Public Views About Reintegrating Child Sex Offenders via Circles of Support and Accountability (COSA): A Qualitative Analysis

ABSTRACT Circles of Support and Accountability (COSA) have recently become part of the criminal justice landscape. However, little has been documented on public views about COSA. The existing research on this topic is entirely quantitative, providing no insight into the reasons behind public support for COSA. This study addresses this gap by analyzing comments made on four online forums following the announcement of Australia’s first COSA program. Findings suggest that community education should focus on a number of key messages about COSA to harness public support for this program. Recommendations are made about the content and delivery of these messages.

policy, and in turn divert resources both from measures that better reflect public views, and from interventions that may be more effective in enhancing community safety (Mears et al. 2008:555).
By way of background, much research has been undertaken on public views about sex offender policy in general. This extensive body of research shows that the public hold punitive attitudes toward this group of offenders (Bollinger et al. 2012;Katz Schiavone et al. 2008;Katz Schiavone and Jeglic 2009;Kleban and Jeglic 2012;McCartan et al. 2015;Mears et al. 2008;Olver and Barlow 2010;Rogers and Ferguson 2011;Shackley et al. 2014;Sundt et al. 1998;Willis et al. 2013), and support punitive and exclusionary policies such as sex offender registries, community notification, preventative detention, and residency restrictions (Brown et al. 2008;Comartin et al. 2009;CSOM 2010;Katz Schiavone and Jeglic 2009;Mears et al. 2008;Thakker 2012), even in the absence of any evidence that these policies work (Levenson et al. 2007). The public supports harsher penalties for sexual than non-sexual offenders (Rogers and Ferguson 2011), and for child sex offenders than those who offend against adults (McAlinden 2007;McAvoy 2012;Mears et al. 2008;Viki et al. 2012). Despite this, the research also clearly demonstrates that the public support treatment for sex offenders (Kleban and Jeglic 2012;Levenson et al. 2007;Mears et al. 2008;Willis et al. 2010), including child sex offenders (Esser-Stuart and Skibinski 1998;, despite being doubtful about its efficacy (Katz Schiavone et al. 2008;Mancini 2014;McCartan et al. 2015;Payne et al. 2010;Sundt et al. 1998;Willis et al. 2010). As Esser-Stuart and Skibinski (1998:101) eloquently summarize, "the social response is complex" (see generally . Despite this substantial body of research literature on public opinion about other sex offender policies, very little has been documented on public opinion about COSA specifically. In fact, only three previous studies could be located, which provide any insight into this topic. In the first, as part of a larger study, Wilson et al. (2007) surveyed members of the community (n = 77) in South-Central Ontario. Once they had been informed about the COSA program in their community, 69 percent of respondents reported that they were "glad", and 62 percent that they were "'relieved", that this group of offenders received support via a COSA. Only small proportions reported negative views about the program, with 14 percent being skeptical that it would reduce reoffending, 8 percent being angry that sex offenders would receive extra support, and 3 percent feeling irritated that people would want to help these offenders ).
In the second, McAvoy (2012) used an online survey to examine public support for COSA in Ireland. She used vignettes of fictional sex offenders to ascertain whether members of the community agreed that a COSA would be a good idea. The vignettes examined the impact of different genders and ages of sexual abusers and victims across a variety of sexual offences (for example, "George is 34. He is convicted of raping a 10-year-old boy while working as a school caretaker"; "Karl is 19. He has a sexual relationship with a 15-year-old girl. He is convicted of statutory rape"). Participants in the online survey (n = 84) were asked to respond to two statements about COSA in relation to each vignette: "[The offender] should be provided with a circle of support and accountability when in the community"; and "I would volunteer to be part of [the offender's] circle of support and accountability". In all cases, a majority of respondents agreed (or strongly agreed) that the offender should be provided with a COSA. Support for a COSA was highest in relation to molestation of a 12-year-old child (75%), rape of a 10-year-old child (73.8%), rape of an adult woman (73.8%), and possession of child pornography (72.6%). Lower levels of support for COSA were found in relation to indecent exposure (61.9%), and statutory rape (59.5% and 52.3%). Despite this support, however, much smaller proportions of respondents indicated that they would volunteer in a COSA for any of the offenders in these fictional scenarios. Importantly, smaller proportions indicated that they would volunteer in a COSA for those offenders who likely pose the greatest threat to community safety. The highest proportions of respondents agreed that they would volunteer in a COSA for statutory rape offenders, while the lowest proportions agreed they would volunteer with child rape and child molestation offenders.
Most recently, Hӧing et al. (2016) used a web-based panel survey across nine countries (United Kingdom, Ireland, The Netherlands, Belgium (Flemish region), France, Spain, Latvia, Bulgaria and Hungary; total n = 1873) to assess public opinion about sex offender rehabilitation generally, and COSA specifically. Hӧing et al. found that 12.3 percent of respondents agreed with the statement "I would be interested in becoming a CoSA volunteer, if such a project was nearby" (ranging from 6.8% in Latvia to 17.3% in Bulgaria). Echoing McAvoy's findings, however, Hӧing et al. found greater support for others becoming COSA volunteers than willingness to volunteer on the part of respondents themselves. Overall, 55.5 percent of respondents indicated they would approve if a friend became a COSA volunteer. This decreased to 49.2 percent who would approve if a family member was to volunteer, and to 40.6 percent if a partner was to volunteer.
While these three studies make an important contribution toward understanding the extent of public support for COSA, none provides any qualitative insight into why members of the public support or oppose COSA. The research reported in this article begins to address this gap in the literature by exploring public opinion about COSA using qualitative data sourced from online forums. Specifically, it sought to identify the reasons that members of the public gave for supporting or opposing Australia's first COSA program, which was announced in March 2015. The program is managed by a non-government offender support organization in Adelaide, South Australia (Worthington 2015). In this context, qualitative research can provide an insight into the views and beliefs of members of the public, and thus an important complement to the existing quantitative research outlined above.
The remainder of this article is presented in four main parts. The first outlines the conceptual background to the study, and introduces the concept of "multiple publics" that provided a conceptual orientation for the study. The second provides a discussion of the methodology utilized for the current study, including data collection and analysis procedures and ethical considerations. Following this, the findings of the studythe primary reasons that members of the public support or oppose the COSA programare presented. Finally, the article discusses the implications that stem from these findings and makes a number of recommendations based on these.

Conceptual background
As is being increasingly recognized, governments create criminal justice legislation and policy not necessarily in direct response to what the public wants, but rather in response to what they imagine the public wants: "there is something of a 'comedy of errors' in which policy and practice is not based on a proper understanding of public opinion" (Allen 2002:6; see also Quinn et al. 2004). Furthermore, while research on public opinion is often presented in monolithic terms, it is most useful to governments and best able to constructively inform policy debates when "presented in a way that depicts the full panoply of public opinion" (Mears et al. 2008:555; see generally Allen 2002).
As discussed above, the research on public opinion about sex offenders shows that the public hold contradictory views about how this group of offenders should be dealt with. In contrast to literature (for example, Olver and Barlow 2010; Willis et al. 2010;Zilney and Zilney 2009) that portrays the public as homogeneous, passive recipients of sensationalistic media accounts of sex offending, this suggests that members of the public hold more heterogeneous beliefs and opinions than is often thought to be the case. This section uses this as a platform from which to propose a different conceptualization of public views about sex offendingone that attempts to account for the nuances and complexities of public views about COSA presented below.
As Kitzinger (2004) outlined, a key debate in the media studies literature exists between those who theorize members of the public to be passive recipients, even dupes, of the media, and those who (perhaps somewhat romantically) theorize members of the public as active, diverse agents and even resisters. The current study follows Kitzinger (2004) by resisting the conceptualization of the public as a homogeneous group of passive dupes, and instead considers community members "not only as 'receivers' of information but as activists and message creators" (Kitzinger 2004:158). This conceptualization is particularly relevant when analyzing data taken from social media. Indeed, the very existence of online expressions of opinion, especially posters' interactions with one another and attempts to challenge or inform one another, demonstrate agency. As El Gazzar (2013) argues, online expressions of opinion both reflect the views of individuals and (seek to) shape the views of (other) individuals; in this sense online posters could be considered "message creators" rather than passive recipients of information. This is particularly the case when posters act as resisters to dominant discourses about sex offenders, and/or as educators of other participants in an online forum.
The current study thus highlights that in contrast to the monolithic public of governments' imaginations (Allen 2002) and indeed, of many researchers' imaginations (Harper and Harris 2016) public views about sex offender policy might be better conceptualized as the domain of "multiple publics" (Kitzinger 2004; see also McCartan 2014), with highly varied, competing and contradictory beliefs and opinions. As outlined in the following section, a critical application of this conceptual orientation provided a basis for examining public opinion about COSA expressed on social media forums.

Methodology
Social media analysis is a relatively new method in public opinion research (American Association for Public Opinion Research [AAPOR] 2014; Anstead and O'Loughlin 2015;Prichard et al. 2015). A number of limitations of using social media in place of traditional data collection techniques to gauge public opinion have been identified, and are important to discuss in the context of the current study. Perhaps the principal criticism is that data sourced from social media lack the representativeness that can be attained by traditional surveys that use random sampling techniques. This is thought to be exacerbated by disparities in access to and proficiency in online technologies, and result in the exclusion of older people and those from low socioeconomic, non-English speaking and/or rural backgrounds (AAPOR 2014; Department for Work and Pensions 2014), instead favoring "people with firm opinions on the subject who enjoy expressing themselves publicly" (Department for Work and Pensions 2014:21). There are, however, a number of counter-arguments to this claim. First, while data obtained from social media sources may not be representative of the community, they can provide an insight into the qualitative nature of individuals' views. As such, the current study seeks to examine the nature of support for or opposition to COSA, rather than to quantify the level of support present in the community. Furthermore, governments are likely to be most interested in the views of lobby groups and cultural influencers or those who are likely to react to government policy, so the views of those who post on social media, while not necessarily representative, may well be the most relevant.
Another criticism of using social media data to examine public opinion is that the demographic characteristics of those who post comments on social media usually cannot be ascertained (Malesky and Ennis 2004), leaving researchers unable to explore relationships between public opinion and characteristics such as gender, ethnicity and age (AAPOR 2014; Department for Work and Pensions 2014). The current study does not, however, seek to examine such relationships, and in any case, the literature on public opinion about sex offenders clearly shows that demographic characteristics are not consistently related to individuals' opinions (e.g. CSOM 2010; Rogers and Ferguson 2011;Willis et al. 2013;cf. Comartin et al. 2009;Mancini 2014;. Finally, data obtained from social media have been characterized as involving a low signal-to-noise ratio (AAPOR 2014); that is, without the guidance of a researcher or instrument such as a questionnaire, those who make comments on social media tend to go "off topic". The "topic", of course, is what the researcher wants "participants" to discuss, rather than what "participants" themselves want to discuss. A counter-argument to this criticism is therefore that social media allow data to emerge organically rather than according to constraints imposed by the researcher; this arguably provides the researcher with an insight into what is important to posters and to what topics emerge naturally, without the input of the researcher. A related concern is that "Unlike survey respondents who typically only provide information when prompted, those who use social media tend to post what they want, when they want, prompted or not" (AAPOR 2014:21). This may be considered to limit the comparability of posters' comments, as they are not responding to the same set of questions as they would be if completing a traditional questionnaire or structured interview. In the current study, however, posters were responding to the same media stimulusthe stories on the Adelaide COSA program described below. In this sense, their responses might be considered akin to data produced via a loosely-structured focus group in which participants are given scope to raise topics of importance to themselves in response to an initial question or topic posed by the researcher.
Data for this research came from four online sources: • all comments (n = 361) made by members of the public on the Facebook page of current affairs television program Insight in response to Worthington's (2015) article Australia Immediately" (see https://www.change.org/p/south-australian-parliament-stop-the-COSA-trial-in-south-australia-immediately-south-australia-are-trialing-a-program-to-offerchild-rapists-friendship-and-support-instead-of-prison-time#petition-updates) between early April (when the petition was created) and 18 April 2015 (the period during which the vast majority of comments were posted).
Comments included in the analysis ranged in length from one word to long paragraphs. In all four data sources, comments that only "tagged" another person into the conversation rather than making a contribution to the debate were excluded. In the case of the Insight, Today Tonight and FACAA Facebook pages, posters not only responded to the original media stimulus (e.g. Worthington's 2015 article), but also interacted with one another, responding to each other's comments, questioning one another, and in a small number of cases, posting links to other sources of information or other news stories. Posters to the www.change.org petition were, however, unable to interact with one another in the same way. Rather, posters to this petition responded to the prompt "I am signing [this petition] because …". As a result, this data source, unlike the other three, captured information only from those opposing the program.
Although other media stories on the COSA program appeared on Australian television and radio in early April 2015, no online comments or other material could be found following these stories that would provide an insight into public support for COSA. A population rather than a sample of comments relating to the introduction of COSA in Adelaide has therefore been analyzed for this article (total n = 768).
All comments were copied and pasted into a Microsoft Word file before being imported into qualitative data analysis software program NVIVO for coding prior to data analysis. Prior to coding, the data were read through twice by the first author in order for a process of familiarization to occur. As Caulfield and Hill (2014) claim, this process is vital when thematic analysis is being undertaken, in order to avoid a superficial analysis. A process of open coding was then undertaken. Open coding involves undertaking a detailed reading(s) of the data and allowing new (i.e. not pre-determined) themes to emerge (Strauss and Corbin 1998).
A thematic analysis, involving "identifying, analysing and reporting patterns (themes) within data" (Braun and Clark cited in Vaismoradi et al., 2013:400;Grbich 2013) was then undertaken by the first author. Thematic analysis is appropriate for research projects that aim to "explore the views, perceptions and/or experiences of groups or individuals, and any differences or similarities in these" (Caulfield and Hill 2014:183). Following Mason (2002), the process of analysis focused on both common themes and counter themesthat is, on points of consensus and divergence among participants. This was particularly important in light of the conceptual framework of "multiple publics", discussed above. However, following Caulfield and Hill (2014:185), care has been taken in the discussion of themes below to give readers a sense of "how much data each theme consists of"; in other words, how frequently particular themes appeared in the dataset.
The use of comments posted to social media for research raises a number of ethical issues (Prichard et al. 2015). Chief among these in the current study is that posters' comments have been used without their consent. As obtaining consent from every individual poster would have been impracticable, and as comments were posted to public Facebook pages (and the public website www.change.org), a waiver of consent was obtained from Queensland University of Technology's Human Research Ethics Committee to undertake the research (approval #1600000209). In accordance with this approval, all posters have been deidentified in this article. To further protect the confidentiality of posters, the forum to which they posted is not indicated alongside their comments.

Findings
The process of data analysis described above yielded a number of distinct, but interrelated, themes. Those who opposed COSA (the vast majority of the online posters) did so predominantly because: they believed that victims of child sexual abuse should receive priority for government spending; they viewed child sex offenders as incapable of changing; and/or because they (erroneously) believed the COSA program would be used either instead of prison or to facilitate early parole release for offenders. Supporters of COSA held positive views of the program because: they believed the program might meet its aim of reducing child sexual abuse; they (erroneously) assumed that COSA work with people who are sexually attracted to children but who have not yet acted on this attraction; and/or they embraced a spirit of experimentalism and championed the introduction of new, innovative justice measures. These themes are discussed in turn below.

Opposition to COSA
The overwhelming majority of members of the public who posted comments to one of the four forums analyzed for this study opposed the pilot COSA program. The program was opposed on a number of discrete but related grounds; these are discussed in turn in the subsections that follow.
Victims should receive priority for government spending The most common reason given for opposing COSA was that victims of child sexual abuse, not perpetrators, should be the beneficiaries of government funding designed to address this issue. Nearly 100 comments of this nature were posted across the four forums. Many dozens of these comments expressed this view in general terms, such as: I think the $ would be better spent helping survivors. Useful, tangible support for the victims FIRST, then for the perpetrators LAST. Children and families need support funding not the perpetrator. Victims need way better support first. This money would be better spent supporting the VICTIMS. Another upside down idea.
A number of sub-themes also emerged within this broad theme. Most frequently, comments within this theme expressed opposition to the COSA pilot because child sexual abusers were not seen as being as deserving of government funding as victims. Comments of this nature thus expressed moral outrage that funding would be directed toward a program for perpetrators, as the following comments demonstrate: I have no sympathy for these deviants and we should not be spending our money on them…. the victims are the sufferers here not them. More funding is needed for the victims not perps. They don't deserve a second chance do their victims get a second chance at life. Why do they need our help? The ones that need help are the poor innocent victims whose lives have been destroyed by this vermin. They don't deserve any rights! How about rights for the children that have been abused! Spend the money on the victims, they are the ones who deserve our support.
A related sub-theme within this broader theme was a sense of frustration and anger that government expenditure would be directed toward perpetrators given what many members of the public perceived as the poor state of service provision that currently exists for victims of child sexual abuse. For example, posters who identified themselves as victims of child sexual abuse described having to actively seek support rather than being offered it, and being offered only "a pittance" by way of crime victim compensation. Similarly, a poster who identified as the parent of a victim described their daughter as being "still…on a waiting list for counselling". Other posters lamented what they perceived as "insufficient access to psychotherapy services", service providers having to "continually fight for funding", and "support services for victims that no longer exist due to government funding cuts". This perception of a lack of funding for suitable and accessible services for victims of child sexual abuse underpinned community members' sense of injustice that any government spending would be directed toward a program for perpetrators. This sentiment is perhaps best captured by the following comment: "What angers me is adult survivors caant [sic] get the help and support they need and Id [sic] rather see them shown the support and compassion that these scum of the earth [perpetrators] are getting".
Such comments reflect an imagined "zero sum game" in which a finite quantity of government funding can either be spent helping victims or offenders. This sentiment was most clearly reflected in one poster's comment that "taking from a victim to make a perpetrator have a great life is unthinkable". While this view was expressed very frequently in the four online forums, however, it was by no means universally subscribed to. Indeed, a small number of posters resisted this dominant view by making the counter-argument that while victims undoubtedly deserve government support, a program like COSA should help prevent future victimization, and ought to be supported on these, more instrumental, grounds. For example, posters stated: Repairing the damage done alone does not prevent future generations from abuse. Attempts by the government to support offenders must be welcomed. There are many support programs out there for victims of sexual assault….Taking one of these approaches does not eliminate the need for the other approaches.
These members of the public thus support preventative measures over (or in addition to) measures that seek to address victimization that has already occurred. These comments, which resist the dominant view that the funding for the COSA program should be redirected to victims' services, demonstrate both the diversity of views held by members of the public (or "publics"), and the agency exercised by posters as they seek to simultaneously challenge and shape the opinions of others. A small number of posters who resisted the dominant view directly challenged it, by making comments such as "It's [the COSA program] not designed to help the criminal. It's designed to avoid more victims", and "[the program] is not taking anything from the pain of the victim, just reducing the number of future victims". One poster even directly confronted those who had identified themselves as victims of child sexual abuse in the forums, but who opposed the COSA program on the grounds that the money should be spent on victims' services, stating that: "I would think as being victims yourself you would welcome a program that hopefully reduces future victims.. Obviously not! I for one welcome a program that can hopefully assist these people in not offending or reoffending".

Child sex offenders cannot be rehabilitated
The belief that child sex offenders cannot be "cured" or rehabilitated underpinned many of the comments made to the four forums, and commonly informed posters' resistance to the COSA program. Many dozens of posters made comments of this nature, opposing COSA on grounds such as: "It's idealism to think these monsters can be 'fixed'", "once a pedophile always a pedophile", and "[a] leopard never changes his spots!!!". This is largely unsurprising, given that the notion of untreatability has been identified as a "deep-seated, powerful and pervasive assumption" (O'Neil and Morgan 2010:23; see also Quinn et al. 2004) and "probably the most deeply entrenched belief about sex offenders" (Federoff and Moran in Thakker 2012:160).
Within this theme, posters engaged in a type of contest of expertise, with many seeking to support their claims about child sex offenders' inability to change by referring to the "science", "evidence" or "research" on this topic. For example, posters commonly opposed COSA based on their belief that "it has been proven" that child sex offenders cannot be rehabilitated, and that this view is not mere opinion but "a fact". As the following comments demonstrate, others made more explicit reference to the "evidence" about child sex offenders' inability to be rehabilitated: Studies have shown a million times over they can not be "cured". All the studies show you can not stop or fix them. There is no evidence that paedohpiles can be reformed. Studies have shown these monsters can't be rehabilitated.
Another group of posters proffered their own experiences as victims of child sexual abuse as evidence that perpetrators cannot be rehabilitated, making claims such as: I've been there as a child and know, god I know, that this person, nothing could change him! As someone who has been affected by one of these people, I don't think they can be changed. My offender doesn't even acknowledge that he did anything wrong. So how can they counsel someone who doesn't even admit to having a problem. I'm also a survivor & no they cannot be rehabilitated.
In a small number of cases, victims of child sexual abuse explicitly challenged the expertise of the COSA program funders and operators, and used their own status as a victim to lend weight to their claim that perpetrators cannot be rehabilitated. For example, one poster asked: "Do you (morons) know what it is like to be sexually abused?? Well I was, from age 5yrs til I was 14 yrs….YOU CAN NOT REFORM A PEDOPHILE". This comment echoes numerous other similar comments throughout the dataset that seek to portray advocates of the COSA program as naïve to believe that COSA could reduce offending. Posters made frequent reference to COSA program coordinators and advocates as "idealistic", "naïve", and "do gooders".
Comments made by members of the public within this theme again demonstrate the diverse publics that participated in the four online forums, as well as posters' efforts to act as message shapers. Posters directly engaged with and challenged one another's views. While only a small number challenged the dominant belief that child sex offenders cannot be rehabilitated, as outlined above, posters variously credited, discredited, supported and dismissed the claims of COSA experts presented in Worthington's (2015) article, and posited other hierarchies of expertise, or in the case of victims, moral authority (Harper and Harris 2016), as discussed further below.
A sub-theme that emerged from the thematic analysis of the data was that because child sex offenders cannot be rehabilitated, COSA represent a waste of government money (e.g. "You can't rehabilitate Paedophiles because they are incurable, so don't waste tax payers money!!"; "I completely refuse to have my tax dollars pay for this. I do not believe child rapists can be rehabilitated"). In some cases, this lead to the view (discussed above) that COSA program funding should instead be spent on victims of child sexual abuse ("There is no such thing as rehabilitation for child molesters! Stop wasting tax dollars on them & start aiming it at programmes for the victims"; "Instead of wasting money trying to fix them…spend it on healing those they have forever harmed!"). Comments of this nature reflect utilitarian beliefs; that is, that program funding would be better spent on victims because perpetrators are not able to be rehabilitated. In the theme discussed in the previous section of this article, however, members of the public opposed money being directed to a COSA program on symbolic groundsbecause perpetrators do not deserve government funding. This is an important distinction for COSA program coordinators, as it may usefully inform their communications with the public about COSA programs, as discussed in more detail later in this article.
That members of the public strongly opposed COSA based on the belief that child sex offenders cannot be rehabilitated raises another key point of relevance to COSA program operators. First, posters' comments demonstrate a pervasive lack of understanding about the difference between pedophiles (individuals with a sexual attraction to prepubescent children) and child sex offenders (individuals who have offended sexually against children irrespective of sexual attraction or orientation) (Richards 2011b). It has been well-documented that much child sexual abuse is opportunistic (Smallbone and Wortley 2001;Wortley and Smallbone 2006), and that it is not the case that "most sexual offenders are dedicated, serial offenders driven by irresistible sexual urges" (Wortley and Smallbone 2006:11). This is important in the context of the current study, as COSA do not seek to "cure" pedophiles, and indeed do not solely recruit pedophiles as core members. Core members may or may not have an exclusive sexual interest in children. Some will have convictions for sexual offences against both adults and children, some will have a sexual interest in both children and adults, and some will have sexually offended against children due to opportunistic factors rather than sexual interest. Rather than seeking to "cure" core members, COSA aim to reduce the risk that they pose to children. COSA do not seek to change an offender's sexual orientation or attraction, but rather to challenge and change their behaviors and the thought patterns that inform behaviors. Thus while it is useful to understand public resistance to COSA on these grounds, the belief that child sex offenders cannot be "cured" need not inform opposition to COSA. Again, this information could usefully inform COSA program operators' communication with communities about the goals of COSA programs and how these are to be met.

Misunderstandings about the COSA program
Finally, opposition to COSA was frequently premised on misunderstandings about the program. For example, members of the public resisted the program because they (incorrectly) believed that COSA would be used instead of a prison sentence for offenders or be used to facilitate early parole: I think these people should be punished for their crimes not to be accomadated [sic] for their heinous crimes….
[or be]….released from prison early. They can't just think they can mess with a child and just go to counselling. They should be punished not supported.
A related reason for opposing the COSA program was the mistaken belief that this program reflects lenient sentencing, and that the availability of a COSA program would result in the judiciary sentencing sex offenders to shorter prison terms: If I was to break into someone's car I would have a much harsher penalty, even if I didn't take anything. These children need justice, not to see their attackers walking free. I'm appalled at the leniency given to convicted paedophiles. There is very little punishment for their crimes. Protection of children should be the role of the government and judiciary not making their sentences lighter or easier.
While community members' concerns about offenders being released from prison early if COSA are available are understandable, this is not how COSA operate, either in the current South Australian program, or internationally. Rather, COSA emerged as a mechanism to reduce the risk posed by sex offenders who had served their entire sentence (including their parole period) in prison, and were to be released into the community with no monitoring or government oversight ). As discussed further in the following section, this suggests that providing information to the public about this aspect of the operation of COSA is a vital step toward securing public support for this program.

Support for COSA
As noted above, the vast majority of comments posted online following the announcement of the Adelaide COSA program expressed opposition to the program. However, small numbers of posters resisted the dominant discourses outlined above, and expressed support for COSA. In addition to comments that expressed general support for the COSA program without explaining the reasons for this (such as "FINALLY", "Brilliant idea" and "SA [South Australia] laws makers are quite forward thinking"), posters expressed support for COSA for a number of reasons: due to the belief that COSA could prevent child sexual abuse; based on the (incorrect) assumption that COSA work with people who are sexually attracted to children but who have not yet acted on this attraction; and/or because they embraced a spirit of experimentalism and championed the introduction of new, innovative justice measures. These are discussed in turn in the remainder of this section.

COSA could prevent child sexual abuse
The most frequently expressed reason for support of the program was the belief that such an approach could prevent sexual offending against children. Numerous posters made comments of this nature, such as: Anything that prevents children becoming victims, yes. Any program that is proven to reduce or prevent the abhorrent act of child sexual abuse is money well spent. Proactive solutions are ALL-WAYS better than reactive solutions. These programs help with recidivism… Big tick from me! Finally! We are acting on prevention. I think this is an intelligent way to handle a wide spread and terribly destructing problem.
A small number of posters provided more specific reasoning for their belief in the ability of the COSA program to prevent child sexual abuse. For example, some claimed that they support the program because it addresses the stigmatization of offenders or their ostracism from the communityfactors that these posters believed would contribute to reoffending: I think this is an excellent idea, and long overdue. Our current stigmatisation of child sex offenders has meant we approach this as an emotional issue in a reactive manner rather than attempting to reduce offending proactively. Fantastic! About time we recognise that ostracism is only likely to push peadophiles into that extreme lonely space where they're more likely to offend! Absolutely in support of this program.
At times, posters' comments about the ability of COSA to prevent child sexual abuse were framed in response to arguments made by previous posters against the COSA program. For example, a number of posters pointed out that despite the debate taking place about the program, both "sides" (i.e. those for and against the program) share the same goal of protecting children: I think it's important to remember what the goal is here. It's a safer community and to reduce offenders…. While I am absolutely disgusted by paedophilia if this program is reducing re offenders and making the community safer, then it has my support.
If it helps to stop offending how could anyone be against it? Who are we trying to protect here? What is the end result we want? Isn't it for all of our children to be safe.
In other cases, comments about the prevention capacity of COSA were framed in response to the frequently-made argument that more should be done to assist the victims of child sexual abuse (as outlined above). In the examples below, supporters of COSA argued that while supporting victims of sexual abuse is important, preventing new victims should also be a priority: What your family has gone through sounds terrible….Victim support needs to be MUCH better….Hopefully a program like COSA can prevent others suffering in the future as your family has.
[T]here already are many programmes for victims. This programme is about reducing the number of future victims.
Misunderstandings about the COSA program An important caveat to the above discussion is that although many posters supported COSA on the grounds that the program may prevent child sexual abuse, some comments appeared to be based on the mistaken belief that COSA support "virtuous pedophiles" -those who are sexually attracted to children but have not yet acted on this attraction (Cantor and McPhail 2016). In reality, COSA are a form of tertiary crime prevention (Richards 2011a) in that they work with convicted child sex offenders. While the following comments suggest support for COSA, they simultaneously suggest that this support might be withdrawn if the nature of COSA was better understood by the posters (italics added): If these people have not offended -and there is a way to 'help/stop them' from ever offending, shouldn't we try. I've seen documentaries where there are people who have the urge but say they have never acted upon it….if there are people who want proper treatment or guidance that prevents children being hurt in such a horrific way, shouldn't the option be there? It seems counterintuitive to deny them support to NOT offend. If they recognize they need help before they act then we should embrace it and help them.
These comments appear to reflect contradictory views about the instrumental and symbolic features of COSA. While members of this group of posters support the instrumental aim of preventing child sexual abuse, they do so only insofar as an individual with a sexual attraction to children has not acted on this attraction. Once an individual has offended, support for this instrumental aim is withdrawn, as offenders no longer deserve assistance or supporteven if this is designed to prevent them offending again. The ways in which posters ascribed symbolic and instrumental values to COSA is considered further in the discussion section that follows.
Other supporters of COSA nonetheless explicitly understood that COSA work with known child sex offenders as a tertiary prevention measure. These posters argued that there is value in preventing reoffending among this group of offenders. For example, posters claimed that "if this program is reducing re offenders and making the community safer, then it has my support", and "Any program that keeps offenders from re-offending is a win" (italics added). This suggests that for a small number of posters at least, COSA's instrumental aim of preventing child sexual abuseeven by convicted child sex offendersprovides valid enough grounds on which to support the program.

Spirit of experimentalism
Finally, a small group of posters adopted what might be considered a spirit of experimentalism, and expressed support for trialing any type of program or approach that might reduce the sexual victimization of children. Typical comments included: Sure give it a crack. Can't hurt to try. We need to do whatever it takes. If this helps then let's do it! If there is the slightest hope that this program will help to reduce the number of children abused than it is worth trying it… For a small number of COSA supporters, such a view was premised on the belief that new approaches are needed to tackle child sexual abuse, given that current approaches seem to be ineffective: "If this helps, I'm all for it. Our current approach clearly isn't working"; "It's high time we trial a new strategy". Others believed that COSA have been successful in reducing child sexual abuse in international jurisdictions, and ought to be trialed locally on these grounds. For example, posters commented that: "The COSA program has been shown to be very effective in other countries, so why wouldn't we set it up here?!" and "The stats from other areas certainly validates the trial!".

Discussion
As the above findings demonstrate, members of the public hold a diverse array of opinions about COSA, from vehemently opposing it to enthusiastically supporting it. While there was much resistance to the announcement of the South Australia COSA program, these findings suggest that there is in fact a diversity of attitudes toward COSA among multiple "publics", and that information about sex offender policy is understood and reproduced in highly varied ways by a diversity of publics. These findings thus reflect Brown et al.'s (2008:272) claim that "what people think about the management of sex offenders in the community spans a broad spectrum of public opinion" (see also CSOM 2010), as well as Esser-Stuart and Skibinski's (1998:101) claim that the social response to sex offending is complex. Furthermore, they support Kitzinger (2004) and McCartan's (2014) calls to reconsider the homogeneous public of the government's imagination and instead embrace the notion of "multiple publics". The findings outlined above also provide an insightlacking in the predominantly quantitative research about this topicinto the ways in which members of these publics can reach the same conclusions via very varied logics (see also Harper and Harris 2016), again suggesting that when we speak of "public attitudes", there is no singular, homogeneous public, but a diversity of publics.
One way of making sense of posters' diverse views is to consider the symbolic and instrumental values they variously ascribed to COSA, and the ways in which these perceived values informed opposition to or support for COSA. It is clear from the findings presented above that those who support COSA do so on almost universally instrumental grounds. For example, COSA supporters highlighted the capacity of the program to reduce reoffending of child sexual abusers, and thus prevent victimization and enhance community safety. Conversely, COSA opponents' comments predominantly reflect symbolic concerns. Specifically, opponents argued for greater victim support and harsher penalties for offenders, and resisted the provision of support for offenderseven though such support may reduce child sexual abuse. The findings of our research thus suggest that for the publicor rather, for some publics -COSA lack symbolic value; they do not "reassure the public by helping to reduce angst and demonstrate that something is being done…[or]…solidify moral boundaries by codifying public consensus of right and wrong" (Sample et al. 2011:28).
Indeed, to suggest that COSA lack symbolic value undoubtedly understates the issue at the heart of public resistance to COSAthe deep-seated belief systems that pre-determine how members of the public interpret information and evidence about sex offender policyoften in ways that simply confirm their existing beliefs. These will undoubtedly be difficult to shift. However, community education can be successful in changing attitudes if delivered effectively (Kleban and Jeglic 2012). The remainder of this article therefore considers both what key messages about COSA are key to changing community views, as well as how and by whom these messages might most effectively be communicated to the public.

Ways forward
It is clear from the analysis presented in this article that better information needs to be communicated to the public about sex offending, and policies and practices introduced to prevent it, including COSA. The results indicate that in particular, members of the public confuse and conflate pedophilia and child sex offending, and this in turn informs strong opposition to the COSA program. Community education about the COSA program thus needs to be clear about who core members are, and that the program does not try to "cure" pedophiles, but rather to reduce the risk of reoffending that child sex offenders pose (by addressing risk factors such as social isolation). That is, COSA seek to change thoughts and behaviors rather than an individual's sexual preference or orientation. Many members of the public in the current study opposed COSA due to inaccurate beliefs, such as that COSA would be used instead of (rather than following) a prison sentence. While it has been well-established in the literature that "acceptance of treatment of sex offenders as an alternative to long punitive incarcerations is low" (Rogers and Ferguson 2011:398), if COSA were to be correctly understood by the community to involve support and assistance with reintegration following prison, this program may well be better supported by the public. Similarly, the cost benefits of COSA could be better communicated to address public concerns about government spending on COSA programs. A further recommendation that stems from our analysis is that COSA's key aim of preventing child sexual abuse should form the focus of any public communication about the program, given the strong negative reaction to the image of "supporting" child sex offenders. Finally, more focus could be placed on the monitoring and accountability functions of COSA over their support function. While Hannem (2011:279) cautions against avoiding discussion of the "more radical aspects of its philosophy" in public discussion of COSA, she acknowledges that in the Canadian context, emphasizing their risk management function "has been a useful strategy". In line with this, our findings suggest that having at least some focus on the risk management and accountability aspects of COSA would go some way to addressing their lack of symbolic potency.
As noted above, however, individuals' views about child sex offender policy are deep-seated and firmly-held. More and better information about the issue such as that suggested above is unlikely to be effective at changing community attitudes in and of itself, as "people feel a response before they deliberatively consider the issue" (Harper and Harris 2016:7;italics in original). There are nonetheless a number of ways that this problem can be addressed; it is not simply what must be communicated, but how and by whom it ought to be communicated that is critical in this context. In relation to the former, recent research shows that the terminology used when discussing sex offenders has a significant impact on public attitudes, since "labelling an individual as a 'paedophile' (or a 'sex offender') evokes a visceral reaction that is independent of a rational understanding of that individual's offending behaviour" (Harper and Harris 2016:7). Indeed, both Imhoff (2015) and Harris and Socia (2014) found greater support for punitive measures when emotive terms (e.g. "pedophile") were used than when more sanitized descriptors such as "people with a sexual interest in children" were used (see generally Harper and Harris 2016). Similarly, Viki et al. (2012) conducted four interrelated studies that demonstrated that the more that individuals view sex offenders in humanizing terms, the more likely they are to support rehabilitative measures for this group, and the less likely they are to support exclusionary measures. Viki et al. (2012Viki et al. ( :2364 usefully conclude that based on these findings, "interventions designed to increase community perceptions of sex offenders as being human might result in more support for rehabilitation programs". McCartan et al. (2015) made the related suggestion that public communication about child sexual abuse might be best delivered within a public health rather than a criminal justice framework, in order to avoid othering offenders and highlighting the social distance between "them" and "us". McCartan et al. (2015) further suggested that tone is important when communicating with the public about sex offender policy, and recommend speaking "with" rather than "at" communities.
A final consideration is who ought to deliver such messages to communities. Given that COSA program staff and academics were roundly dismissed as "idealistic" "do gooders" by members of the public, a further recommendation would be to involve COSA volunteers, especially volunteers who come from professions such as the police, in community education. Victims and victim advocates are also often involved in COSA programs as volunteers (Richards 2011a), and should likewise be drawn on in public communication about COSA, since victims are deemed to possess the authority that professionals and academics lack. Victims make vital contributions to COSA programs, in particular to holding offenders accountable; this should be communicated to the public to lend gravity to the program and foster community support for COSA.

Conclusion
Much public resistance followed the announcement of Australia's first COSA program. Nonetheless, online comments analyzed for this article demonstrate a diversity of public opinion, and suggest the need to engage with multiple publics around the issue of responding to child sex offending. Based on this analysis, this article has made recommendations both about key messages that should be communicated to the public about COSA, and about effective strategies for delivering these messages. In particular, using non-emotive and humanizing language will assist in harnessing community support for COSA.
A final point is that more qualitative research is neededboth into public opinion about sex offenders generally, and about COSA specifically, in order to complement the substantial body of the existing quantitative literature. Qualitative research is necessary to provide insights into the nuances and complexities of public opinion about sex offenders and what ought to be done about them. Qualitative research using a large, representative sample of the public would best meet this aim, build on the research presented in this article, and contribute to our understanding of this important topic.