Data collection

Data Collection

Estimating Homelessness: Towards A Methodology for Counting The Homeless in Canada

October, 1995

Prepared by:
T. Peressini, Ph.D., L. McDonald, Ph.D., and D. Hulchanski, Ph.D.
Centre for Applied Social Research
Faculty of Social Work
University of Toronto
Project Manager: T. Peressini

EXECUTIVE SUMMARY

Background
Counting the homeless has proven to be a arduous task. Since the early 1980s American activists, government officials and social researchers have produced estimates of the size of the homeless population which have ranged anywhere from 250,000 to three million or more. (Jencks, 1994; Burt, 1992; Cordray & Pion, 1991) Needless to say, this wide range of estimates has been the source of an acrimonious debate about the actual size of the population between advocates and activists on the one hand, and government officials and social researchers on the other. (Jencks, 1994; Kondratas, 1991) The two most contentious aspects of the numbers debate are centered around the issues of who should be included amongst the ranks of the homeless, and how they should be counted. While a consensus has been achieved about the difficulties inherent in defining and counting the homeless, little agreement has been reached as to how to remedy them. At best, we can conclude that most researchers agree on one fact: who we define as homeless determines how we count them.

Canadian researchers have not been immune to this debate, although the issue of homelessness and counting the homeless has received far less attention in Canada. (O'Reilly-Fleming, 1993) Only a handful of attempts to count the homeless have been carried out at both the local and national levels (Fallis & Murray, 1990); these estimates, like their American counterparts, have been similarly called into question. The issues are the same: who we include in a definition of the homeless, and how we measure them. However, unlike their American counterparts, those working in the area in Canada have only recently begun to address these issues.

Defining Who the Homeless Are
As noted, determining who to include in a definition of the homeless is a difficult task. Part of the problem in terms of arriving at an agreed upon definition of the homeless is that the issue of homelessness is an emotionally charged one. No one, be they researcher, government official, advocate, activist or service provider will deny that the condition of homelessness is one that requires immediate and effective action. What they cannot agree upon is who to target for assistance. What action should be taken, and what services are required to alleviate and, if possible, eliminate this social condition? In a time of strained and dwindling fiscal support, on what scale do we provide services? These are the issues that have fueled the debate about 'who' should be counted amongst the ranks of the homeless. For 'who' we define as homeless determines who will and will not be counted and, ultimately, who will be the recipient of the funding, resources and services that are provided to deal with the problem.

While many different definitions of the homeless are used in the research literature, in general they can be classified into one of three types: definitions focusing on the literally homeless; definitions focusing on those who move into and out of homelessness on a regular basis | the marginally homeless/housed; and, definitions focusing on that component of the broader Canadian population who are 'at risk' of becoming homeless.

Counting the Homeless
Many different methods have been employed to study and count the homeless. These range from the more recent telephone surveys of the general population to attempts to include the homeless and other disadvantaged groups in the Canadian and American national censuses. In general, eight basic strategies for enumerating or measuring the homeless population have been used: censuses and one-night counts; key-person surveys; partial counts; heroic extrapolations from partial counts; windshield street surveys; adaptations of area probability designs; service-based designs; and, automated (computerized) client tracking systems. (Iachan & Dennis, 1993; Rossi, 1989)

It should be noted that, apart from attempts to take a census of the homeless population, all of the methods listed above have not been developed for the purpose of 'counting' the homeless. Rather, these techniques have been developed and used to survey members of the homeless population using non-traditional sampling techniques. Typically, a sample of the total population is surveyed, and estimates of the total size of the population are then calculated on the basis of the sample selected. Estimates calculated in this manner are a function of the precision of the sampling design and, for the most part, none of the techniques listed above are as precise as anyone would like.

There are many obstacles to deriving precise estimates of the population. The most pernicious obstacles to developing a standard method for estimating the size of the population include: problems of duplication; timeframe; variations across geographical locality; the number and type of services available in a locality; the degree of service utilization by the homeless; and, the level of cooperation with efforts to enumerate the homeless in the community. To examine these obstacles and the challenges they create for those trying to estimate homelessness, Canada Mortgage and Housing Corporation (CMHC) sponsored a special workshop.

The Workshop on Homelessness
The Workshop on Homelessness brought together a panel of experts from the United States and Canada to critically review the various definitions and methods that are currently available, and to offer recommendations concerning those that are preferred or optimal. The panel consisted of experts from government, the service community, the research community and academia. The workshop was carried out over a three-day period. On the first day of the workshop, the issue of the types of definition of homelessness that should be used was discussed and debated. Participants emerged from this debate agreeing that researchers prefer definitions that are focused on the literally homeless. These types of definition are chosen because they are relatively easy to operationalize and implement, and they provide the highest return in terms of cost effectiveness and representation of the population.

The topic for the second day was the methodological techniques that have been used to count the homeless. The participants arrived at three main conclusions concerning the methodologies for counting the homeless. First, service-based designs provide the highest degree of accuracy, reliability and representation of the population. They are also the most cost-effective of the variety of methods available. Second, it was concluded that attempting to enumerate the homeless on the street, is extremely expensive and has met with very little success. None of the methods that are currently available for counting the homeless includes a street component that is inexpensive or has been deemed successful. Lastly, service-based methods are preferred, but must be adapted to the community/area they are being used in because of variations in the number of services available and the degree to which the homeless use the services.

The topic for the third day was how to ensure that the chosen method for counting the homeless will work in the community. In this regard, as was repeatedly emphasized by participants throughout the workshop, any attempt to enumerate or count the homeless must be accompanied by a great deal of cooperation on the part of everyone involved. Counting the homeless is, of necessity, a collaborative process that requires the cooperation of everyone who is concerned with and involved in the care of the homeless: government officials, service providers, researchers and academics.

Estimating Homelessness: List of Recommendations
In conclusion, three key steps should be kept in mind in any attempt to estimate homelessness. First, support must be garnered for the initiative through development of a community involvement process. Second, community agreement needs to be sought on a precise definition of who is to be considered as homeless. Third, given the definition selected, an appropriate method for estimating homelessness should be chosen. Complete documentation of the initiative should then be kept, detailing; the definition of the homeless; who is and is not counted by what sampling frame; what services are included and excluded in the count; and, how the data are collected.

The key recommendations of workshop participants have been summarized under each of the three critical steps below:

STEP ONE: FACILITATING COMMUNITY INVOLVEMENT

That a committee be formed at the outset of any endeavor to count the homeless, consisting of all relevant members of the service community (directors of services, outreach workers, frontline workers, etc.), as well as relevant members from the government and academic communities. The committee should be responsible for all aspects of the research program, from its development and implementation, to the dissemination of the results.
That the investigators must recognize and acknowledge that providers are primarily concerned with the management, organization and funding of their facilities and programs, and with the establishment of policy. Therefore, efforts should be made at the beginning of a research program to determine exactly the types of information that providers require to address their interests and concerns.
That the interests and information needs of the service community should be incorporated into the research program. This may involve carrying out two counts: an unduplicated estimate of the total number of homeless; and, a count of the number of individuals using services. The research program may also require a survey component in which the specific information required by providers can be collected. Providers may be interested in the following kinds of information about their clients: admission rates, length of stay, recidivism or re-admission rates, use of other services, quality of service received, etc.
That the investigators recognize that such a research program may pose a considerable inconvenience to the participating organization. Therefore, the implementation of the research program must attend to the day-to-day operations of the organization and ensure it is as unobtrusive as possible. Such actions as funding additional staff to assist in the process of collecting data may be effective in alleviating the inconvenience to the organization.
That the rights and privacy of the homeless, and the organization, are guaranteed. The investigators must demonstrate at the outset that the rights of the homeless will not be compromised, that the privacy of the homeless will not be violated, and that information provided by both the homeless and the service providers will be held in total confidentiality. Informed consent must be obtained prior to the implementation of the research program.

STEP TWO: DEFINITIONS

That a definition of the homeless should focus on those individuals who are most in need: those who possess no permanent form of housing and who reside either on the street or in temporary or emergency shelters. The definition should target anyone who is either literally homeless or at imminent risk of becoming homeless.
That the following guidelines be used for defining whether an individual is literally homeless. A person will be considered to be literally homeless if they meet any one of the following criteria: (1) they stayed overnight in a shelter designated for homeless people, runaways, or neglected or abused women; (2) they stayed at least one night in a house, apartment or room paid for with municipal, provincial or federal emergency housing funds; (3) they stayed overnight in a place not meant for human habitation (e.g., a vacant building, a public or commercial facility, a city park, a car or on the street; (4) they have a regular place to stay that is not their own (e.g., people who traded sexual favors for shelter or spent one night in a hotel or hospital); or (5) they use a soup kitchen or emergency food bank for the homeless population.
That, regardless of the definition used, who is included in and who is excluded from the definition is clearly indicated.
That the committee unanimously agree that obtaining information on those who are not included in the definition is extremely expensive, and attempting to include them will compromise the success of the research program.

STEP THREE: METHODS FOR COUNTING THE HOMELESS

That the standard method for enumerating the homeless be service-based.
That the committee unanimously agree that counting the street population (e.g., who do not use services and shelters) is extremely expensive, and attempting to include them will compromise the success of the research program.
That the committee unanimously agree that the use of a service-based methodology which excludes the street population results in an underestimate of the total size of the population, and that the estimate produced reflects only those homeless who are sheltered and use the services included in the research during the time period within which it is carried out.
That the minimum number of sampling frames employed in a service-based method include: shelters; soup kitchens; day programs and outdoor encampments. And, that they be designated as the primary or standard frames to be used in taking a census of the population in urban (service-intensive) areas.
That standardized procedures for the adaptation of such a methodology for use in rural (service-unintensive) areas be developed.
That, where feasible, the shelter-component of service based methods be replaced with an automated client tracking system, such as the ANCHoR (Automated National Client-specific Homeless services Recording) system developed in the United States with funding from the U.S. Department of Housing and Urban Development and other government and non-government actors.
That funding be provided to support ongoing efforts to develop and improve existing methods for counting the homeless; in particular, to develop reliable and accurate methods for counting the street component of the homeless population, as well as methods for enumerating the homeless in rural areas.

back to Research main page

Last updated June 16, 1998

For more information, please contact CHIC@cmhc-schl.gc.ca

CMHC Home | About CMHC | Site Map | Français