Defining “At-risk of Homelessness”: Re-connecting Causes, Mechanisms and Risk

Abstract

A key focus of homelessness policy across the Anglophone world is to prevent homelessness by targeting interventions to those considered “at-risk”, yet the term “at-risk of homelessness” remains ambiguous. A solid definition is required. Typically, risk is defined using those factors that are over-represented in the population of interest. However, this approach tends to obscure the relationship between risk factors and their broader causes. It is also unfeasible given the characteristics of the current body of knowledge on homelessness. This paper takes a different approach. It argues that clarifying what is meant by “at-risk of homelessness” requires understanding the type(s) of causation involved, the various causes of homelessness and the mechanisms through which they act. The paper addresses each of these three requirements to propose a provisional definition of “at-risk of homelessness”. Following further empirical work, this definition can be used to enumerate the “at-risk of homelessness” population and design prevention strategies.

Keywords:

• Homelessness
• at-risk
• cause
• new consensus
• definition

Acknowledgements

I would like to thank my PhD supervisors Wendy Stone, Beth Webster and in particular Kath Hulse (Swinburne University) for their support. Thanks also to Dr Karina Roberts, Assoc. Prof Tim Moore, Ms Lee Palmer and my anonymous peer reviewers for their helpful feedback.

Notes

1. Chamberlain and MacKenzie (2014) make this same argument in relation to homelessness.

2. The term “hybrid model” is also used by Bramley and Fitzpatrick (2017) to refer to the new consensus view. Their article was published after this article was submitted for review and this similarity is entirely coincidental.

3. For example, both critical realism (Fitzpatrick 2005) and analytical sociology (Hedström and Ylikoski 2010) use a mechanistic account of causation.

4. For further information on the diversity of causal relationships and theories of causation see: Reiss (2007, 2009); Williamson (2006); Cartwright (2004); Casini (2012)

5. A further problem with the new consensus view is that it confuses the level of measurement of a cause or risk factor with the level of the cause per se. Garside’s example of ethnicity, cited earlier, is illustrative. While ethnicity is measured at the individual level, the reason that certain ethnic groups have poorer health incomes is not anything inherent in their ethnicity, but the broader social context in which these groups are more socially and economically disadvantaged. The level of measurement of the variables does not automatically correspond to the level of cause.

6. For a summary of this literature see Wood et al. (2014, 2015)

7. Hodgson (2006) defines organizations as: “special institutions that involve (a) criteria to establish their boundaries and to distinguish their members from non-members, (b) principles of sovereignty concerning who is in charge, and (c) chains of command delineating responsibilities within the organization”(Hodgson 2006, 8).

8. It will be important to distinguish between whether past experiences of homelessness make homelessness more likely because of the mechanisms outlined here, or whether past experiences of homelessness indicate higher risk of homelessness due to the presence of factors that brought about the first episode of homelessness – which may be the case in some instances and so this type of cause would not, in those cases, play an actual causal role.