Background on the Point-in-Time Count
For many communities, January is an important time of the year to collect data on homelessness
through the U.S. Department of Housing and Urban Development’s (HUD) Point-in-Time Count
(PIT Count) — an attempt to count the number of people experiencing homelessness on a single
night. The count has been mandated by the US Congress, and local communities employee volunteers and service providers to comply with PIT Count guidelines published by HUD.
HUD’s annual PIT Count is meant to provide critical data about individuals living outside and
in shelters; to demonstrate the overall need for services; and to understand homelessness trends across the nation. However, rather than effectively showing the need for increased attention and funding for homelessness, the PIT Count results in inaccurate numbers that undercount the true level of homelessness.
Communities have dramatically different approaches to “counting.” Minimal training for volunteers is only provided at the time of the count, and there is little state or national oversight of the local process. The PIT Count misses large numbers of people, such as those who ride public transportation all night, people staying with friends or family temporarily, or those staying in vehicles, abandoned buildings, and motel rooms (some of which are not officially counted as ‘homeless’ by HUD).
A snapshot of homelessness on one day of the year does not equal a trend – more information is needed to understand changes in homelessness over time. The one day of the outdoor count may be an especially cold night or a time when family decides to take in their relatives. There are too many variables in a one night count to allow any useful discussion about the meaning of this number, even if the volunteers covered every inch of the city and were meticulous talking to every single individual outside.
The flawed census also results in an inaccurate definition of homelessness. According to HUD, the broad definition of homelessness is
“Individuals and families who lack a fixed, regular, and adequate nighttime residence… [and] individuals who will imminently lose their primary nighttime residence… [and] Unaccompanied youth under 25 years of age, or families with children and youth who are defined as homeless under other federal statutes … [and] Individuals and families who are fleeing, or are attempting to flee, domestic violence, dating violence, sexual assault, stalking, or other dangerous or life-threatening conditions and has no other adequate residence or support networks to obtain permanent housing.”
More specifically, an individual qualifies as homeless if their residence will be lost within 14 days of the date of application for homeless assistance and no subsequent housing arrangement has been identified and the family or person lacks the networks for alternative housing. Additionally, a homeless person under this definition cannot have had a lease, ownership interest, or occupancy agreement in permanent
housing within sixty days of applying to the homeless assistance population. A homeless person qualifies as someone who has experienced persistent instability as measured by two moves or more in the past 60 days and is expected to continue under these circumstances.
This definition excludes people living in other homeless situations such as those who “double up” or stay in motels due to lack of alternatives. Moreover, this definition disproportionately excludes homeless families with children and youth. Living on the street puts children at inconceivable risk, which is why many homeless families are forced to stay in motels, with friends, or undesirable circumstances. The inadequacies of the PIT Count give community leaders, government officials, and the public a distorted picture of the true levels of homelessness in America. However, the PIT Count is a large factor in determining a community’s budget, and HUD requires the count in order for a community to receive federal funding for homeless programs. To continue to use this inaccurate data to make funding decisions and effective strategies to address needs and end homelessness is a mistake.
National Coalition for the Homeless (NCH) supports the Department of Education definition over that used by HUD. We also would prefer a more research-based study that includes rigorous controls and methods that could be tested and replicated for accuracy. In the interim, local communities can look to other sources of data such as public schools, early childhood programs, and youth-serving programs such as HeadStart in order to get a more accurate picture of who is experiencing homelessness. The Department of Education data can be better relied on because teachers and supervisors are reporting data from the ground, where they interact with students on a regular basis and understand the unique
situations of families and youth. Lack of appropriate shelter options, fear of child welfare authorities and the safety of shelters, and reductions in transitional housing are the main reasons why most families and young people are not in shelters or on the streets. For example, HUD’s 2018 Annual Homeless Assessment Report estimated that on a single night 53,692 parents and children were experiencing homelessness. But data from the Department of Education shows that 1,508,265 homeless children and youth were identified in the 2017-2018 school year by public schools.
How is the Point in Time data harmful to the population of those without housing?
- These are done by volunteers and there is no standard for how this count is designed and implemented. This report when issued by HUD has an entire section on why this data is not a reliable census, but that is all ignored by the media who use this data as an official count delivered by the federal government in a report every year.
- The Point in Time is a one day count, which is scientifically useless. It cannot be made into a trend or cannot even be compared to the previous year. It could have snowed on the night of the count one year and then the next was a beautifully clear night. These variables have a huge impact on the numbers.
- Some communities use law enforcement as either security for the volunteers or just to provide expertise of those who drive the city every night. There is no prohibition against the police coming back the next night to arrest or harass the individuals counted the previous night.
- The Point in Time Count has some serious deficiencies in being able to count youth, especially LGBTQ+ youth, and many people of color thus diminishing the extent of need within certain populations.
- The Point in Time Count does not result in any changes in the system so therefore it serves no purpose. If a city finds a dramatic increase in families counted on the one day in January compared to the previous January, it does not result in any changes in funding, shelter beds or programming.
- Cities are rewarded with incentives if they can count a smaller number compared to the previous year. This makes the data useless since the outcome is preordained.
- There is not a clear definition of homelessness making it very difficult for the volunteers to understand. This only exacerbates the undercount in homelessness.
The National Coalition for the Homeless believes that it is time to end the annual Point in Time
Count conducted by homeless social service staff and volunteers throughout the United States
because of the harm caused by this one day snapshot to those who experience homelessness.
This mandate by Congress was put in place to verify that the billions going toward addressing
the homeless crisis in the United States was not being wasted. Unfortunately, because of the lack
of any other credible data and the difficulty in counting a migratory population, this annual
report has improperly turned into a census of the total homeless population in a community.
Further, through the complicated funding process, the Department of Housing and Urban Development has incentivized collecting numbers equal to or below the count from the previous year. The media has misreported and misused this data presenting the public with a huge undercount and undermining activists who report fact based statistics on the numbers of homeless people. There is also the limited definition of homelessness used during the Point in Time Count that misses millions of families staying in motels, vehicles, and young people who move frequently.
Alternatives for Congress
In order for Congress to receive solid data, we must engage professional researchers to conduct a more scientifically rigorous survey, and then employ statistical sampling to get a legitimate number for how many people are living without housing. It will require a large investment, but NCH believes it would be the right way to ensure accurate data is being used to allocate federal resources. Further models of data collection are listed below.
- Fund University backed research in a number of cities. HUD would need to increase the allocation of funds to get an accurate count. They would need fund to fund a sample of large, small, and suburban communities to come to a better estimate of the number experiencing homelessness. There are central intake to the shelters, poverty figures from the US Census, eviction data, calls to government telephone helplines, housing waiting lists, food requests, and accurate school data that could be aggregated to come up with an accurate number for those who lose their housing and spend a period of time unhoused. These samples could then provide an estimate for the national population that is far more accurate compared to the guessing game done every January. This would need to be tested and monitored so that it would be able to be published in a scientific journal. It would need to be peer reviewed and would be held to the highest academic and scientific standards.
Alternatives for Local Media and Communities
Instead of using the flawed Point in Time counts, communities and the media should use a number of other data sources to create a more comprehensive understanding of homelessness that better represents existing populations and to inform policy decisions and funding. None of these sources are comprehensive or scientifically accurate census information of those without housing, but they are far more accurate compared to the Point in Time Count. These data sources are more specific, and are not as likely to be confused with a comprehensive census like the point in time count is almost always misconstrued. Point in Time counts are not a census nor can they show trends. Here are some alternatives for the local community to use:
- University based research on the numbers. Some communities with large research based universities do try to compile local data sources to come up with accurate numbers of homeless people in a community. There are a few cities in the United States with poverty centers that are able to use massive data dumps to come up with a fairly accurate number of people experiencing homelessness in a city or accurate numbers experiencing poverty.
- Real-time Data and By-Name Lists. Communities can create new systems to collect current data on their homeless populations using central intake and requests for help. The Los Angeles Homeless Services Authority, for instance, established an online portal that allows anyone to report any unhoused person who needs help in the community.
- U.S. Department of Education Data. The Department of Education uses a more comprehensive and accurate definition of homelessness than HUD, leading to more accurate numbers. Communities can use the data collected on homeless children and youth to have more accurate numbers and better understand trends. (https://schoolhouseconnection.org/the-pitfalls-of-huds-point-in-time-count/)
- The US Census and the American Community Survey. The US Census numbers are important scientific markers that are far more accurate than the PIT counts. Since a percentage of persons who live in poverty become homeless every year, it is easier to pick that number for a good estimate of homelessness. For communities with a weak housing market, unhoused persons typically represent about 5%-8% of the those living in poverty; in a tight housing market, it is 10%-16% of those living in poverty. Suburban homelessness translates to around 3%-7% of those living in poverty. Census figures are not as accurate for rural homelessness because of a number of other factors.
- Housing Waiting Lists. If your housing waiting list is allowed to increase without shutting down that is a good number of self-declared housing insecurity in a community. It is far more accurate compared to the PIT counts.
- 2-1-1 Calls. 2-1-1 connects provides information and connects individuals to health and human services and other homeless services. The number of people who called 2-1-1 can be a helpful number to track homelessness trends. Some communities break that down by housing related or shelter calls and can separate out duplicate callers. The accuracy of these numbers vary dramatically by community, but they are not confused with a census, and they are good for showing trends.
- Homeless Deaths. December 21 is Homeless Persons’ Memorial Day when communities commemorate the lives of those lost among the homeless community each year. These memorials often have a list of names for everyone who died that year, which can serve as a good indicator of trends. Communities should also have their medical examiner record housing status in their records in order to better track homeless deaths in their community to increase the accuracy of these statistics.
- Data Sampling. Cities can use existing data from sources outside PIT counts and use data sampling methods to estimate the number of unsheltered people in a community.
- Coordinated Entry. Coordinated Entry System/Central Intake is a system which allows the shelters and services to collect data in a centralized location. Central intake serves as a single point of entry for all homeless and at-risk persons within the community connecting to homeless services. Central intake can provide good data on the number of people seeking assistance in the community. They use weekly data updates to determine the effectiveness of various strategies and can use leveraging technology and data from around the city or county to provide a more accurate picture of homelessness.
- Stratified Sample Design/ Horvitz-Thompson (HT) estimator. This is a design in which one samples a subset of areas based on the density of homelessness, and estimates country- wide data based on the results of the sampled regions. It provides a way to more systematically deploy existing resources in a way that allows for estimation. Additionally, it gives additional and more comprehensive canvassing in low-density, suburban regions. Some suggestions include using plant-capture and post-count surveys in the sampling approach. Plant-capture methods involve sending homeless decoys or “plants” to various locations on the night of the count, and then documenting which decoys are or are not surveyed. Post-count surveys: CoC staff enlist the support of service providers to survey individuals experiencing homelessness in the days after the count. (https://www.hennepin.us/-/media/hennepinus/your-government/projects-initiatives/coc/evaluating-hc-PIT-count-report.pdf)
- Multiplicity sampling. Residents of apartment, condominium, or single room housing were called and asked to provide estimates of hidden homeless on complex property. The calls would be made at different times of day, also allowing calls to be scheduled, ensuring that the researcher can reach the respondent. This method takes into account the network size of individuals being surveyed. Respondents are also asked to report the number of hidden homeless people on their neighborhoods property. Interviewers are taught techniques on dealing with reluctance and refusal. (https://www.proquest.com/docview/1821868775?accountid=8285&parentSessionId=aE9BjKdbEJ5uYozSrldTx5eP3if1z07klj6BMT0SAWY%3D&pq-origsite=primo, https://sciendo.com/pdf/10.2478/jos-2014-0014)
- Nationally representative phone survey. This is a method which interviews a representative broad base of the population in the nation. In the past, Voices of America focused this research on young adults and youth ages 13-25 with a sample of approximately 26,000 people. Respondents answered questions about youth homelessness and their personal experiences. Although, this plan could be implemented for the entire population if capacities warranted it. This method is cost efficient and replicable. Additionally, it is reinforced through a sample of follow-up interviews in an effort to learn more. Supporting procedures in this technique are in-depth interviews and brief surveys. (https://voicesofyouthcount.org/wp-content/uploads/2017/11/VoYC-National-Estimates-Brief-Chapin-Hall-2017.pdf)
- Next-day survey in social services. This method conducts surveys in feeding programs, warming drop-in centers, outreach programs, healthcare for the homeless and asked people where they slept the night before. (Ending Homelessness: Burnes and DiLeo)
- HMIS (Homeless Management Information System) plus survey. HMIS data is collected throughout the year to create an aggregated list of unhoused persons by name who are currently connected to homeless services. Researchers could use this method to observe the quantity of people making use of these programs in order to track the total number of homeless individuals. NCH has always opposed HMIS data as extremely harmful to the privacy of those who lose their housing, but HUD does require the use of HMIS in much of the US. One benefit to this method is that the information is already stored and available in the HMIS system. Cities such as New York City and Philadelphia have used this count. (Ending Homelessness: Burnes and DiLeo)
- NSHAPC (National Survey of Homeless Assistance Providers and Clients). This is a service-based enumeration technique that used statistically valid sampling methods in order to utilize sixteen types of services homeless people may utilize, and sampled 4,000 people. (Ending Homelessness: Burnes and DiLeo)
- Peter Rossi Method. This method divides geography into blocks depending on the high and low probability of encountering homeless people, goes to all the high-probability blocks and a sample of the low probability blocks, and estimates the number of people in a low probability spot. This method depends on planners’ capability to identify high probability blocks. Cities such as New York City and Los Angeles utilized this method. (Ending Homelessness: Burnes and DiLeo)