added, “I think if you’re assessing someone’s vulnerability, you have to take into account all of the factors in their life … The inherent structure of racism in our country puts populations that are at a disadvantage when it comes to accessing resources.” Several participants referenced academic research to support their claims, “The HMIS Lead spearheaded this … She had been reading the research that the VI-SPDAT tended to prioritize White men over others” [Heather]. Others cited research conducted by their CoC, “I do not think that it [the VI-SPDAT] is effective … We just had our most recent coordinated entry evaluation … We can see racial differences in the scores” [Anne]. In some cases, higher VI-SPDAT scores were said to facilitate inequitable housing outcomes, “We were looking at the impacts of the VI-SPDAT tool and how people were being referred … [We] saw that people of colour were not scoring as high overall and were not being prioritized … ” [Debbie]. Participants often used colour-consciousness to interrogate the VI-SPDAT by citing racial biases that deprioritised BIPOCs. This does not mean every participant provided a detailed account of racial bias. Some cited hearsay whilst others mentioned analyses conducted by the CoC.

A few respondents defended the VI-SPDAT. They thought it was an effective triage tool, “I’ve heard all the scuttlebutt around it, but it’s a tool. It’s not the be all and end all … I think that it gives you an excellent place to start … I stand by the VI … We’re going to go down in flames with it” [Tonya]. Some cited research conducted by their CoC that contradicted recent studies:

We do an annual data analysis of homelessness by race … Folk of colour are more highly represented in our homeless population, but we serve them at higher rates in our housing programmes … We haven’t noticed any inequities in how we’re serving folk using VI-SPDAT. [Beau]

Whereas Tonya and Beau supported the VI-SPDAT, other participants qualified their endorsement of the tool, “I heard that there were challenges to it … But when it comes to assessing people’s vulnerability there’s always going to be flaws in whatever you do because that’s a hard task in itself” [Ginger]. Whilst others said the racial dynamics of their community rendered critiques irrelevant:

OrgCode is no longer updating [the VI-SPDAT] because of all of the complaints … What we experience is totally different … The majority of our population is Hispanic … What we experience as far as racial inequalities is a family being denied to go into a shelter because they don’t speak Spanish … [Sophia]

This illustrates variation in how participants evaluated the VI-SPDAT. Despite recommendations from national organizations, some who disagreed with critiques continued using the VI-SPDAT. This does not mean defenders advanced a colourblind project. Beau did not observe racial biases when he tested the hypothesis against his CoC’s data. Others like Ginger normalized racial disparities as inevitable. Passive acceptance of racial disparities advances a racist project if the VI-SPDAT facilitated racial exclusion(s).

Mechanisms

Participants identified mechanisms that created racial disparities. Some had heard criticisms but could not specify how the VI-SPDAT excluded BIPOCs in their community, “There has been some talk that it could be that it’s not racially equitable … I don’t see that in my population, but then we’re predominantly largely a White community … I’ve never really gotten a specific, ‘This question is the question that makes it not equitable … ’” [Rachel]. This led some to conclude the tool unintentionally excluded BIPOCs, “We looked at our data, and were like, ‘This is not doing what it was intended to be doing.’ Not out of malice … It’s just not doing what it’s intended to be doing” [Thomas]. A consensus emerged that the VI-SPDAT is racially biased and needed replacement. This critique did not resonate with participants like Rachel who worked in a predominately White rural community where racial disparities were less apparent. In contrast, Thomas worked in a community that neighboured a large city with a racially diverse population. Critiques of the VI-SPDAT were possibly more salient to him because racial disparities were observable. Rachel’s CoC nevertheless joined others in retiring the VI-SPDAT.

Most participants identified problematic questionnaire items. A key flaw in the tool, from their perspective, is it was not developed with BIPOCs, “The tool itself was not developed in collaboration with people of colour … So, I think there was some inherent bias … ” [Debbie]. Some thought excluding BIPOCs in this way facilitated White prioritization. Several respondents challenged questions about healthcare utilization for not reflecting BIPOC experiences:

There were some questions like medical access questions that you’re scoring higher … Some people don’t have the same access to medical services … If people that don’t insurance or just had Medicaid, they could only access certain doctors … That was a big thing that stuck out to me. [Yolanda]

Participants were sensitive to negative experiences BIPOCs had with doctors. Some referenced this distrust whilst criticizing health-related items, “A notable category in the VI-SPDAT is how many times one seeks medical care … It doesn’t fully capture the folks that are distrusting of the system … And we find a large proportion of those folks are likely to be people of colour” [Ted]. Others attributed misreported medical history to perceived norms in BIPOC communities:

The vast differences when you ask a Black head-of-household about their health and wellness compared to a White head-of-household is astronomical … It falls into that whole what is the norm for sharing that information. If you’re a Black man that’s head-of-household, I’ll not be answering that because that is a weakness. [Adriana]

This demonstrates the use of colour-consciousness to critique the VI-SPDAT. Respondents used their knowledge of racial history to identify components of the VI-SPDAT that could deprioritise BIPOCs for permanent housing. This does not mean their analysis was always correct. Most respondents were White and may have lacked knowledge of BIPOC communities. That does not negate our observation that they were using colour-consciousness to resist biopolitical exclusion.

Participants also raised concerns about items measuring exploitation. The VI-SPDAT probes whether a respondent has been taken advantage of on the street. Some thought asking these questions before caseworkers developed rapport invited racial bias:

There is inherent bias in the tool … It specifically limits the options for Black women … Questions are asking, “Have you ever exchanged sex for money?” The VI-SPDAT is meant to be assessed [within] the first week … They don’t really know the case managers … They are going to probably answer “no” for this question … Because they’re just not comfortable … [They’re] not going to score as high. [Lisa]

Lisa used colour-consciousness to criticize VI-SPDAT questions that BIPOC women might be less willing to answer honestly. The VI-SPDAT is often administered by White assessors who ask questions about sexual exploitation, physical violence, drug use, self-harm, and legal troubles. Most people would probably hesitate to answer these questions whilst getting assessed by a stranger, especially if they are a BIPOC who views the welfare system as a racist project. Colour-consciousness therefore gave Lisa a tool to identify a mechanism of racial exclusion.

Respondents also criticized the VI-SPDAT for giving BIPOCs higher scores than Whites. Recall that the VI-SPDAT separates clients into three categories that inform a housing referral. To follow this guidance, homeless systems must have enough programme slots to accommodate demand. One respondent said this was never the case in his CoC:

A lot of [rapid rehousing] programmes [don’t] want number 12 and up … They want between six and ten … That’s normally a person who hasn’t been homeless for a very long time … It sounds like the higher number is going to get the housing, but the higher number normally was not getting the housing … [William]

This would have been inconsequential if William’s CoC had enough permanent supportive housing. When we asked, “Were there racial disparities in who was getting scored higher,” William answered, “Yes, just because some of those clients are from villages … Any actual village, it’s going to be an [indigenous] village … If you’re coming down here, and you’re trying to get help, you’re already going to be pretty high on that score.” This shows how the proper use of the VI-SPDAT can unintentionally generate racial disparities. If a CoC offers more rapid than permanent housing, then people who score higher than seven may get deprioritised by providers who refuse highly vulnerable clients. Because BIPOCs usually scored above seven on the VI-SPDAT and local service providers specialized in rapid rehousing, the tool re/produced racial disparities in homelessness by keeping BIPOCs on the street in William’s community. This illustrates why questions need to be uniquely tailored to each CoC.

A minority of respondents raised concerns about assessors. Who can conduct an assessment varies by access point model. Some CoCs use a single-site model that requires homeless persons travel to a centralized location for assessment. Other CoCs use a No-Wrong-Door model that allows hotline staff, emergency shelter providers, street outreach teams, etc. to do assessments. The latter model includes more assessors with less oversight. Participants believed inter-assessor variability could account for racial bias, “It’s also very dependent on the assessor and how well we train everybody … Folks might be leading with their questions or delve more into the questions than they should” [Ted]. Respondents varied in interpretation of assessor bias. One said caseworkers manipulated VI-SPDAT scores to help clients, “ … One of the negative pieces about having a decentralized coordinated entry system is that you have multiple parties doing the same assessment … We have seen some agencies where we’ve had to remove them … because they have been skewing scores … ” [Rene]. Another respondent thought distrust discouraged BIPOCs from giving honest answers to White assessors, “People maybe don’t trust the person that they’re giving their information to … They don’t trust the person that is the asking the questions because they don’t trust institutions … If the assessor is not a person that looks like them, then maybe there is concern … ” [Marie]. Some participants thus echoed OrgCode by attributing racial disparities to assessor bias (OrgCode 2022). Respondents like Rene said dishonest caseworkers inaccurately scored clients. Other participants like Marie used colour-consciousness to racially contextualize assessments. Marie’s point was previously echoed by Lisa and Adriana when they critiqued items on the VI-SPDAT.

Resistance

Participants described ways they resisted biopolitical exclusion. Biopolitical resistance meant finding alternative measure(s) to assess and prioritize clients for permanent housing. This meant using colour-consciousness to challenge colourblind racism. Although many respondents expressed need for a new vulnerability measure(s), several said their CoC had just started to look for an alternative, “I just assigned homework for the rest of our team, about watching video, where the Austin CoC has created their own vulnerability index tool … We’re kind of thinking, ‘Is that a possibility here?’ Questions like that are only just starting now … ” [Dustin]. Although most participants said their CoC wanted a replacement, some thought this would be an arduous task that they were not ready to start, “I know [the VI-SPDAT has] come across some criticism … I know there’s a number of communities that are wanting to move away from it … We would like to eventually, but it’s so integral to our system … It’s a daunting prospect to move away from that” [Ginger]. Biopolitical resistance started with attributing racial exclusions to the VI-SPDAT. Many CoCs who viewed the VI-SPDAT as institutionally racist sought a colour-conscious replacement. As we showed, this was not always the case. Some participants whose CoC evaluated their system, found the VI-SPDAT was race neutral, and therefore planned to keep using it. Other CoCs sought an alternative based on hearsay even though they had not observed racial disparities in their system. And some had no plans to test the tool or look for a replacement because they disregarded critiques. Biopolitical resistance therefore was not ubiquitous nor did every CoC have a locally informed reason to find a replacement.

Several respondents said their CoC had created a workgroup to design an original assessment tool. William previously demonstrated a localized assessment was needed by some CoCs to account for their unique racial dynamics. This did not always mean complete rejection of the VI-SPDAT, “We have a workgroup that’s currently taking a look at that and seeing what other assessment questions we might add in addition to the SPDAT … A lot of that is in the testing and development right now … ” [Rachel]. Creating a new assessment was a protracted process, “We’re actually undergoing a process of redesigning our coordinated entry system and one of the top things is actually replacing the VI-SPDAT, but that is a long process, probably take another year” [Ted]. Some participants identified issues that needed to be addressed whilst making a new assessment tool:

We also have seen some racial disparities on our priority list, but I can’t say for certain that it’s because of the assessments. Is it the person doing the assessment? Is it because we have a really high incarceration rate in [my county] and so people’s chronic homelessness gets messed up because they’re in-and-out of jail? [Tara]

Biopolitical resistance is a collective process that extends over time and includes multiple stakeholders in or adjacent to the homeless system. Resistance does not start from scratch. Participants used available resources to create a colour-conscious assessment. That meant respondents identified ways the VI-SPDAT deprioritised BIPOCs to ensure their replacement did not perpetuate institutional racism. Participants used colour-consciousness to think of ways institutional racism within and outside the homeless system impacted entries to access points and exits into permanent housing. The results of this analysis informed item construction and protocols for administering the new tool at access points.

Workgroups used various strategies to create a new assessment tool. Many participants were not actively involved in questionnaire design and thus provided general reflections of what their CoC’s workgroup was doing. One participant was involved in this process and detailed strategies their CoC used. Her workgroup hired a consultant who sampled tools that other CoCs experimented with, “We hired a local consultant who knows the system very well, and we said could you please conduct a landscape assessment of all the CoCs in the country and see what they’re doing about the SPDAT retiring … So, she did” [Adriana]. The workgroup adapted these tools to their homeless system:

She found some tools that they were utilising … They really didn’t work for us … They have ripped apart the VI-SPDAT, torn apart other assessments that they liked … What do we truly see in the system? … And then they took these priority characteristics and went through the VI-SPDAT and started creating their own questions. [Adriana]

They also incorporated people with lived experience of homelessness into the survey design process:

We had individuals with lived experience on that workgroup … They have been helping us frame the questions … The number one thing that people with lived experience continuously remind us is, “Don’t tell me what I need. Ask me what I need. Don’t ask me what happened … Ask me what I need.” So, that way we’re reforming those questions to be a lot more friendly. [Adriana]

Whilst respondents previously showed how resistance includes multiple stakeholders, Adriana provided evidence that managerial staff hire outside experts to gather information from other homeless systems. Her team compared/contrasted these tools to see if they addressed racial dynamics in her homeless system. This was a creative process that combined, adapted, and/or replaced tools developed elsewhere. This included homeless persons who had accessed coordinated entry. Requesting their input helped Adriana’s workgroup create colour-conscious assessment items that invited honest responses from BIPOCs.

Participants also described resistance strategies that they had already implemented. Some CoCs added race-related items to their assessment questionnaire:

[New questions] speak to a person’s experience over their lifespan that are contributing factors … Have you ever been denied housing, public benefits or employment due to a criminal record or due to a credit history … It’s more about the pragmatic day-in-and-day-out things that people of colour are more likely to be more discriminated against … [Heather]

This helped respondents situate their clients’ homelessness in US race relations. Participants also obtained information from outside sources to fact-check BIPOC clients whose distrust of the medical community led them to misrepresent their psychiatric history, “We have just used community mental health as an additional resource … It’s mostly our community mental health providers saying, ‘We actually know that they have schizophrenia’ … And that’s typically what’s happening is they’re bringing in case notes or something to support maybe a change in the score” [Marie]. Triangulating data helped participants resist institutional racism by supplementing deficiencies in the VI-SPDAT. Doing so accounted for risk factors that enhance morbidity for homeless BIPOCs and circumvent barriers that dissuade BIPOCs from engaging homeless services.

Many participants said their CoC adopted a new vulnerability measure. From their perspective, new tools were imperfect, but out-performed the VI-SPDAT. One respondent said their CoC replaced the VI-SPDAT with a new scale. When asked how she thought the new scale performed against the VI-SPDAT, she replied:

I wouldn’t say it fully corrects it because every tool is going to have its issues, but it does give us another perspective, and it’s a little more in-depth and allows us to understand what the needs of the clients are and how successful they can be in certain housing options. [Lisa]

A different CoC continued using the VI-SPDAT but incorporated a questionnaire that measured medical declination to prevent BIPOCs who avoided healthcare services from getting deprioritised:

We have implemented a revision worksheet that is called a “Medical Decliner.” If someone has a history of declining medical procedures or emergency care that would still give them the point there. So, there’re ways to band-aid the tool … [Ted]

Some respondents said their CoC replaced the VI-SPDAT because other CoCs in their state had done so. Biopolitical resistance by the largest CoC therefore diffused to smaller CoCs in the surrounding area:

We are using something called the [vulnerability tool]. We started out with the VI-SPDAT … And then [the CoC who developed the vulnerability tool is] our state’s biggest CoC … They took an initiative to develop the [vulnerability tool] … It provided us the opportunity to have a state-wide assessment. [Kelly]

Some CoCs had thus made a lot of progress in shifting from or reducing dependency on the VI-SPDAT. Our data shows biopolitical resistance occurred through supplementation and replacement. Supplementation may make sense in CoCs that have a predominately White homeless population where BIPOCs infrequently access coordinated entry. Replacement may be more appropriate for CoCs in diverse areas with a history of racial conflicts. BIPOCs in those communities may be less trusting of homeless service providers and therefore need a different assessment tool than the VI-SPDAT.

Deciding to supplement or replace the VI-SPDAT did not mean CoCs chose a colour-conscious alternative. A colour-conscious tool eschews colourblindness by making assessments that reflect hardships that BIPOCs uniquely experience. However, some respondents said their CoC replaced the VI-SPDAT with colourblind assessment questions:

We stopped using [the VI-SPDAT] probably four years ago because … There were reports that came out about it having some racial inequities … We just came up with our own assessment tool … Assessing their chronic homelessness, the time homeless, their disabilities, and then maybe some housing preferences. [Pamela]

Colourblindness thus informed some efforts to promote racial equity. Such well-intentioned acts risk creating new biopolitical exclusions that disadvantage BIPOCs. Variation in the racial project that bureaucrats advance may create differences across CoCs in racial equity. Such variation would mean the ability of BIPOCs to escape homelessness will differ from one CoC to the next.

Discussion & Conclusion

This paper used interview data from 28 homeless systems to analyse the way system managers understand the VI-SPDAT in light of critiques from race scholars, mechanisms through which the VI-SPDAT deprioritises BIPOCs, and any strategies they are using to address racial disparities in their assessment process. Most participants were aware the VI-SPDAT had been critiqued as racially biased; however, some respondents were better informed about the details of those critiques than others. Although some defended the VI-SPDAT, most problematized it as racially biased. Respondents attributed racial disparities to questionnaire design, insensitivity to local race relations, assessor characteristics, and/or institutional mistrust among BIPOCs. From their perspective, the VI-SPDAT’s racial bias was colour-blind institutional racism re/produced by well-intentioned White people. Participants varied in anti-racist resistance. At one extreme, respondents were starting to think of ways to replace the VI-SPDAT whilst others were either experimenting with alternatives or implementing a new assessment tool. Although some participants were looking for an equitable alternative, they adopted a colour-blind replacement that might reproduce exclusions associated with the VI-SPDAT.

We used Foucault’s (2004) biopolitical framework to interpret our findings. The VI-SPDAT measures morbidity and bases programme referrals on these estimates (Grainger 2022). It is thus a biopolitical technology that managerial staff used to allocate an essential resource that clients need to mitigate their exposure to mortality risks (Minnesota Department of Health 2023). This means the VI-SPDAT measures the health of clients so managerial staff can make resource allocation decisions that can impact their survival. Because the VI-SPDAT scores White people higher than BIPOCs (Cronley 2020), it is institutionally racist and creates unequal opportunity for BIPOCs to escape homelessness. Respondents used colour-consciousness to challenge these exclusions. This kind of anti-racist governance resisted biopolitical exclusion by interrogating/changing the way questionnaire items and assessment processes account for racial inequalities. By revising or replacing the VI-SPDAT, participants made BIPOC lives matter by accounting for the impact of racism on their health and chances of escaping homelessness.

Our analysis extends homeless scholarship in several ways. Although BIPOCs are overrepresented in the US homeless population, scholars have under-examined institutional forces that create this disparity (Fowle 2022). Research on the VI-SPDAT shows it deprioritises BIPOCs for permanent housing assistance (Cronley 2020; Kithulgoda, Vaithianathan, and Parsell 2022; Wilkey et al. 2019), thereby perpetuating exposure to mortality risks (Minnesota Department of Health 2023). Recent protests by anti-racist groups correlate with efforts in US homeless systems to create racially just allocation schemes. Key stakeholders have criticized the VI-SPDAT as institutionally racist and advocate its replacement with a colour-conscious alternative (Petry et al. 2021). Whilst previous studies document service providers’ racial critique of the VI-SPDAT, this study showed managerial staff are resisting biopolitical exclusion with colour-conscious alternatives to the VI-SPDAT. This observation extended past research that relied on small case studies (Fritsch et al. 2017; Slota et al. 2021) by presenting data from 28 CoCs that are scattered across the USA. That showed the VI-SPDAT is getting challenged by homeless systems across the USA.

We also advanced scholarship on supportive governance by analysing assistance that managerial staff indirectly give clients. Past research on supportive governance has neglected race relations in homeless systems (DeVerteuil and Wilton 2009; Fowle 2022). The few studies on race have been conducted outside the USA, ignore how institutional racism is fortified or challenged in homeless systems, and overlook the way(s) managerial staff address racial disparities (Clarke and Parsell 2019; Dyb 2021; Lancione 2016; Mostowska 2020; Pauly et al. 2019). This hinders knowledge about mechanisms of racial inclusion/exclusion in US homeless populations. We extended that research by showing how managerial staff use biopolitical resistance to support service recipients. Most participants used colour-consciousness to evaluate their system’s performance. This showed how supportive governance can get expressed as biopolitical resistance (see Selmeczi 2009). This type of anti-racist support connected morbidity, mortality, and race in efforts to replace the VI-SPDAT.

Lastly, our analysis demonstrated colour-consciousness does not always replace colourblind policies. Lancione (2016) shows Italian caseworkers have created stigmatizing environments where interpersonal racism is practised between homeless persons. Whilst that study focusses on race relations, it ignores the bureaucratic logic(s) producing that environment. Furthermore, Lancione’s (2016) analysis overlooks how the design of homeless systems impact the racial composition of homeless populations. Our findings extended that study by specifying ways earnest bureaucrats can unintentionally fortify racial exclusions in a homeless system. Some participants said they replaced the VI-SPDAT with a colourblind alternative despite using colour-consciousness to critique the previous assessment tool. Not accounting for race could perpetuate disparities that they wanted to end. This shows how sincere people can re/produce institutional racism whilst they try to resist it. In doing so, they harm BIPOCs by deprioritising them for permanent housing assistance and perpetuate racial disparities in homelessness (see Fowle 2022).

That said, our study has several limitations. First, we only sampled BFZ communities in the USA. Community Solutions regularly shares information with BFZ communities. It is possible that information about the VI-SPDAT and colour-conscious alternatives was filtered to participants through this network. Non-BFZ communities may not have the same information. It is therefore unclear if our findings are applicable to non-BFZ communities. Relatedly, we conducted this study in the USA where unique racial projects have unfolded over time. This means the VI-SPDAT’s relevance to racial equity may vary in countries with different racial histories. Homeless scholars should thus exercise caution when applying our findings to cases outside the USA. Second, we sampled few purely rural CoCs that did not include a city in its jurisdiction. Rural communities often lack the racial diversity of large US cities (Johnson and Litcher 2022). This study suggests managerial staff in predominately White rural areas were less critical of the VI-SPDAT. That we only interviewed two rural CoCs makes it difficult to know if our findings are transferable to other cases. Third, we largely sampled CoCs in Western and Southern US regions. Racial groups are unevenly distributed across US regions (Frey 2019). Under sampling CoCs in the Northeast and Midwest makes it difficult to know how the VI-SPDAT affects race relations in those areas. These limitations reduced the nuance of our data analysis. Fourth, the racial dynamic of our interviews raises concerns about data validity. Both authors and most participants identified as White. Although this may have created a safe space for White participants to discuss racial equity, BIPOCs may have felt less comfortable talking about those issues. Furthermore, the overrepresentation of White women ironically means that standpoint is dominant in our study.

Our findings can be extended by future research. Above, we stated the VI-SPDAT has been adopted by homeless systems in other Anglo countries. The Homeless Hub of Canada advises homeless systems to replace the VI-SPDAT (Shinn and Richard 2022b). We could not find evidence that advocacy organizations in Australia have issued similar guidance. Future research is needed to understand how system managers in different countries interpret and respond to racial critiques of the VI-SPDAT. Moreover, we showed some US homeless systems are seeking alternative assessment tools. Regional variations in US race relations means comparative ethnographic research is needed to understand how new assessments are being created. Any assessment will likely create new exclusions. It is important to identify new exclusions and reforms that are needed to help those subpopulations. Lastly, homeless scholars can extend this study by addressing the racial dynamics that we listed as limitations. Researchers might consider using BIPOC interviewers to collect data and/or purposefully sample CoCs who have BIPOC system managers. Doing so will help scholars create a more holistic representation of this phenomenon.

Data availablity statement

Masked data is available upon request.

Acknowledgments

We would like to thank the participants who contributed their time to this research, the anonymous reviewers for providing helpful comments on this paper, and the University of Glasgow for giving us small grants to finance this research.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. HUD defines permanent housing as “a community-based housing model, the purpose of which is to provide housing without a designated length of stay … PH programme participants must be the tenant on a lease (or sublease) which must: have an initial term of at least one year; be renewable for a minimum term of one month; [and] be terminable only for cause” (Federal Register 2012, 45426:).

2. Resource constraints forced us to stop sampling.

3. Most participants identified as White. This might reflect the overrepresentation of Whites in social service management positions and non-Whites in social service assistant positions (see US Bureau of Labor Statistics 2022).