This is a post peer-review, pre-copy-editing author’s version of a paper now published as Walker MJ, Franklin J. 2018. An argument against drug testing welfare recipients. Kennedy Institute of Ethics Journal 28(3): 309-360. Available at: https://muse.jhu.edu/article/707066. AN ARGUMENT AGAINST DRUG TESTING WELFARE RECIPIENTS Mary Jean Walker and James Franklin Abstract: Programs of drug testing welfare recipients are increasingly common in US states and have been considered elsewhere. Though often intensely debated, such programs are complicated to evaluate because their aims are ambiguous – aims like saving money may be in tension with aims like referring people to treatment. We assess such programs using a proportionality approach, which requires that for ethical acceptability a practice must be: reasonably likely to meet its aims, sufficiently important in purpose as to outweigh harms incurred, and lower in costs than feasible alternatives. In the light of empirical findings, we argue that the programs fail the three requirements. Pursuing recreational drug users is not important in the light of costs incurred, while dependent users who may require referral are usually identifiable without testing and typically need a broader approach than one focussing on drugs. Drug testing of welfare recipients is therefore not ethically acceptable policy. Programs of drug testing welfare recipients have been introduced in a number of states of the USA and in New Zealand. The practice has also been proposed but not implemented in Canada and the United Kingdom (Wincup 2014). Recently, legislation was prepared to introduce 1 drug testing of welfare recipients in Australia, and more US states have drafted legislation. Stated aims of these programs include: to identify people with drug problems in order to refer them to treatment, with the longer-term aim of facilitating their gaining employment; to prevent welfare payments being spent on illicit substances; and protecting the children of people dependent on drugs (see, e.g., Bolen 2014, 86; Schaberg 2012, 575). But these different aims imply quite different responses to drug test results, and ambiguity surrounding what these programs aim to do complicates attempts to evaluate their success. In this paper, we argue that programs of drug testing welfare recipients1 are not ethically acceptable. We approach assessing the programs’ ethical acceptability as a question of proportionality. This is an appropriate approach because drug testing imposes some costs and burdens on welfare recipients, for the purpose of achieving aims thought to outweigh those costs and burdens. Thus, their ethical acceptability relies on an assessment of the worth of their aims, considered in relation to the costs and burdens. 2 Since the proportionality analysis allows an ethical assessment that does not rely on any particular position about welfare rights or government paternalism, this approach may have a broader appeal than ethical assessments made from such positions (though it may also be consistent with some such assessments). We first briefly overview information on existing programs, in section 1. In section 2 we explain our approach to evaluating the ethical acceptability of these programs. The approach implies that for ethical acceptability, a program of drug testing must meet three criteria: that the program can plausibly be expected to meet its aim(s); that its aim(s) are sufficiently important to justify its costs and burdens; and that it is not substantially more costly and burdensome than feasible alternatives. This implies the need to further clarify what the aims of drug testing welfare recipients are, undertaken in section 3; and to examine empirical studies relevant to 2 whether programs will meet the three criteria, undertaken in section 4. Drawing on this, in section 5, we examine whether programs of drug testing can meet the three criteria in relation to any of its possible aims. We argue that it cannot, and conclude that it is not ethically acceptable policy. 1. EXISTING PROGRAMS OF DRUG TESTING WELFARE RECIPIENTS In the USA, the federal Personal Responsibility and Work Opportunity Reconciliation Act 1996 (PRWORA) granted states authority to drug test welfare recipients. In 1999, a pilot program of drug testing was introduced in Michigan. This program was terminated following a legal challenge on the basis that it “violated constitutional protections against unreasonable searches”, as was a program in Florida a decade later (NCSL 2017).3 Despite this, drug testing of some welfare recipients is increasingly becoming the norm. As of early 2017, 17 US states had drafted and 15 states had passed legislation for a drug testing program (NCSL 2017). Details of how the programs work differ considerably by state. Most relate to recipients of the Temporary Assistance for Needy Families (TANF) payment, which provides cash assistance, and programs promoting preparation for work, to low-income families. Some states have introduced legislation to test recipients of other assistance programs such as the Supplemental Nutrition Assistance Program (SNAP) or Medicaid (NCSL 2017).4 Some test new applicants, others test recipients on an ongoing basis. In most states, applicants or recipients are first screened for potential drug problems using a psychological assessment tool or questionnaire, and only those indicated as likely to be drug dependent are tested. This process circumvents legal challenges based on drug testing constituting a suspicionless search proscribed by the Fourth Amendment (NCSL 2017). In some locations, people testing positive or refusing a test are 3 excluded from benefits; in others, benefits may be retained on condition of attending drug treatment, or (for payments intended for the support of dependent children) benefits may be paid to an alternative adult (Hall 2016). There are also various procedural differences, such as in how programs guard against the possibility of false positives (DHHS 2011, 6). Data collection and evaluation of the programs appears to be relatively scant, and differences between the programs complicate drawing general conclusions. In a pilot program run in Florida during 1999-2000, among 6,462 applicants, 1,447 or 22.4% were indicated by a psychological assessment tool to have a drug dependence problem. Of the 1,447 referred for urinalysis, 335 or 5.1% of the total screened population tested positive (Crew and Davis 2003). Testing was in most cases undertaken immediately after the application was made, but occasionally was done on the following day (Crew and Davis 2003, 42). Hall (2016, 2 of 7) reports positive test rates within a range of 3.2% to 16.9% among those referred to a test following screening. Hall’s data shows that the percentage testing positive of the total population screened in 9 states was 0.19%; if refused tests (where reported) are treated as positives, the rate is 0.57% (Hall 2016, appendix D). The cost per positive test varies considerably, from $200 in Tennessee to $7,006 in Missouri (Hall 2016, 2 of 7). The program in New Zealand is quite different, and could be considered a method of subsidizing employer costs for pre-employment drug testing rather than a program of drug testing welfare recipients. Since 2013, people receiving some welfare payments must undertake a drug test where this is a requirement of a potential employer or trainer. 5 Though employers usually pay for pre-employment testing in New Zealand, the reforms enable employers to invoice the state for the cost of a positive test. New Zealand’s Ministry of Social Development passes this cost on to the welfare recipient, and may impose a sanction. The sanction does not 4 automatically remove welfare payments, but is part of New Zealand’s system of graduated sanctions (Wincup 2014; New Zealand Ministry of Social Development 2016). Some data have been released in response to Freedom of Information requests. In 2015, there were just under 32,000 referrals of job seekers to employers who required a preemployment drug test. In this time, 55 sanctions were imposed in areas which were reported on (some areas where numbers were low were omitted to protect privacy), and less than 5 people had their benefits cancelled or suspended. From September 2013 to September 2016, there were 466 drug test-related obligation failures, including test refusals as well as positive tests (New Zealand Ministry of Social Development 2016). 2. ETHICAL ASSESSMENT OF DRUG TESTING WELFARE RECIPIENTS We will assess the ethical acceptability of programs of drug testing welfare recipients using a ‘proportionality’ approach. The approach is suitable for analysing policies which impose costs or burdens in order to achieve some valued aim. This approach involves assessing whether drug testing programs: 1) Are reasonably likely to meet their aim(s); 2) Impose costs/burdens that are not out of proportion to the importance of their aim(s); and 3) Do not impose substantially more costs/burdens than feasible alternatives. The criteria are similar to some of those used to assess the justness of wars, analogous to principles of a reasonable chance of success, proportionality, and necessity (e.g., Walzer 1977). 5 Similar criteria are used in the jurisprudence of a number of countries and in international law, as a tool for assessing laws that involve rights infringements (e.g., Cianciardo 2010). Similar principles have also been proposed as useful in assessing the ethical acceptability of public health measures (Childress et al 2002; Kass 2001) and in ethical decision-making in healthcare generally (Hermerin 2012). While we shall not attempt an in-depth justification of this approach, it is worth some brief comments. The three criteria can be justified quite intuitively: a policy intervention cannot be justified on the basis of its aims if those aims are not empirically linked to the policy; if it imposes costs and burdens out of proportion to the importance of its aims; or if the aims could be met in another way that involves fewer costs or burdens. This approach might appear consequentialist, since it involves comparison of outcomes with and without the programs, in terms of their costs and benefits. However, as Rodin (2011) has argued, proportionality assessments incorporate both consequentialist and deontological thinking. Rodin notes that harm can be justified either when someone has made themselves liable to being harmed (e.g., by committing an unjust attack), or when the harm is a lesser evil (i.e., the harm prevents a greater harm occurring). Similarly, the imposition of a rights infringement (whether or not this counts as a ‘harm’) could be justified because the person on whom it is imposed is liable to it, or because it will lead to the best overall consequences. In the former, we draw on deontological intuitions, involving normative assessment of the liable person; in the latter, we draw on consequentialist thinking and compare outcomes. Rodin argues that judgements that someone is liable to harm – that it is justifiable to regard their rights to be suspended in a localized way – relate to judgements of proportionality, because liability is from the start assessed in relation to some good to be achieved by the harm (2011, 77). 6 For example, by threatening violence in the form of a punch, an aggressor might make themselves liable to force that would prevent the threatened punch, but is not thereby liable to force of any kind and degree. That is, the judgement as to whether someone is liable or not already involves a judgement about proportionality, because the person can only be liable to a harm that is proportionate (2011, 79). In the case of drug testing, analogically, we could consider the general right not to be drug tested to be suspended, either because some feature of being a welfare recipient makes people liable to bearing this rights infringement, or because imposing the rights infringement leads to better outcomes overall than not imposing it. But at base, these two approaches are linked. On the one hand, assessment of whether welfare recipients are liable to accept drug testing must take into account the degree of seriousness of the rights infringement in relation to the purpose it serves. This involves judging how valuable the aim is – and this could derive from various ethical views including non-consequentialist theories. On the other hand, the weighing of outcomes would include, among various other factors, the rights infringement itself on the ‘cost’ side of the equation. Here again, the weighing would require an assessment of how important the aim is (2011, 97). We recognize that there are other kinds of ethical argument against drug testing welfare recipients, for instance, those deriving from particular positions on welfare rights, government paternalism, or considerations about ethics of drug use (Perez-Munoz 2017). But any such argument is likely to be highly controversial, and its acceptance may rely on underlying normative commitments or political preferences notwithstanding its philosophical rigour. Our approach does not rely on premises about these matters, and as such, we hope, may have a broader appeal. The proportionality approach implies a structure for our assessment in the remainder of the 7 paper. First, we need to understand what the aims of drug testing welfare recipients are (section 3). Second, we need to consider whether the programs are likely to meet these aims, what costs and benefits they are likely to result in, and whether there are other, less costly, methods of meeting the aims (section 4). We can then apply this to assessing whether the programs meet our three criteria (section 5). 3. AIMS OF DRUG TESTING PROGRAMS In subsection 3.1 we examine aims that have been stated in media statements, legal debate, and policy documents. We analyse these suggestions to show there are (at least) five proposed final aims of drug testing, served by two intermediary aims, which drug testing might achieve via different mechanisms. Some of the aims might, however, also be undermined by the alternative aim or the mechanism used to achieve it. In subsection 3.2, we consider the suggestion of various commentators that drug testing welfare recipients serves unstated or implicit aims related to social biases, broader policy agendas, and/or political expediency. 3.1. Stated Aims Media statements indicate several aims of drug testing programs: preventing welfare payments being spent on illicit drugs, protecting the children of drug-using welfare recipients, helping drug-dependent people on welfare, deterring drug use, and reducing government spending. For example, explaining his support of a bill to drug test welfare recipients in Alabama, one representative stated, “I don’t think it’s right for taxpayers to have to fund somebody’s drug habit” (cited in Chandler 2014). A representative from Michigan explained his support of a drug testing bill by saying, “[w]e have children that are starving in this state because 8 moms, dads or both are on methamphetamines or cocaine or some other horrible substance, and they sell their bridge cards instead of feeding their children” (cited in Oosting 2014). An Australian minister stated of the recently-planned trial that its aim was to help people overcome drug problems and obtain work, rather than leaving them “at risk of a cycle of welfare dependency" (cited in AAP 2017). In New Zealand, following news reports that during the first six months of the program, only 22 out of 8001 tests were positive, government officials stated that this showed the program was successfully deterring drug use (Fisher 2014). A news story reporting that in Utah, only 262 of 30,000 tests during 2012-13 were either positive or refused, and the total program cost was $350,000, quoted officials as saying this represented a good investment, as denial of payments to 262 people would have amounted to more than the program cost over time (Price 2013). The aims of drug testing programs have also been discussed in relation to legal cases surrounding several of the US programs. The Fourth Amendment requires that searches be justified by reasonable suspicion of illegal activity, or serving a special need of the government. Drug tests are legally recognized as searches, and while most programs now aim to meet the requirement of ‘reasonable suspicion’, there has been debate about what ‘special needs’ they could serve. In relation to the early program in Michigan, Courts considered government needs to protect children from child abuse and neglect, and to ensure that public funds are not used to buy illicit drugs (Schaberg 2012, 575). McLaughlin (2013, 581) suggests three special needs: “promoting self-sufficiency, ensuring public funds are used for their intended purpose, and protecting children in homes with drug addicted parents”. ‘Promoting self-sufficiency’ is explained as helping people to stop using drugs in order to find work (2013, 581-2). Widelitz (2011, 299-300) proposes the policy could serve five special needs: combating drug use among 9 people receiving welfare; protecting children in families receiving welfare; removing drug use as a barrier to work; preventing children in families receiving welfare from developing future drug problems; and public safety. The latter is explained by noting that drug use is statistically linked to crime (2011, 305). Similarly, the explanatory memorandum to the proposed Australian legislation stated two aims of the testing. First was improving “a recipient’s capacity to find employment or participate in education or training by identifying people with drug use issues and assisting them to undertake treatment” (Parliament of the Commonwealth of Australia 2017, 67). Second was to “maintain the integrity of, and public confidence in, the social security system by ensuring that taxpayer funded welfare payments are not being used to purchase drugs or support substance abuse” (2017, 156). Of course, a policy’s having more than one aim is not a drawback; it may be a positive. But consideration of exactly how drug testing could bring about the various aims reveals some underlying tensions. Notice that different aims imply two different responses to a positive drug test result. The aim of reducing government spending indicates excluding people from payments upon a positive or refused test, and contraindicates referral to treatment, which would increase costs. The aim of improving employment outcomes would indicate referral to treatment as most appropriate.6 The aim of ensuring welfare payments are not spent on drugs could be met by removing payments, or by reducing drug use itself. Drug testing could lead to reduced drug use either by referring those testing positive to treatment, or by deterrence (which could be an effect of testing itself, or an effect of a threat of exclusion from benefits). The aim of reducing public safety risks linked to drug-related crime might, again, be achieved by reducing drug use, while exclusion from payments could increase crime, as people seek income from other sources. This 10 indicates the response of treatment referral (though the threat of a test or of exclusion upon a positive test might, again, reduce use via deterrence). If the aim is to protect children, this might be achieved by reducing use (again via treatment referral or deterrence), while exclusion from benefits seems more likely to harm children. The various possible causal links indicated between drug testing and its aims are mapped in Figure 1. The diagram involves simplifications (e.g., we represent drug testing itself as a deterrent but omit the deterrent effect of the knowledge that a positive test could lead to exclusion from benefits; deterring people from claiming benefits might also impact on several of the aims; some of the final aims might causally impact on each other; and so on). However, any attempt at policy analysis of this kind must involve some such simplifications. The aim of ‘excluding drug users from benefits’ is also somewhat artificial as it is not typically explicitly stated as an aim, but is an inferred intermediary aim serving the further stated aims. Fig 1. Pathways for proposed effects of drug testing welfare recipients This indicates that there are two ‘intermediary’ aims, of reducing drug use amongst welfare recipients, and of excluding those using drugs from this population, and two mechanisms by 11 which drug tests could achieve the former aim. All but one of the five possible final aims could be met by reducing drug use, while two could be met by payment exclusion. 3.2. Possible Unstated Aims It has also been suggested that drug testing programs are actually motivated by negative feelings about or moral disapproval of welfare beneficiaries, the poor, drug users, and/or members of some races. For instance, media commentators have stated that drug testing is a way to punish welfare recipients, that it reflects attitudes related to the stigmatisation of welfare recipients, and that it relates to constructing poverty and/or drug use as individual rather than societal problems (e.g., Forsyth and Banham 2017; Pollack 2013). Welfare recipients have long been the subject of various stereotypes, including being likely to use or abuse drugs. Evidence indicates that this stereotype is unwarranted: estimates of drug use among welfare recipients range from 4 to 37%, but overall, drug use prevalence is likely to be only several percentage points higher in the unemployed population than among the general population (DHHS 2011, 2). Despite this, there is a widespread belief that drug use is much higher among welfare recipients (Amundsen et al 2015, 446). US Policy discourse surrounding drug testing programs often presents welfare recipients as not only generally drug-users, but also as unworthy, and unemployed due to personal failings (including lack of moral principles) or by choice (Amundsen et al 2015, 448; Amundsen et al 2014, 15). Drug testing may thus be construed as having a punitive aim towards this stigmatized group. In addition, drug testing welfare recipients is part of a broader trend towards welfare conditionality, occurring in many countries over the last quarter-century. In the USA, this coincided with moves toward harsher criminal sentencing, and harsh criminal justice policies can 12 also be seen to affect welfare policy. For instance, PRWORA bars those with drug felony convictions from many kinds of welfare entirely, though states may opt out of this eligibility rule. US drug testing programs can thus be interpreted as another extension of harsh criminal justice policy into welfare systems (Amundsen et al 2015, 447), penalising not only drug users but the entire welfare recipient population (Amundsen et al 2014, 23). Further, drug testing of welfare recipients can be argued to have political motivations. As with other anti-drug action, it has the effect of linking drug use to a marginalized group with low electoral power (Amundsen et al 2015, 443-4). This has been raised in particular with regard to US programs, where stereotyping of welfare recipients intersects with racial stereotyping, such that welfare recipients are stereotypically Black people (Amundsen 2015, 445; Alexander 2010, 47-8). Indeed, Ledford (2016) shows that drug testing of welfare recipients is more likely to be adopted in states where there is a higher proportion of Black people on TANF, and where there are higher levels of ‘symbolic racism’ (a combination of negative affect towards a group and beliefs that they violate certain cherished values, measured using survey data on social attitudes). Ledford suggests that discrimination against Black people may be more fundamental to the introduction of drug testing programs in the US than discrimination against welfare recipients in general. In concert with various other policies (such as barring convicted felons from voting), drug testing could be interpreted as not only reflecting racial biases, but as part of a widespread effort to reinforce inequalities and social oppression, particularly of Black people, while maintaining a veneer of ‘colour-blindness’ (Alexander 2010, 52). In this context, some suggest that drug testing of welfare recipients has unstated and perhaps implicit aims, of punishing welfare seeking, reducing access to government support in general, or more generally of reinforcing current social inequities. (On the part of politicians proposing drug testing programs, 13 there could also be an aim of appealing to an attitude of approval of harsh treatment of welfare recipients among the general public.) Notice that if the aim of drug testing welfare recipients is understood in this way, the policy would automatically fail on the proportionality approach. As explained above, the proportionality approach is useful for evaluating policies that introduce costs or burdens for the sake of some valued aim. But these unstated aims are ones that, even if they are valued by some, are themselves unethical and should not be valued; or (we might say in an objectivist mood) are valued but not valuable. If it is true that the aims of drug testing programs are actually unethical, then drug testing cannot be an ethical policy, let alone do any work in justifying the harms or costs its achievement involves. 7 Further possibilities are that some proponents of drug testing programs are motivated by these unstated aims, while others genuinely believe the stated aims; or that, though the programs are motivated in ways that are ethically suspect, they might still meet their stated aims. In these cases – assuming the stated aims have some value – it is not clear what the moral relevance of the motives behind a policy (as distinct from stated policy aims) should be. At the very least, we would need the achievement of the stated aims to do the work of justifying or compensating for not only the harms and costs of the programs, but also for introducing a policy on the basis of unethical motives (raising further difficult questions about whether anything could ‘compensate’ for such motives, which may be assessed differently on different normative theories). That is, if drug testing programs meet (at least some of) their stated aims in a way that meets the criteria of the proportionality approach, we would need to assess whether these unstated aims are playing a role, and how the ethically problematic status of these aims affects the overall ethical acceptability of the policy. Thus, in order to have any chance of being 14 ethically acceptable policies, drug testing must meet some of its ethically acceptable, valued aims – to do the work of justifying not only the costs and burdens imposed in introducing them, but also of outweighing or silencing ethical concerns about such motivations. Since we will argue that drug testing welfare recipients cannot meet their stated aims in a way that satisfies the proportionality approach, however, this further question will not be dealt with here. 8 As such, we will not discuss the unstated aims further below as aims of drug testing programs. We do of course recognise that drug testing welfare recipients may in fact have such results as greater burdens on welfare recipients that are experienced as punishments, reduced access to government support, and the reinforcement of oppressive, often racist, norms, inequities, and stereotypes. There is empirical evidence for some of these claims, though some are not easy to measure, which is discussed below. These important issues are thus recognised in the proportionality analysis, although we will treat them as costs or negatives of the policies, rather than as aims that might help justify their other costs. We are therefore treating them as unintentional consequences, whether or not this is always true, and we take this to be a way of applying a principle of charity to the arguments provided for drug testing programs. It is possible after all that, even if some proponents of drug testing programs are motivated by implicit biases, others genuinely believe they will benefit drug users and their children, or hold a combination of the ‘stated’ and ‘unstated’ aims. Focusing solely on the unstated aims runs the risk of targeting a straw man by failing to apply the principle of charity sufficiently. In the next section we thus turn to empirical evidence on the question of whether drug tests are likely to meet any of their stated aims, what costs they will impose in the process, and whether there are other means of achieving these aims. 15 4. EMPIRICAL EVIDENCE 4.1. Is Drug Testing Welfare Recipients Likely to Meet its Aims? There is relevant empirical evidence available on some of the proposed causal links represented in Figure 1, and on some of the assumptions that underlie them. The first proposed causal link is between a drug testing program and a reduction in drug use. There is no direct evidence available on whether reduced drug use would be achieved via deterrence. Such evidence would be very difficult to obtain, though there is evidence that deterrence can be effective in some similar cases, such as random breath testing of drivers (Terer and Brown 2014). As to drug testing leading to reductions in drug use via treatment referrals, we must first note that most people who use drugs are not dependent, and therefore not candidates for treatment (Grant 1996). The outcomes of treating non-dependent people for drug dependence are not known, though it is hard to see how it could produce desired outcomes. This raises questions about the assumption that drug testing is a good method to identify those with drug dependence, insofar as this is the targeted group (as treatment referral practices, and some of the rationales linking to the final aims imply). Positive drug test results do not provide information on whether someone is dependent, or uses drugs occasionally or recreationally – and nor does a negative result establish that someone is not dependent. 9 The sensitivity and specificity of drug testing for detecting dependence will differ depending on how a program is run, but some indicative figures are developed by Pollack et al (2002a). This study used survey data on drug use among welfare recipients in the USA to estimate that approximately 4% of welfare recipients would be drug dependent, between 11 and 18% would use drugs recreationally, and 22% would have alcohol use or other psychiatric disorders, but not use drugs (2002a, 25). These estimates indicate that drug testing could identify as many or even 16 more recreational as dependent drug users (2002a, 30). Thus (even if drug tests could provide completely accurate results), using drug testing to identify dependence will mean there will be some ‘false positives’ (non-dependent drug users) and ‘false negatives’ (people who are drug dependent but test negative). For the ‘true positives’ (people who test positive who are dependent on drugs), evidence about drug treatment suggests that it can effectively reduce drug use. There is a great deal of research on a range of different treatment modalities, which cannot be discussed in detail here. Reviews note some limitations on effectiveness (comparable with treatments for psychiatric disorders), and many treatment facilities struggle for adequate resources which may reduce potential effectiveness (Pollack 2017; Dutra et al 2008). Thus “[v]iewed at the individual level, relapse and polysubstance use are typical experiences among clients in SUD [substance use disorder] treatment interventions” (Pollack 2017, 165). Coerced or mandatory treatment might have different outcomes to voluntary treatment (Werb et al 2015). Nonetheless, the large evidence base surrounding drug treatment indicates that some reductions in use are likely. There is some evidence indicating that drug treatment can increase welfare recipients’ likelihood of finding employment, but also some conflicting evidence. One US study found that recipients who completed drug treatment were twice as likely to be employed two years later than members of a control group, but no benefits were achieved when the program was scaled up (Bloom et al 2001, 4-6). Metsch et al (1999, 42) found that amongst women leaving a drug rehabilitation program in Florida, those who had completed the program were more likely to be working than those who had not (37 versus 13%) (though such correlations might be best explained by a common cause – a feature that could make people more likely to complete treatment and to find work). Analysis of Florida’s pilot drug testing program concluded that 17 neither an indication of dependence on the psychological screening tool, nor testing positive to a urinalysis, made much difference to the likelihood of employment or earnings (Crew and Davis 2003, 48-52; 2006, 76). The idea that drug treatment could improve employability is given some credence by the fact that drug dependence is a recognized factor in the literature on barriers to employment (Atkinson et al 2001; Taylor and Barusch 2004; Pilkinton 2010). In one study of long-term welfare recipients, about 20% were indicated by a questionnaire to have possible drug or alcohol problems (Taylor and Barusch 2004). Nam’s (2005) analysis of welfare exits shows that substance dependence significantly decreased the likelihood of a welfare recipient finding work. Of course, a correlation between drug dependence and unemployment leaves open what the direction of causation is, or whether there is a common cause (such as earlier life trauma). Other evidence suggests that the causal relationships are complex. Several studies report that long-term welfare recipients experience multiple barriers to employment. Among participants in Taylor and Barusch’s study, 35% reported a physical health problem; 23% had possible learning disabilities; 32% had not finished high school; 31% had a child with a medical condition, disability, learning disability or mental health condition; up to 57% had clinical levels of depression; and 36% had missed work due to domestic violence (2004, 180). A study in Michigan reported that 25% of TANF recipients had major depression, 15% had PTSD, and 7% had generalized anxiety (in Stromwall, 111; see also Morgenstern 2008; Nam 2005). Further, those experiencing barriers to employment are likely to experience more than one barrier. In Taylor and Barusch’s study, 8% of the sample had experienced none of the barriers being investigated, 13% had one barrier, and 57% had 2-4 barriers (2004, 181). Atkinson et al report that welfare recipients who use drugs are more likely than those who do not to report various 18 other barriers to employment (2001). Both welfare seeking and drug use are correlated with poverty, low levels of education, and trauma (Pollack et al 2002b, 261). While drug dependence may make it more difficult to gain employment, its relationship with employability is likely to be causally complex, with each of these variables related to various common causes as well as to each other. This indicates the presence of a group within the population of welfare recipients whom we class as ‘polydysfunctional’, who experience drug dependence problems, but also a range of other problems. In relation to increasing the employability of non-dependent drug users by reducing use, we have not identified any directly relevant studies. That some proportion of the working population also uses drugs indicates that drug use is not always a barrier to working. This may differ depending on the kind of work, and the kind of drug taken. However, if employers undertake pre-employment drug tests, or otherwise suspect an applicant uses drugs, drug use might be a barrier to being offered work, if not to undertaking work. There is conflicting evidence relating to the link between reducing drug use and protecting children. A correlation has been shown between child maltreatment and drug use, particularly dependent use (Kepple 2017) – though neither drug use nor dependence necessarily lead to child maltreatment or even reduced parenting capacity (Kepple 2017; Testa and Smith 2009). Research into whether parental drug treatment has positive outcomes for children has conflicting results. Several studies comparing parents in contact with child services receiving and not receiving drug treatment found no significant differences in terms of later child maltreatment allegations, and two studies reported an increased probability of later allegations (Testa and Smith 2009, 155-6). This could reflect inadequacies in the treatment programs themselves. The prima facie causal link is also challenged by recognizing that the population of drug-dependent 19 parents with children at risk of maltreatment also typically experience a number of other risk factors for child maltreatment, e.g., low level of education, poverty, domestic violence, and alcohol dependence (Testa and Smith 2009). They may fall into the polydysfunctional group, suggesting that drug treatment, even where successful at reducing drug use, could be limited in its effects for protecting children. There are well-established correlations between drug use and criminal activity (even excluding drug use itself as a criminalized activity). A recent review summarizes: Psychopharmacological properties of certain substances directly increase individual risks of violent behavior and victimization. […] Illicit drugs are costly, and this is a key driver of acquisitive crime. (Pollack 2017, 164) There is also some reason to think that drug use can be a cause of crimes, in the senses that the intention to use drugs may motivate acquisitive crime to finance drug use, and that intoxication (with certain drugs such as ice) could make a causal contribution to crime through heightening recklessness or aggression (Reuter and MacCoun 2001, 21-2). An Australian study of police detainees found that 19% self-attributed the offence to drugs (excluding alcohol), primarily for one of these reasons (Payne and Gaffney 2012). There is also evidence that drug treatment can reduce crime, though limitations on treatment effectiveness noted above apply (Pollack 2017, 166). It seems safe to assume without evidence that reduced drug use among welfare recipients would lead to reduced spending of welfare payments on drugs, and that being excluded from receiving benefits would also achieve this aim. Finally, while it might be expected that a drug testing program could save the state money 20 by excluding people testing positive from benefits, in fact the evidence indicates that programs typically increase government spending overall. Drug testing programs vary considerably in cost, reflecting variations in how the programs are run. Drug tests themselves differ in cost depending on the type of test. Programs will also involve delivery costs, such as staff training, staff time diverted into test administration, and the costs of repeat tests. Where the program is linked to treatment requirements, states have needed to increase treatment funding. Carley (2012) estimates that for a program with a 6-month exclusion after a positive test to save money, between 1.9 and 3.8% of welfare recipients would need to test positive (with the range reflecting differences in program procedures). This estimate assumes that payments for child upkeep would also be removed; if those payments were continued, the program would have “minimal impact” on caseload or savings (Carley 2012, 4). Carley’s estimate also excludes the costs of increased treatment funding, but notes in its review of other state programs that this is their major cost. The US Department of Human Services review of drug testing programs reports that “[n]one of the State cost estimates identified for this paper showed net savings resulting from proposed drug testing programs” (DHHS 2011, 7). 4.2. Costs of Drug Testing Programs We have already noted that drug testing welfare recipients is likely to have net financial costs rather than savings. In addition, there are ethical ‘costs’, including infringing the rights of welfare recipients, and other potential harms. As noted above, legal scholars have debated whether and how an infringement of the rights of welfare recipients can be justified, in terms of the Fourth Amendment’s provisions for individualized suspicion or a governmental special need (e.g., Perez-Munoz 2017; Bolen 2014; 21 Schaberg 2012; Newell 2012). 10 This approach results from the recognition that the drug test is a search, and thus involves some infringement on privacy rights. Part of consideration of the justification of this infringement is its nature and extent. A drug test could infringe either bodily or informational privacy. With regard to bodily privacy, different kinds of tests have different levels of intrusiveness. Blood testing, though the gold standard for detection of drugs or drug metabolites in the body, is highly invasive and generally not proposed for drug testing programs. Urine tests, where samples are collected with supervision (which may be judged necessary to prevent sample substitution or manipulation), are highly invasive; sweat, hair, and oral fluid tests are less so (Pidd and Roche 2011). With regard to information privacy, drug testing would require disclosure of personal information for some, such as prescription medications or health status which could interfere with the test. We shall not attempt to provide a general defence of privacy as a right. We assume for the purposes of this argument that privacy has some importance, such that measures to protect it have some ethical weight, but that it can be legitimately infringed upon in some situations and for some purposes. This is consistent with our aim of providing an argument that does not rely on potentially contentious normative commitments, and is broadly consistent with current legal treatment of privacy in liberal democracies. Thus, the privacy infringement will be counted as one cost of drug testing programs to be considered in assessing whether they can meet the three criteria, in the following section. Other than the loss of privacy, the process of drug testing may not impose a particularly high burden on welfare recipients in its tangible aspects: the time to take a test is not long, and methods used are not painful. There is evidence that being subjected to a drug test can be psychologically burdensome: it is experienced as degrading or humiliating (Wincup 2014; Perez- 22 Munoz 2017). Some have argued that it is particularly ethically problematic to impose a degrading experience on a group whose members are often already disadvantaged (in PerezMunoz 2017, 7). It has also been suggested that punitive measures related to welfare receipt can reduce people’s confidence and self-efficacy, which might contribute to lowering their employability (Murphy et al 2011; Perez-Munoz 2017). These effects could occur for the general population of welfare recipients, i.e., a larger population than those using drugs. In addition, since the burden of drug testing will be borne by a group that is the most socioeconomically disadvantaged, it may enhance existing inequalities, such as racial inequalities. In the US, for example, there is evidence that Black and Hispanic people are more likely to experience poverty (and a number of its determinants, such as lower levels of education), and are disproportionately enrolled in welfare schemes (McDaniel 2017, 22). 11 Insofar as drug testing constitutes an additional burden on those tested, it adds to the existing inequalities that this represents.12 Another set of concerns relates to the effects of welfare exclusion. No studies have been identified on the effects of removing welfare payments on the basis of a positive drug test. However, research on welfare sanctions shows that they can increase mental health problems, economic hardship (measured by utility shutoffs, food insecurity, lack of medical care, and homelessness or eviction), and the risk of child maltreatment – though they may also increase compliance and welfare exits (Wu et al 2015, 2-3). Several studies examined effects on drugdependent people who lost their eligibility for benefits after dependence was declassified as a ‘disability’ for the purposes of some US welfare payments. This group reportedly later experienced worse psychological comorbidities, but no reduction in drug use (in Watkins and Podus 2000). There was also an increase in arrests (Chatterji and Meara 2010, 239), and “one 23 year postdisenrollment, about half of the recipients reported monthly legal earnings below $500 and received no cash public aid” (in Pollack et al 2002b, 270). There is some evidence that welfare sanctions can increase risks to children of hospitalisation and food insecurity, but also that there are potential benefits for children if drug use is actually reduced (DHHS 2011, 8). Other possible societal costs are an increase in crime, and increased financial pressure on families and associates, as excluded people seek other sources of income. Other concerns relate to referring people to treatment in response to a positive test. First, there is an ethical concern about making welfare receipt conditional on attending drug treatment. Since people in this situation may lack other options, this can be construed as coerced treatment and so conflict with a generally recognized right to refuse medical treatment (insofar as drug dependence treatment counts as ‘medical’). 13 The acceptability of coerced or mandatory drug treatment (linked for instance to criminal justice procedures) differs greatly by jurisdiction, however, and is somewhat normalized in some locations. Second, some have worried that drug testing programs could inflate the population of people in drug treatment beyond those who really need it. Even among those who use drugs in ways consistent with criteria for dependence, not everyone needs treatment, and many ‘age out’ of such use (Pollack 2017, 166). Increasing the treatment population could thus waste resources. In drug treatment programs that are grouprun, it could also alter the nature of the treatment experience for those who could benefit from treatment, altering the effectiveness for this group. Finally, some suggest that there are (non-financial) societal costs as drug testing programs encourage stereotyping of welfare recipients as drug users, and contribute to a public discourse that is derogatory and disparaging about people on welfare, portraying them as undeserving of social supports (Amundson et al 2015). As a result, drug testing could contribute to 24 marginalisation and disconnection of unemployed people (Wodak 2017), and damage welfare recipients’ job prospects (Crew and Davis 2003, 41). The implication that unemployment is caused by individual rather than systemic factors could also lead to neglecting other barriers to employment (Berger 2001). Such effects on societal attitudes may have follow-on effects that exacerbate existing inequities. There are a number of studies showing that amongst welfare recipients, Black and Hispanic people are more likely than whites to be sanctioned, particularly where sanctions are applied at the discretion of case workers, demonstrating racial prejudices (McDaniel 2017, 23-5). Drug testing programs play into and reinforce these harmful prejudices (Ledford 2016). 4.3. Are There Other Potential Methods of Meeting the Aims? There are several alternative possibilities for meeting the aims of drug testing programs. Here we note just a few options by way of example. There are alternative methods for identifying people with drug dependence problems. Prior to the introduction of drug testing programs, some US states screened welfare recipients using questionnaires administered by welfare caseworkers, with those indicated to have a potential problem being referred for assessment by a professional (Morgenstern 2001). There was some dissatisfaction with such systems because they resulted in treatment referral rates of around 1-5%, perceived to be much lower than the rate of welfare recipients with drug problems (Henderson et al 2006, 218). However, the rates of people testing positive on drug tests are not higher. Indeed, if the aim is to identify drug dependent rather than recreational users, psychological screening may be a more accurate measure. The psychological screening tool utilized as part of some drug testing programs (the Substance Abuse Subtle Screening Inventory 25 or SASSI) has tested, in its most recent form, as having a sensitivity of 93% and specificity of 90% (Lazowski and Geary 2016). 14 This would also improve accuracy in terms of capturing people dependent on substances that are not tested for (including alcohol, dependence on which affects more people, and which is also statistically linked to lower levels of income, child maltreatment, and crime). Though the accuracy of drug testing welfare recipients will depend on a program is run, it is likely to be much lower than this (Pollack et al 2002a, 28). Another option is the incorporation of ‘brief interventions’ (BIs) at welfare centres. BIs involve a range of strategies that can be adapted for use in different settings. BIs utilize ‘motivational interviewing’ techniques, with a focus on encouraging people to regard their substance use as not worth its costs, and themselves as capable of changing their behavior. Measures include short, 5-minute advice, intensive counselling sessions, and provision of selfhelp information (Kumar and Malhotra 2000). There is extensive evidence for their effectiveness and cost-effectiveness for alcohol use, and some evidence for its effectiveness in relation to other drugs, though it may be more effective where problems related to the drug use are mild to moderate (Bray et al 2017; Parmar and Sarkar 2017). There are also alternative methods to meet the aim of preventing welfare payments from being spent on illicit drugs. For instance, governments might provide support via methods other than cash payments. Examples are the USA’s SNAP, or the ‘cashless welfare card’ used in some areas of Australia, which provides credit that cannot be spent on alcohol or gambling, or to obtain cash (Australian Government DSS 2017). 5. PROPORTIONALITY ANALYSIS We have proposed that for a program of drug testing welfare recipients to be ethically 26 acceptable, it must be reasonably likely to meet its aim(s), not impose costs out of proportion to the importance of its aim(s), and not be substantially more costly than feasible alternatives. For brevity, we will refer to these below as success, proportionality, and necessity criteria. Since achieving reductions in use via treatment referral and deterrence, and achieving other aims via exclusion from benefits, seem mutually exclusive, we assess the aims in two groups, first in relation to the aim of reducing drug use (and its various further aims), and then in relation to aims served by welfare exclusion. The analyses also make clear the necessity of distinguishing between two quite different populations of people who use drugs, for whom the effects of these policies may be radically different. 5.1. Reducing Drug Use We begin with the success criterion. People using drugs recreationally would not be affected by a treatment referral. The probability of recreational users being deterred from using by drug testing is not known. If drug tests are given at predictable times, such as on application, deterrence might not be lasting. Random scheduling of drug tests might be more effective as a deterrent, but does not appear to be used in any current programs, and may be practically difficult to implement given the need to establish reasonable suspicion and the short detection window for most drugs. Recreational use seems the least likely kind to impact on employability, and recreational users are less likely to be implicated in child maltreatment cases or involved in crime. They perhaps spend welfare payments on illegal drugs, but would be less likely to do so in high amounts, or where this would divert funds needed for basic necessities. Thus, while there may be some deterrence, overall it seems unlikely that drug testing programs will greatly reduce recreational drug use, and so achieve the four further aims to a significant extent. 27 People who are dependent on drugs are unlikely to achieve reduced use via deterrence. Drug dependence is known to change decision-making surrounding drug-taking and is a chronic relapsing condition. While dependent people appear to retain some measure of choice over their use, an intermittent or one-off drug test is unlikely to impact much on overall use. Some proportion of drug-dependent welfare recipients might achieve reduced drug use through treatment. The evidence does not provide clear guidance on whether they are thereby more likely to gain employment, and there are reasons to doubt this, especially if other employment barriers are unaddressed. We also lack good evidence that treatment benefits children, and there are again reasons to doubt it will have much effect if other issues are not addressed. The evidence-base surrounding drugs and crime indicates that some reductions could be expected to follow reductions in use, though it is not known how many drug-dependent people receiving welfare are also involved in crime. These would also reduce the amount of welfare money being spent on drugs. Noting some lack of evidence, we will allow for the sake of argument that there might be effects for increased employability and improvements for children, and consider that a drug testing program could meet the success condition in relation to these four aims, for dependent users, to some extent. However, this is unlikely to be a very great extent, given what is known about the low-to-high-moderate effectiveness of treatment for reducing drug use (Dutra et al 2008), and the low rate of positives detected by drug testing programs. In terms of the aim of reducing drug use (and its further aims), then, programs seem more likely to meet the success criterion via treatment than deterrence. We will therefore focus on this mechanism in the remainder of this assessment, in order to follow the principle of charity. We then need to assess whether such an outcome is sufficiently important to justify the various costs of the program. These include: the broader population of welfare recipients have their privacy 28 infringed and might be negatively impacted; people testing positive are enrolled in treatment that could be considered coerced and may be irrelevant for them; financial costs (quite high given the need for additional treatment investment); and the reinforcement of harmful stereotypes, including in some locations racial stereotypes. Though these various problems would be difficult to quantify, it seems implausible that these costs would be compensated for by reductions in drug use among a small proportion of welfare recipients, with some (partly uncertain) ensuing benefits. For this end to justify such costs, it would need to be considered highly important. Some might consider it so, perhaps due to holding a ‘zero-tolerance’ approach to drug use or considering drug use intrinsically unethical. While some do hold such a view, it is difficult to justify once its basis made explicit: it appears to rely on premises that one’s mind should not be manipulated, and/or that there is something intrinsically wrong with seeking pleasure through substances, but in general these are not held by those arguing that drug use is itself unethical, or are not held consistently (Husak 2004, 407-11). While some reduction in drug use among welfare recipients may have some value, then, it is implausible to consider it to have sufficient value to justify its costs and thus it will not meet the proportionality criterion for this aim. It is also unlikely that a program could meet the necessity criterion for this aim. There are alternative methods for the identification of people with drug dependence, which is the kind of use most relevant to this aim. These could avoid some of the costs (enrolling non-dependent people in inappropriate treatment, and many of the financial costs), while having similar effects for drug use reduction and ensuing benefits. There may be other policy options possible to better further children’s interests. Since families with at-risk children and drug-dependent parents are likely part of a polydysfunctional group, they are likely to come to the attention of relevant 29 authorities through other means than drug tests performed by welfare agencies – and are likely to need a broader set of interventions if outcomes are to be improved. The aim of reducing use in order to increase employability, protect children, reduce drugrelated crime, and prevent welfare money from being spent on drugs, is thus largely irrelevant to recreational drug users, either by deterrence or treatment. For dependent users, available evidence indicates it might meet the success criterion via treatment, although perhaps not to a great degree, and it is unlikely to meet proportionality or necessity criteria. 5.2. Exclude People Who Use Drugs from Welfare The aim of excluding people who use drugs from welfare – as a means to stated aims of preventing welfare payments being spent on drugs, and reducing government spending – could be achieved by a drug testing program. Such a program might either test all recipients, or use pre-screening, excluding those who test positive or refuse a test. The former case would exclude some recreational as well as some dependent drug users, while the latter would be likely to exclude mainly dependent users (assuming the screening tool is reasonably accurate). Though some drug users would likely test negative (if they have not used recently, or have used drugs not tested for), either of these methods would reduce the amount of welfare payments being spent on illicit drugs to some degree. This may or may not meet the aim of saving state money overall, depending on program details such as rules relating to payments intended to support children. For the sake of argument, we grant that this is possible, and will count programs which exclude recipients who test positive as meeting the success condition for its further aims. To meet the proportionality criterion, meeting these aims would need to be regarded as highly important, as they need to outweigh high costs. Importantly, exclusion from benefits will 30 reduce people’s capacities to meet their basic needs. The evidence indicates that this is likely to exacerbate mental health problems and economic hardship, both for the welfare recipients themselves and their dependants and associates. In addition, it may increase crime, particularly acquisitive crime, as people seek alternative means of support (Pollack 2017); could have negative effects for children (Wu et al 2015); and, insofar as disadvantaged racial groups are over-represented among those undergoing testing, could worsen existing socioeconomic inequalities. In addition, non-drug-using welfare recipients would also have to undergo a rights infringement; damaging stereotypes about welfare recipients, disadvantaged minorities, and drug users would be encouraged; and marginalisation of drug users would be perpetuated. We do not consider the aim of preventing welfare payments being spent on drugs to be sufficiently important that meeting it would compensate for all of these costs. It does not seem plausible to us that preventing a person from spending welfare payments on illicit drugs is more important than that person, and others, being able to access basic necessities – and there would be more costs than this. Nor can aims concerning government spending be plausibly considered to outweigh these negative effects, particularly considering that they are likely to increase financial costs in other areas in any case (the costs caused by increased crime are likely to be particularly high, especially in locations with high rates of drug-related incarceration) (Pollack 2017, 168-9; 174-5). Accordingly, drug testing programs that exclude people testing positive from payments will not meet the proportionality criterion. We recognize that, again, some might consider the aim of preventing welfare payments being spent on illicit drugs to be extremely important, again perhaps due to beliefs held about the ethical status of drug use, or to beliefs about the obligations of citizens receiving government assistance. Thus some might regard this aim to be sufficiently important that meeting it would 31 compensate for all these costs, and so take drug testing for this purpose to meet the proportionality criterion. However, even if this possibility is granted, following a principle of charity, such a program would be unlikely to meet the necessity criterion. In relation to this aim, alternative methods for identifying people who are drug-dependent may not be relevant, if the aim applies to spending on recreational drug use as well as dependent use. However, there are alternative methods available to prevent spending of welfare payments on illicit drugs, such as cashless support. While there are also various objections to this form of welfare to be considered, any costs associated with it may be less than the costs of drug testing combined with exclusion from welfare. Thus drug testing to exclude people from welfare payments does not meet the necessity criterion. 6. DRUG TESTING OF WELFARE RECIPIENTS IS NOT ETHICALLY ACCEPTABLE POLICY We conclude that programs of drug testing welfare recipients, however their aims are construed, are unlikely to be able to meet those aims, or where they do meet them, will not be able to do so in ways that meet proportionality and necessity criteria. As such, programs of drug testing welfare recipients are not ethically acceptable. Our analysis suggests a number of general points that often remain implicit in discussion of drug testing welfare recipients, contributing to lack of clarity in both public and academic debate. Programs are likely to affect different populations of drug users in quite different ways. There is little point targeting recreational users with drug testing, as this will not have great effects for any of the aims of programs, and will have some costs. A smaller, dependent group could 32 possibly derive some benefits from drug testing linked to treatment interventions. However, members of this group, many of whom will be polydysfunctional, are likely to come to the attention of authorities in other ways, and often face a complex set of difficulties which require a broader set of interventions. REFERENCES Alexander M. 2010. The New Jim Crow: Mass Incarceration in the Age of Colorblindness. New York: The New Press. Australian Associated Press (AAP). 2017. Drug test trial for welfare recipients ‘all about love’, says Turnbull. SBS News, 22 August 2017, https://www.sbs.com.au/news/article/2017/08/22/drug-test-trial-welfare-recipients-all-aboutlove-says-turnbull, accessed 13 December 2017. Amundson K, Zajicek AM, Kerr B. 2015. A social metamorphosis: Constructing drug addicts from the poor. Sociological Spectrum 35(5):442-464. Atkinson J, Lee D, Dayton-Shotts C, French C. 2001. Self-perceived job skills and employment barriers among non-drug using and chronic drug using welfare-to-work participants. Journal of Drug Issues 31(3):747-756. Australian Government Department of Social Services (DSS). 2017. Cashless debit card – Overview. https://www.dss.gov.au/families-and-children/programmes-services/welfareconditionality/cashless-debit-card-overview, accessed 13 December 2017. Berger PB. 2001. Science misapplied: mandatory addiction screening and treatment for welfare 33 recipients in Ontario. Canadian Medical Association Journal 165(4):443-4. Bloom D, Loprest PJ, Zedlewski SR. 2011. TANF recipients with barriers to employment. Office of Planning, Research and Evaluation, Urban Institute, Brief #1. Available at https://www.urban.org/research/publication/tanf-recipients-barriers-employment, 13 Nov 2017. Bolen J. 2014. Screening the poor: The legality of drug testing for welfare benefits. Richmond Journal of Law and the Public Interest 18(1):77-88. Bray JW, Del Boca FK, McRee BG, Hayashi SW, Babor TF. 2017. Screening, brief intervention and referral to treatment (SBIRT): rationale, program overview and cross-site evaluation. Addiction 112(Suppl. 2):3-11. Carley F. 2012. Drug testing welfare recipients: a review of potential costs and savings. State Notes: Topics of Legislative Interest. Michigan Senate Fiscal Agency, Summer 2012. http://www.senate.michigan.gov/sfa/publications/notes/2012notes/notessum12fc.pdf, accessed 13 Nov 2017. Chandler K. 2014. Alabama lawmakers approve welfare drug testing. The Washington Times, 3 April 2014. https://www.washingtontimes.com/news/2014/apr/3/alabama-lawmakersapprove-welfare-drug-testing/, accessed 13 December 2017. Chatterji P, Meara E. 2010. Consequences of eliminating federal disability benefits for substance abusers. Journal of Health Economics 29:226-240. Childress, JF, et al. 2002. Public health ethics: mapping the terrain. Journal of Law, Medicine and Ethics 30:170-178. Cianciardo J. 2010. The principle of proportionality: The challenges of human rights. Journal of Civil Law Studies 3(1):177-186. 34 Crew RE, Davis BC. 2003. Assessing the effects of substance abuse among applicants for TANF benefits. Journal of Health & Social Policy 17(1):39-53. Crew RE, Davis BC. 2006. Substance abuse as a barrier to employment of welfare recipients. Journal of Policy Practice 5(4):69-82. Department of Health and Human Services (DHHS). 2011. Drug Testing Welfare Recipients: Recent proposals and continuing controversies. ASPE Issue Brief, US Department of Health and Human Services. https://aspe.hhs.gov/basic-report/drug-testing-welfare-recipientsrecent-proposals-and-continuing-controversies, 13 Nov 2017. Dutra L, et al. 2008. A meta-analytic review of psychosocial interventions for substance use disorders. Am J Psychiatry 165:179-87. Fisher D. 2014. Minister claims low drug result as victory. New Zealand Herald, 11 January 2014, http://www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11184479, accessed 13 December 2017. Forsyth M, Banham C. 2017. Drug testing welfare recipients is not about love, Malcolm Turnbull, it’s about punishment. The Guardian Australia, 11 May 2017. https://www.theguardian.com/commentisfree/2017/may/11/drug-testing-welfare-recipientsis-not-about-love-malcolm-turnbull-its-about-punishment, accessed 31 January 2018. Grant BF. 1996. Prevalence and correlates of drug use and DSM-IV drug dependence in the United States: Results of the National Longitudinal Alcohol Epidemiologic Survey. Journal of Substance Abuse 8(2):195-210. Hall R. 2016. Drug testing and public assistance. CLASP: Policy solutions that work for lowincome people. Washington, DC. https://www.clasp.org/sites/default/files/publications/2017/04/2016.02.04-Drug-Testing- 35 and-Public-Assistance-Brief-FINAL.pdf, accessed 13 Nov 2017. Henderson S, Dohan D, Schmidt LA. 2006. Barriers to identifying substance abuse in the reformed welfare system. Social Service Review 80(2):217-238. Hermeren G. 2012. The principle of proportionality revisited: interpretations and applications. Medicine, Health Care and Philosophy 15:373-82. Husak DN. 2004. The moral relevance of addiction. Substance Use & Misuse 39(3):399-436. Kass NE. 2001. An ethics framework for public health. American Journal of Public Health 91(11):1776-1782. Kayanan M. 2013. Shaping a legal challenge to the drug testing of welfare applicants. Clearinghouse REVIEW Journal of Poverty Law and Policy 47:114-118. Kepple NJ. 2017. The complex nature of parental substance use: Examining past year and prior use behaviors as correlates of child maltreatment frequency. Substance Use & Misuse 52(6):811-821. Kumar S, Malhotra A. 2000. Brief interventions in substance use. Indian Journal of Psychiatry 42(2):172-183. Lazowski LE, Geary BB. Validation of the Adult Substance Abuse Subtle Screening Inventory4(SASSI-4). European Journal of Psychological Assessment (online first). https://doi.org/10.1027/1015-5759/a000359. Ledford C. 2016. Symbolic racism, institutional bias, and welfare drug testing legislation: racial biases matter. Policy Studies Journal (online ahead of print). DOI:10.1111/psj.12192 Metsch LR, McCoy CB, Miller M, McAnany H, Pereyra M. 1999. Moving substance-abusing women from welfare to work. Journal of Public Health Policy 20(1): 36-55. 36 MacCoun RJ, Reuter P. 2001. Drug War Heresies: Learning from other Vices, Times and Places. Cambridge: Cambridge University Press. McDaniel M, Woods T, Pratt E, Simms M. 2017. Identifying racial and ethnic disparities in human services. Urban Institute, Washington, DC. https://www.urban.org/research/publication/identifying-racial-and-ethnic-disparities-humanservices/view/full_report, accessed 22 December 2017. McLaughlin BW. 2013. Drug testing, welfare, and the special needs doctrine: An argument in support of drug testing TANF recipients. Cleveland State Law Review 61:567-98. Morgenstern J, et al. 2001. Specialized Screening Approaches Can Substantially Increase the Identification of Substance Abuse Problems Among Welfare Recipients. Research Notes. Office of the Assistant Secretary for Planning and Evaluation, US Department of Health and Human Services. https://aspe.hhs.gov/report/specialized-screening-approaches-cansubstantially-increase-identification-substance-abuse-problems-among-welfare-recipients, 14 Nov 2017. Morgenstern J, Hogue A, Dasaro C, Kuerbis A, Dauber S. 2008. Characteristics of individuals screening positive for substance use in a welfare setting: Implications for welfare and substance-use disorders treatment systems. Journal of Studies on Alcohol and Drugs 69(4):561-70. Murphy J, Murray S, Chalmers J, Martin S, Marston G. 2011. Half a Citizen: Life on Welfare in Australia. Allen and Unwin, Sydney. Nam Y. 2005. The roles of employment barriers in welfare exits and reentries after welfare reform: Event history analysis. Social Service Review 79(2):268-93. National College of State Legislatures (NCSL). 2017. Drug testing for welfare recipients and 37 public assistance. 24 March 2017. http://www.ncsl.org/research/human-services/drugtesting-and-public-assistance.aspx, accessed 14 Nov 2017. New Zealand Ministry of Social Development. 2016. FOI requests. Available at https://www.msd.govt.nz/about-msd-and-our-work/publications-resources/officialinformation-responses/2016/index.html, accessed 14 Nov 2017. New Zealand Ministry of Social Development. 2018. Obligations. https://www.workandincome.govt.nz/on-a-benefit/obligations/index.html, accessed 30 January 2018. Newell W. 2011. Tax dollars earmarked for drugs – The policy and constitutionality of drug testing welfare recipients. Columbia Human Rights Law Review 43:215-254. Oosting J. 2014. Welfare drug testing bills win final approval in Michigan Senate, head to governor’s desk. Mlive, 10 December 2014, http://www.mlive.com/lansingnews/index.ssf/2014/12/welfare_drug_testing_bills_win.html, accessed 13 December 2017. Payne J, Gaffney A. 2012. How much crime is drug or alcohol related? Self-reported attributions of police detainees. Trends and Issues in Crime and Criminal Justice 439. Australian Institute of Criminology, Australian Government. http://www.aic.gov.au/publications/current%20series/tandi/421-440/tandi439.html, accessed 13 Dec 2017. Parliament of the Commonwealth of Australia. 2017. Senate Social Services Legislation Amendment (Welfare Reform) Bill 2017 - Revised Explanatory Memorandum. https://www.aph.gov.au/Parliamentary_Business/Bills_Legislation/Bills_Search_Results/Re sult?bId=r5927, accessed 14 Nov 2017. Parmar A, Sarkar S. 2017. Brief interventions for cannabis use disorders: A review. Addictive 38 Disorders and Their Treatment 16(2):80-93. Perez-Munoz C. 2017. What is wrong with testing welfare recipients for drug use? Political Studies (online ahead of print). Pidd K, Roche A. 2011. Workplace Drug Testing: Evidence and Issues. National Centre for Education and Training on Addiction, Flinders University, Adelaide. http://nceta.flinders.edu.au/files/5813/2200/5230/Workplace%20Drug%20Testings%20ml% 202.pdf, accessed 13 Dec 2017. Pilkinton M. 2010. TANF recipients' barriers to employability: substance abuse and domestic violence. Journal of Human Behavior in the Social Environment 20(8):1011-1023. Pollack HA. 2013. States want drug tests for welfare recipients. That’s a terrible idea. Washington Post, 5 June 2013. https://www.washingtonpost.com/news/wonk/wp/2013/06/05/states-want-drug-tests-forwelfare-recipients-thats-a-terrible-idea/?utm_term=.887d7ed8156a, accessed 31 January 2018. Pollack HA, Danziger S, Jayakody R, Seefeldt KS. 2002a. Drug testing welfare recipients—false positives, false negatives, unanticipated opportunities. Women’s Health Issues 12(1):23-31. Pollack HA, Danziger S, Seefeldt KS, Jayakody R. 2002b. Substance use among welfare recipients: trends and policy responses. Social Service Review 76(2):256-274. Pollack HA. 2017. Dealing more effectively with problematic substance use and crime. Crime & Justice 46(1):159-200. Price ML. 2013. Utah’s welfare drug testing saved more than $350,000 in first year, officials say. Deseret News, 8 September 2013, https://www.deseretnews.com/article/765637435/Utahofficials-say-welfare-drug-tests-save-money.html, accessed 13 December 2017. 39 Rodin D. 2011. Justifying harm. Ethics 122(1):74-110. Schaberg AE. 2012. State drug testing requirements for welfare recipients: Are Missouri and Florida’s new laws constitutional? Missouri Law Review 77:567-89. Stromwall LK. 2002. Mental health barriers to employment for TANF recipients. Journal of Poverty 6(3):109-120. Taylor MJ, Barusch AS. 2004. Personal, family, and multiple barriers of long-term welfare recipients. Social Work 49(2):175-83. Terer K, Brown R. 2014. Effective drink driving prevention and enforcement strategies. Trends and Issues in Crime and Criminal Justice 472. Australian Institute of Criminology, Australian Government. http://www.aic.gov.au/media_library/publications/tandi_pdf/tandi472.pdf, accessed December 14 2017. Testa MF, Smith B. 2008. Prevention and drug treatment. The Future of Children 19(2):147-168. USA.gov. 2018. Government benefits. https://www.usa.gov/benefits, accessed 30 Jan 2018. Walzer M. 1977. Just and Unjust Wars. Basic Books, New York. Watkins KE, Podus D. 2000. The impact of terminating disability benefits for substance abusers on susbtance use and treatment participation. Psychiatric Services 51(11):1371-1372;1381. Werb D, et al. 2015. The effectiveness of compulsory drug treatment: A systematic review. International Journal of Drug Policy 28:1-9. Widelitz J. 2011. Florida’s suspicionless drug testing of welfare applicants. Nova Law Review 36:253-398. Wincup E. 2014. Thoroughfares, crossroads and cul-de-sacs: Drug testing of welfare recipients. International Journal of Drug Policy 25:1031:1037. 40 Wodak A. 2017. People who think punitive measure help drug addicts haven’t seen what I have. The Guardian, 22 August 2017, https://www.theguardian.com/commentisfree/2017/aug/22/people-who-think-punitivemeasures-help-drug-addicts-havent-seen-what-i-have, accessed 14 December 2017. Wu C-F, Cancian M, Wallace G. 2014. The effect of welfare sanctions on TANF exits and employment. Children and Youth Services Review 36:1-14. 1 Since most existing and proposed programs focus on income support payments made to working-age people who are unemployed (or under-employed), our argument will focus on this group. Any future proposals to drug-test recipients of other welfare payments, particularly payments that support those unable to work, may suffer from similar problems to those discussed below, but will also raise other complex issues. Since there have thus far been no proposals to drug test recipients of such payments (that we are aware of) we leave these aside in the current paper, though we expect that if our argument against drug testing succeeds for a potentially work-engaged population, it would also rule out drug testing of those unable to work. 2 And/or on some feature of welfare receipt that makes welfare recipients liable to bear some burdens. As we argue below, this can be incorporated into the proportionality approach. 3 For detail on the series of legal decisions and appeals related to these programs (and analysis) see, e.g., Bolen (2014); McLaughlin (2013); Newell (2012). 4 Details of TANF programs differ by state. SNAP (‘food stamps’) provides low-income families or individuals with a debit card which can be used to buy food. Medicaid provides healthcare to low-income families and individuals, with eligibility and types of care covered differing by state (USA.gov 2018). Other countries that have considered drug testing programs all have universal healthcare systems, do not distinguish between contributory and non-contributory government assistance, and have only considered drug testing for recipients of unemployment benefits. 5 This applies to recipients of income support payments who have work obligations, and so is relevant to New Zealand’s Jobseeker and Sole Parent Support payments, both means-tested payments to support working-age people who are under- or unemployed. See New Zealand Ministry of Social Development (2018). 6 Some have claimed that the threat of exclusion from benefits could motivate people to stop using drugs and find work (Widelitz 2011, 303). While it may have some such deterrent effect, this seems unrealistic for people who are drug dependent, and the instability involved in lacking a basic income is unlikely to be conducive to finding work. 7 As a separate consideration, politicians’ aims of proposing drug testing programs to appeal to voters may also be ethically suspect, though perhaps not as clearly so, since some political 41 dissimulation may be expected by voters or normalized (and politicians may also genuinely share their electors’ views). But even if a political aim of appealing to voters is ethically permissible, at least in some cases, it is hardly an aim of sufficient value to justify any of the costs or harms of a particular policy. 8 It is worth making explicit that we are therefore treating these unstated aims as irrelevant to the ethical assessment of drug testing welfare recipients for the purposes of the proportionality analysis, even though the proportionality analysis is largely built around an understanding of policy aims. This is not because we think it has no ethical relevance but rather for two other reasons. First, that (as argued below) drug testing programs cannot be ethically justified even on the basis of its stated, potentially ethically acceptable aims, making the inclusion of the unstated aims as aims unnecessary to show that the policy is unethical. Second, we are wary of contributing to further polarisation of drug policy debate. In saying this, we acknowledge that it may also be true that such unstated aims are important to recognize in ethical investigations which have a broader scope than our own, for instance in relation to broader practices of welfare, drug, and criminal justice policy. 9 Most testing methods have a quite short detection window of one to several days (with a longer detection window for cannabis) (Pidd and Roche 2011). People who are dependent might use less frequently than this, or be dependent on alcohol, or illicit drugs that are not tested for. Any program will also need a way to distinguish illicit from prescription use, including opioid painkillers, prescription methadone, and medicinal cannabis in relevant jurisdictions. This issue arose for some tested during the overturned Florida program (Kayanan 2013). Tests can also be inaccurate due to medical conditions, such as kidney problems (DHHS 2011, 4). 10 US Courts have also considered drug testing justified in some situations by implied consent (e.g., in voluntary school sports programs) and by a ‘diminished expectation of privacy’ attached to some kinds of government employment (e.g., customs officials) (Newell 2012). 11 Rates of disadvantaged group members who are subjected to drug tests could be even more disproportionate if people are selected for testing using some sorts of profiling, or informal discretionary mechanisms. 12 Conversely, if drug testing programs actually benefit recipients, perhaps by aiding recovery from drug dependence problems from treatment referrals, it could be regarded as an equalising measure. 13 This might depend on the type of treatment (e.g., pharmacological versus counselling), as well as whether drug dependence is considered a disease. 14 Of course, it is arguable that requiring people to undertake such tests might also constitute a privacy infringement and be similarly stigmatising. Our point here is just that if the aim of the program relates to identifying the population with drug dependence problems (rather than any drug user) it is likely to be more accurate. There may well be good reasons not to adopt it either, but these are beyond the scope of this paper. 42