Imprisonment, Social Support, and Desistance: A Theoretical Approach to Pathways of Desistance and Persistence for Imprisoned Men

Desistance should be the main ground for reentry policies for imprisoned offenders. However, theories on desistance are diverse, and they disagree about the key factors related to the origin, maintenance, and failures of the desistance process. This research considers three main theories of desistance—control, cognitive transformation, and strain-social support—to explain desistance in a sample of imprisoned men in Spain. The main finding of the research is that strain-social support theory may be of primary importance for understanding desistance because of its capacity to explain processes of change that begin during imprisonment and that continue upon release.


Introduction
Given that desistance theories provide the factors and mechanisms that explain why and how offenders break with criminal careers, it seems reasonable that they should be the main grounding for rehabilitation and reentry policies for imprisoned offenders.
However, despite the relevant number of studies over the last 35 years on the factors and processes that explain the abandonment of criminal careers, important theoretical debates about the desistance process remain.
This article examines debates identified among three relevant theories in the field (control theory, cognitive transformation theory, and strain-social support theory). The debates concern the origin of the desistance process (objective vs. subjective factors), the mechanisms for maintaining desistance (such as commitment, attachment, supervision, hooks of change, compensation, and moderation of strain), and the contingencies of the desistance process when the person cannot achieve conventional adult roles (failure vs. success). We address these three debates by analyzing a diversity of desistance and persistence processes in a sample of 36 incarcerated men in Barcelona (Catalonia) who were interviewed twice: in the final weeks of their prison sentences and between 1 and 2 years after the expiration of the prison sentence. The analysis aims to identify which of the aforementioned theories provides stronger support in explaining the factors and mechanisms that are involved in the process of desistance.

Literature Review
To understand changes in criminal careers, two major approaches have been developed by desistance scholars: the revised version of control theory (Laub & Sampson, 2003;Sampson & Laub, 1993) and cognitive transformation theory (Giordano, Cernkovich, & Rudolph, 2002;Maruna, 2001;Paternoster & Bushway, 2009). In addition to these two major approaches, this article examines social support intended as a development of strain theory (Cullen, 1994) 1 -that is why we use the term strainsocial support theory-which has been less explicitly used as a framework for understanding desistance but which appears to have emerging relevance in research (Calverley, 2011;Schroeder, Giordano, & Cernkovich, 2010;Visher & O'Connell, 2012). In this section, we review these three theories to explain the origin, maintenance, and contingencies of the desistance process.

Origin of Desistance: Objective Versus Subjective Factors
Theories of desistance agree that in the whole process of desistance, both objective and subjective factors are relevant. However, a degree of disagreement exists over the type of factor-internal or external to the individual-that initiates the process of change in criminal careers.
Control theory adopts a relational vision of the desistance process, assuming that delinquent acts result when an individual's bond to society is weak or broken (Hirschi, 1969). Although the original theory aimed to elucidate the onset and maintenance of offending behavior, with the "age-graded theory of informal social control," 2 Sampson and Laub (1993) extended it to explain desistance. According to these authors, desistance originates from the formation of both new social bonds in adult life (such as a stable marriage or stable job) that produce a stake in conformity and life routines that are incompatible with an offending lifestyle. Although Laub and Sampson consider agency to be one relevant element of the desistance process, subjective factors can be interpreted to have only a secondary role in their theory because individuals choose to desist because of the desire to live according to the requirements of conventional adult roles. In the words of the authors, " . . . we believe that most offenders choose to desist in response to structurally induced turning points that serve as the catalyst for sustaining long-term behavioural change" (Sampson & Laub, 2008, p. 172). Paternoster and Bushway (2009) consider Laub and Sampson's (2003) expression "desistance by default" (p. 278) to synthesize their position.
The position of Sampson and Laub has been challenged by a number of authors who maintain that before individuals may adopt new roles that promote desistance, they must have a mental state of openness to change (Bottoms & Shapland, 2011;Giordano et al., 2002;Lebel, Burnett, Maruna, & Bushway, 2008;Maruna, 2001;Skardhamar & Savolainen, 2014). Thus, for cognitive transformation theorists, the main catalyst for change is not a turning point that is external to the individual but a subjective reflection on the self. Although some authors have explored the association of this cognitive change with spiritual and religious practices (Giordano, Longmore, Schroeder, & Seffrin, 2008;Hallett & McCoy, 2015) or with certain negative events in offenders' lives (e.g., arrest, incarceration) that induce a reflection on the "feared self" (Paternoster & Bushway, 2009), the factors that generate "openness to change" (Giordano et al., 2002) in some people but not others remain unclear. 3 With respect to strain-social support theory, Cullen (1994) in his pioneering article suggests that "social supports may exert independent (main) effects on crime . . . by . . . transforming deviant identities" (p. 542). Following Lin (1986), Cullen considers social support to entail instrumental and/or expressive provisions that are supplied by people and organizations. Thus, this support can be provided at the micro level, through personal ties, but "can be viewed as a property of social networks and of communities and larger ecological units in which individuals are enmeshed" (pp. 530-531). When applied to desistance research, this argument suggests that the desistance process has an external, objective origin. However, Cullen also underscores the importance of accounting for not only actual but also perceived support "because it leads to the insight that people do not receive support in a mechanical way but interpret, appraise, and anticipate it in the context of social situations" (p. 530). This argument allows for the theoretical integration of the external, objective resources that activate the desistance process with the subjective mechanisms that are involved in the process. This argument has also been developed in other studies, although it has not been explicitly placed within strain-social support theory (Dufour, Brassard, & Martle, 2015).

Mechanisms for Maintaining Desistance
For theories of desistance, similar to theories of crime, it is essential that the description of the psycho-social processes that mediate between the factors that originate desistance and the behavior of abstinence from criminal offending is clearly stated. The three theories under consideration display different mechanisms to explain the maintenance of desistance.
According to control theory (Laub & Sampson, 2003;Sampson & Laub, 1993), once a turning point in offending behavior has occurred, different mechanisms explain the maintenance of desistance: commitment in the new role (for example, a stable partner increases an individual's bonding social capital and raises the individual's stake in conformity), attachment to the persons to whom new bonds have been created and supervision in the context of participation in conventional routines within institutions such as marriage and stable work.
Cognitive transformation theory, especially as formulated by Giordano et al. (2002), states that once a person has shifted to a mental state of openness to change, the person must have access to "hooks for change," that is, new life circumstances, such as involvement in a treatment program, a religious experience, or a new couple relationship. These hooks may be useful for elaborating a narrative of the change process, facilitating new pro-social relationships, and ultimately allowing people to build alternative versions of themselves.
Strain-social support theory as originally formulated by Cullen (Cullen, 1994;Cullen & Wright, 1997) may be considered a development of strain theory (Agnew, 1992;Merton, 1938). Social support is intended as a factor that prevents criminal behavior by moderating the criminogenic effects of strain on crime (Cullen, 1994;Cullen & Wright, 1997). Although strain-social support theory is rarely used as an explicit theoretical framework in desistance studies, the general finding that persisters have experienced (or perceived) more social problems than desisters (Bottoms & Shapland, 2011;Burnett, 1992;Farrall, 2002;Lebel et al., 2008;Zamble & Quinsey, 1997) may be effectively interpreted in the framework of strain-social support theory. Many of the obstacles to desistance that offenders report in these studies-financial problems, lack of work, lack of residence, drug addiction, poor family relationships, and a criminal record-may be considered "stressors" within the context of offenders' lack of support to overcome these obstacles in a conventional way (Agnew, 1992(Agnew, , 2006. Moreover, strain-social support theory is implicitly involved in the implications of these studies' findings based on helping offenders resolve their social difficulties (Bahr, Harris, Fisher, & Harker Armstrong, 2010;Bottoms & Shapland, 2011;Farrall, 2002;Lebel et al., 2008). Recent research has underlined that in addition to lessening strain, family support for some may produce mental and emotional states such as optimism, conventional identity, or willingness to change that may be related with desistance (Calverley, 2011;Cid & Martí, 2012;Schroeder et al., 2010;Visher & O'Connell, 2012).

Outcome of the Desistance Process When Conventional Adult Roles Are Not Achieved
Although research on the relationship between the transition into adult roles and desistance remains inconclusive (Siennick & Osgood, 2008), desistance scholars appear to agree that " . . . desistance is associated with transitions to full-time employment and marriage, as well as other adult markers, such as having children and achieving financial independence" (Massoglia & Uggen, 2010, p. 553). However, we still think that cases in which people have begun a process of desistance but have not been able to achieve these adult markers or have achieved some adult markers (e.g., employment, partner relationship) in the initial stage of their desistance but have lost them later are more problematic. We believe that control theory, cognitive transformation theory, and strain-social support theory may offer different answers regarding whether these threats can derail the desistance process.
For control theory, acquiring these conventional adult roles is an essential part of the desistance process. If individuals are not able to achieve the "full status" of adulthood, their stake in conformity may diminish, and if their lives are not structured in conventional routines, opportunities for crime may appear. Consequently, we assume that control theory would suggest that failure to achieve conventional adult roles may derail the desistance process.
According to cognitive transformation theory, the contingencies that imply the failure to achieve these roles or the loss of them (e.g., from divorce or unemployment) may not threaten the desistance process if the person has already developed an identity in which offending behavior is considered unacceptable (Giordano et al., 2002). Losing a hook for change may increase the difficulty of the desistance process, but the person may actively seek other hooks to maintain the change (Bottoms & Shapland, 2011).
For strain-social support theory, events such as being unemployed after a prison sentence, losing a job, or ending a marriage or a romantic relationship are stressful situations that, according to strain theory, may lead to offending behavior (Agnew, 2006). However, Cullen (1994) suggests that the amount of social support that one receives moderates the relationship between strain and offending behavior. The theoretical implication is that failure to achieve conventional adult roles after the desistance process begins will not lead to a relapse into offending behavior when the person has sufficient social support.
In Tables 1 and 2, we present a synthesis of the explanation of desistance given by the three theories under consideration.
From the previous review, we may conclude that there are competing theories devoted to explain the whole process of desistance. We think that the knowledge we have on the capacity of each theory to clarify pathways of desistance and persistence is already scarce and probably not enough transferable to different social contexts. Furthermore, we believe that sensible policies of reentry may improve when they are theoretically driven. These considerations oriented the research we have conducted in Spain in which we try to analyze the capacity of the three theories in dispute to explain the desistance and persistence processes for the imprisoned men who took part in the study.

Method
The research adopted a qualitative approach aimed at tracing the reentry process of men who were sentenced to prison for acquisitive crimes in Spain. In particular, we used narrative interviews that were conducted in two waves: in the final weeks of their prison sentences and between 1 and 2 years after the expiration of the prison sentence. 4 The research population consists of men who were imprisoned for property offenses and drug dealing in the province of Barcelona (Catalonia). From this population, a purposive sample was selected to include participants of different ages (because desistance pathways may differ between youths and adults) and different criminal backgrounds (because opportunities for desistance may be affected by cumulative disadvantages, as stated in Sampson & Laub, 1997). To ensure the presence of desisters and persisters in these diverse situations, the sample included men who were ending their sentences in both open and closed regimes, which is an effective predictor of recidivism in Catalonia (Capdevila & Ferrer, 2009). To obtain this sample, offenders with expiring sentences were asked by professional prison or parole staff to participate in the study. Individuals with profiles that were underrepresented in the sample were selectively asked to participate in the final sampling phase. The consent rate for the sampling process was 70%, and 67 men were interviewed (see Table 3 for sample characteristics). In the second wave, 36 of the 67 participants were reinterviewed between 1 and 2 years after the expiration of the prison sentence. We were unable to locate 26 of the remaining 31 participants; two refused to be reinterviewed, two passed away during the follow-up period and one was still in prison at the end of the follow-up. The sample for the presented analysis includes data from the 36 men from the follow-up: 21 of them have been qualified as desisters because they have not been reincarcerated in the 2 years since their prison sentences ended, and they have not reported offenses that could result in their return to prison; the other 15 have been qualified as persisters because they have committed new offenses after their prison sentences and have been reincarcerated. 7 As indicated in Table 3, two main differences in the sample characteristics can be identified between the first and the second waves. The first difference is the underrepresentation of foreigners in the second wave due to the high geographical mobility of these participants. Therefore, some pathways of desistance that are potentially more prevalent among this population may have been discarded. The second significant difference between the two waves is the underrepresentation of non-recidivists in the second wave. This difference is due to the higher difficulty of locating participants who have not returned to prison. 5 a Population with expiring prison sentence in the province of Barcelona (April-July 2010). Source. Catalan Prison Administration (SIPC). b New offense that brings to reincarceration committed within 2 years of the sentence expiration. In 14 cases, reincarceration was within the 2 years, and in one case, it was some months afterwards. c Two persons passed away during the follow-up period; one participant was still in prison at the end of the follow-up; one participant with no recidivism data. a Early-onset offenders: First reported property, violent or drug offense committed in late childhood or early adolescence. Late-onset offenders: First reported property, violent or drug offense committed after the age of 18. b New offense that brings to reincarceration committed within 2 years of the sentence expiration. In 14 cases, reincarceration was within the 2 years, and in one case, it was some months afterwards. Table 4 shows the profiles of the 36 analyzed cases, considering the participants' age when they were interviewed in the first wave, the age at which according to their reports they began engaging in offending behavior, and their status of desisting or persisting in offending behavior during the follow-up period are considered. The first profile (men up to 35 years old with early-onset offending behavior) is the most common in both our sample and Catalan prisons. Some of the men with this profile are Spanish, whereas others are foreigners (mainly immigrants from North Africa and South America), but men with both origins grew up mostly in poor families and in criminogenic neighborhoods. The second profile consists of men who are older than 35 and who also show early-onset offending behavior. Most of them are Spanish-born men who grew up in neighborhoods in the metropolitan area of Barcelona, which suffered from high levels of drugs and social exclusion in the 1980s. Most of these participants have spent a significant part of their adult lives in prison, and they have experienced drug abuse-a central topic in their past and present-which has had negative effects on their current health (most of them suffer from illnesses such as HIV, hepatitis, or mental disorders). Finally, late-onset offenders mainly comprised men who were born in non-criminogenic neighborhoods (in Spain or abroad) and who did not report engaging in offending behavior during their childhood or early adolescence. Often, these individuals have a previous work record, but they often report periods of unemployment as a factor related to offending behavior.
As stated previously, the main instrument of this research was narrative interviews, which were applied in two waves. The first interview occurred a few weeks before their sentences expired and was conducted in prison or parole offices by a member of the research team after professional prison staff members made the first contact. This interview aimed to obtain information about the participants' trajectories-with a special focus on their situation and changes during their current sentence-and their expectations regarding the reentry process; it comprised three parts: the person's background (e.g., family, neighborhood, education, job, delinquency, drug use, and imprisonment), the experience of the current prison sentence, and future prospects for after the expiration of the prison sentence. The second wave of interviews, which was conducted between 1 and 2 years after their sentence expired, aimed to determine the evolution of the participants after their prison sentences had ended and their present situations in different social fields (e.g., residence, family relations, work, other sources of social support, drug use, and offending), with a particular focus on how they had addressed their relationship with offending behavior. These interviews were also conducted by a member of the research team, either in or outside prison (depending on the participant's situation during the second wave). The approximate mean duration of the interviews was 90 min in each wave, and interviews were recorded and transcribed to conduct a qualitative analysis. Apart from the interviews, longitudinal data on trajectories were obtained by using a life-history calendar, and data on reincarceration during the 2 years following the expiration of the sentences were provided by the Catalan Prison Service.
The analysis presented in the "Results" section of this article aims to identify the factors and mechanisms that operate at each of the three stages of desistance (origin, maintenance, and outcome when a conventional adult role was not achieved) by considering the commonalities and differences among the participants.

Origin of Desistance: Objective Versus Subjective Factors
This section explores which theory under consideration most accurately describes the factor that influenced these men at the beginning of their desistance. The following results supporting each theory are expected: Control theory predicts that the beginning of a new pro-social adult relationship precedes the emergence of a narrative of desistance. Strain-social support theory would agree with the need for an external source, but it would emphasize the support provided by conventional persons. Unlike the other theories, cognitive transformation theory would place the emergence of the process of desistance in the participant's reflection about the need to change.
When the 21 interviewed desisters talked about the origins of the desistance process, they distinguished two different moments: before they began to serve their prison sentences and during the process of serving their sentences.
One factor among the participants who were incarcerated while in the process of desisting was commencing a new romantic relationship that made them think differently about engaging in offending behavior: Why did you change? Why did I change? Because I met the girl I'm with now, and she persuaded me not to do these kinds of things . . . (E28, Desister, 23, Interview 1) However, this situation was not particularly common in our sample; only some of the young adult interviewees (up to 26 years old) had experienced this turning point. Instead, for other young adult and for most adult desisters (older than 27 years old), prison was the setting where these men reported entering into moral conversations with themselves about regretting their past lives and began to think about changing. As suggested by Paternoster and Bushway (2009), for some of these men, the negative evaluations of their own identities occurred at the very beginning of their prison sentences, arising from thoughts about the consequences of their behavior, whereas for others, the moment of reflection occurred after they had served part of their sentences, arising from meaningful events that had occurred during prison life: I had been sentenced to three and a half years, and as a consequence of the fight I was involved in while in prison, I had to serve 6 years. Then, I realized that I should avoid problems . . . I should change because I didn't want to come back to prison . . . (E363,Desister,28,Interview 1) From this quotation, one may conclude that in the process of serving their prison sentences, the participants engaged in a process of rethinking about their past, which opened them up to change, as observed by Giordano et al. (2002). However, we explored whether the participants who reported to begin this change process during imprisonment were supported by conventional families and/or partners and whether this observation reflected a general pattern. Developing a feeling of agency about change was generally produced within the context of relevant others who were committed to the participant's change process: . . . just after I was released, my brother said to me: "I promised myself that all my worries would finish when you were released." Hearing this from my brother broke my heart, and I felt like shit. Because, Jesus, it's not only about you, is (.). It's everything . . . And then you're released, and you realize all the consequences of your shit, your behavior; that's when you say, "Fuck it, I'm going to do it for me and for them." (E338, Desister, 31, Interview 1) The effect of social support on the participants' openness to change in prison can also be observed in the persisters' narratives. The lack of a change in their identity and their fatalism about their capacity for avoiding crime on release were related to a lack of family support: If I had a different family, if my parents hadn't divorced . . . , if I could find support, support from someone, even if it was only for five minutes, support from someone, this would help me. However, I know that I'm not going to get this support. Not from my father, even if I go to his house, not from my mother, not from my brother, not from anyone. (E58, Persister, 30, Interview 1) In summary, although prison was the setting in which most of the participants began to identify a change process and although certain events in their prison lives were reported by participants as relevant for their cognitive transformations, the analysis indicates that these events had a positive impact only when certain external factors-in particular, relevant others-catalyzed the change. This finding is in line with control and strain-social support theories and is similar to the results reported in Soyer (2014).
By contrast, subjective factors appear to be catalysts for desistance for specific individuals only, and these processes are more understandable within the framework of cognitive transformation theory.
First, subjective factors are catalysts for desistance among young adult offenders who reported comparatively more positive childhood family lives than other participants, who had some job experience during adolescence, and who began to reflect on their lives at some point during their initial arrest and period of imprisonment (Moffitt, 1993;Paternoster & Bushway, 2009). All these factors likely explain a robust sense of agency that appears to act as a catalyst for change: I feel much pity for many people, you know? I feel much pity for these people who are still, for instance, offending or carrying out a bad life, and I . . . understand that everyone has his own circumstances, but I also see that if you want to do something, you can do it; if you want to follow a path, you will follow it. No one has the power to bring you to one side or to another; everyone knows what is good for him and what harms him. (E265, Desister, 24, Interview 1) Second, subjective factors are catalysts for desistance among middle-aged participants (above 40 years old) with a long trajectory of crime and drug abuse and with a lack of social ties. As explained by Shover (1985), such individuals often express a feeling of tiredness about a life in which imprisonment has been constant: Any day, I could die of an overdose, and it's a miracle that I'm still alive. I know that this

Mechanisms for Maintaining Desistance
In this section, we focus on the capacity of the aforementioned mechanisms (i.e., commitment, supervision, attachment, hooks for change, compensation to a partner or to family, and moderation of strain) to explain the maintenance of the desistance process. The following results supporting each theory are expected: Control theory should be favored if the mechanisms that maintained change were the commitment to maintain pro-social relationships, attachment to relevant persons in these new social relationships, and supervision made by pro-social persons to participants. Cognitive transformation theory would expect that desistance was based on the access to hooks for change and the subject's determination to search for them. Finally, strain-social support theory would be emphasized if maintenance of desistance was linked to the moderation of strain produced by support and to the attitude of the participants to preserve change as compensation to the people who provided support.
Compensation is the mechanism that was most widely identified in our sample. Participants of all ages who received support from a family and partner viewed their own change primarily as a moral duty to compensate for the support that they had received: Second, the study participants widely emphasized that assistance received by relevant others moderated the strain that prisoners may suffer during imprisonment and at release. The evidence comes not only from desisters but also from interviews of persisters of all profiles who provide evidence of the relevance of this mechanism for understanding desistance. The interviewees reported that in addition to pressure from peers, the strain from not having a job, having insufficient money to meet their needs and those of their families, or having insufficient support to overcome those difficulties was the main reason for their persistence in engaging in offending behavior. Commitment is a mechanism that was also found in our sample when the participants were involved in conventional partner relationships: I think that if I didn't have a strong relationship with my wife, with whom I have a strong relationship and who has helped me, and if I didn't have a son, I would have stolen again, because I was accused of things that I had not done, and this infuriated me, and I thought "If I'm accused of doing something that I haven't done, why not do it? . . . I think another person would have already relapsed by now." (E80, Desister, 28, Interview 2) Finally, hooks for change constitute a mechanism in different situations. First, they foster desistance when a person lacks both partner and family ties and when the person has become open to change without external support. This situation characterized some of the participants in our sample, and depending on their own resources, they may look for a job, social benefits, or community assistance: . . . I hope to see myself settle down and having quit drugs, or at least having reduced my use to methadone. I hope to get away from this village and to have a bedroom or a house. However, it is difficult, because it costs a lot of money, and I don't know; I don't know. I think what is more feasible for me is to ask my social worker to find a place for me in a therapeutic community. I'm considering that, and I think I'm going to try it. (E330, Desister, 47, Interview 2) 6 Second, hooks for change also fostered desistance for interviewees who received support from partners and/or families who gave them the motivation to devote time to participating in activities (treatment and training programs, looking for jobs) that may have favored their early release and their desistance. I'm waiting for a temporary job at Christmas, and after that, I have to call other people who will offer me work until March. It is clear that I have to work. I have to work because as I mentioned to you before, we have plans with my girlfriend, and the plans have to be accomplished, and for that, you need to work. (E143,Desister,36,Interview 2) In this way, we realize that interplay occurs between social support and hooks for change. The support produced the motivation to search for hooks for change, and this searching of hooks for change is what maintained desistance.

Outcome of the Desistance Process When Conventional Adult Roles Are Not Achieved
This section addresses participants who were classified as having a desistance narrative at their first interview-for having experienced a break with their offender identity or reinforcing their worker identity and for feeling self-confident about achieving their conventional plans-but who faced threats to achieving a pro-social adult role after their release from prison. Our aim here is to explore which of the theories under study may best explain the consequences of these threats for desistance. The expected results supporting each theory are as follows: Control theory would be favored if successful trajectories are exclusively linked to the development of conventional adult roles. For cognitive transformation theory, an identity change would protect the person from reoffending even if the person has not obtained a pro-social adult role. Finally, strain-social support theory suggests that even if an individual has not obtained a prosocial adult role, desisters would not fail if they continued to benefit from social support.
Only a minority of the desisters were able to attain a pro-social adult role. Most of these desisters were unemployed or had lost the jobs from their initial release, their emancipation projects had generally been delayed, and they mostly lived with their parents. However, in their narratives, the participants reported that they were lawabiding people who were committed to their new identities. This conventional identity should be viewed in the context of the ongoing support that they received from families and partners. A first dimension of this support is instrumental and prevents individuals from perceiving strain: Have you thought about reoffending? No, not at all. I'd rather look for a job, ask for money from my mother, or my family, or anyone . . . (E28,Desister,26,Interview 2) A second dimension is expressive support: Having support is very important; I'm telling you. Not necessarily the family, but you need a person who can listen to you and advise you, not a person who tells you what to do, because then you'll be dependent. The day you will feel alone, even though it been six years since release, the day you don't have this person, then you will have a problem. (E338,Desister,33,Interview 2) The relevance of social support to threats to desistance can also be confirmed if we examine failure cases (i.e., people who were in a desistance process at the first interview but who reoffended and were reincarcerated during the follow-up period). In some cases, a lack of economic support and the need to provide money for the family in the context of the male breadwinner model and the criminogenic effects of masculinity (Carlsson, 2013) appear to explain derailments: What happened was that there wasn't any money, to maintain myself and to maintain the children and household expenses. Where could I go? To social services? If there are a lot of unemployed people, Spanish included, they have more rights than foreigners, than us. And then how can I manage to live? (E360,Persister,30,Interview 2) Finally, social support was relevant for explaining the maintenance of the desistance process in some middle-aged offenders in whom cognitive transformation was the catalyst for change. These offenders include participants who did not aspire to acquire conventional adult roles because of their age and their long drug abuse trajectory. Rather, their aims were focused on more immediate concerns such as moderating their drug abuse and avoiding opportunities to reoffend. Despite the lack of social bonds, these participants appear more likely to continue to desist because of the formal support that they receive from the state (permanent pensions or temporary benefits, treatment programs) and/or the occasional informal support that acquaintances provide them. However, because such support is scarce and often irregular, it may be insufficient to meet these men's needs, and these men may drift between accepting living in poverty and reoffending. I am a pensioner; I try to live on the pension. A permanent pension? Yes. Which expenses can you afford with this pension? Pay for the room, food, tobacco, not much more than that . . . but only when everything goes well. When it does not . . ., I don't have money to buy food, to pay for the room, you know? I look for an empty house, and I go in . . . (E330,Desister,47,Interview 2) A detailed description of the 36 cases, which is shown in the appendix, indicates that strain-social support theory has a key explanatory role across the different profiles analyzed, whereas the other theories are relevant specifically in certain profiles and situations, such as the youngest interviewees (in the case of control theory), and lateonset participants and the oldest participants with lack of social ties (in the case of cognitive transformation theory).

Balance of the Three Theories
Regarding control theory, specifically, the age-graded theory of informal social control throughout the life course (Laub & Sampson, 2003;Sampson & Laub, 1993), we had the following findings: First, we obtained moderate confirmation with respect to the relevance of objective factors for the origin of the desistance process. Not all of the desisters showed desistance because of social ties, and only some of the social ties that led to desistance corresponded to the turning points delineated by Sampson and Laub (1993). Moreover, desisters took an active role in the origin of the desistance process. Second, we noted moderate confirmation with respect to the relevance of control mechanisms, with the commitment to a new relationship being the more relevant, for explaining the maintenance of desistance. Rather, in our research, other mechanisms such as compensation, moderation of strain, and hooks for change appear to be more explanatory. Third, we failed to confirm the thesis that desistance is conditional on the development of a pro-social adult role, as the participants' failure to reach certain adult life markers (financial independence and family life) did not often lead to breakdowns in the desistance process.
Regarding cognitive transformation theory, we obtained the following findings: First, we reached moderate confirmation with respect to the idea that the desistance process originates from cognitive change (Giordano et al., 2002;Paternoster & Bushway, 2009). Although some participants did begin their change process by critically reflecting on their past, a more prevalent pattern in the sample is that the cognitive change was preceded by certain social factors such as relationships with family or a new romantic partner (these results are similar to the results reported in Dufour et al., 2015). Moreover, the catalytic character of cognitive transformation is observed under specific conditions only-among men who had not accumulated disadvantages during their initial imprisonment and among middle-aged inmates after long periods of imprisonment. Second, we did get strong confirmation for the thesis that hooks of change constitute a mechanism for maintaining desistance. This mechanism is observed not only when cognitive transformation is a catalyst for desistance but also when the person feels motivated by family and/or partner support, devotes time to rehabilitation and educational programs, or seeks employment-that is, when the person takes an active role in the change process. Third, we got moderate confirmation for the idea that once participants have experienced an identity change, they will be able to resist stressful situations without interrupting the desistance process. Nevertheless, the interviewees were able to resist these threats in the context of social support from family and/or partners.
Finally, regarding strain-social support theory, we obtained the following findings: First, we got moderate confirmation for the idea that receiving support serves as a catalyst for desistance. Although support serves as a catalyst when the person is involved in relationships with significant others, other desistance pathways can emerge when participants react to their initial experience of imprisonment with a strong sense of self-efficacy or when they express a feeling of tiredness with respect to being imprisoned again. Second, we found strong confirmation for the suggested mechanisms for maintaining desistance. Compensation and moderation of strain appear to be relevant mechanisms related to partner and/or family support, providing evidence of the relevance of social support for understanding desistance (Cullen, 1994). Other mechanisms such as hooks for change are also relevant in a social support context. Third, we did have strong confirmation for the role of social support in the evolution of participants who began the desistance process but had not obtained a pro-social adult role. These participants were not able to maintain or find work and were not able to form families, but the ongoing support from families or partners prevented them from reoffending.
To summarize, we have found that, on balance, of the three theories under consideration, strain-social support theory is favored by this research. Regarding the three stages of the desistance process-origin, maintenance, and outcome-we found that control theory contributes to the explanation of the origin and maintenance but does not provide a satisfying answer to the consolidation of desistance despite individuals' non-achieving pro-social adult roles. Cognitive transformation theory provides a reasonable description of the mechanisms of change, particularly regarding the attitude to the person in search of hooks for change, but it does not seem able to explain this attitude without accounting for the external factors emphasized by social control and strain-social support theory. Finally, strain-social support theory seems able to explain the whole process in most situations: It gives a prevalent explanation of the origin of change, identifies important mechanisms to maintain change (compensation and moderation of strain), and is able to explain the desistance outcome despite the inability of participants to achieve pro-social adult roles.

Limitations
This research has the following limitations: First, this research is limited to understanding the desistance of men who were convicted of acquisitive offenses. Second, because of the limited number of participants tracked in the follow-up period (36 of the 67 in the original sample), we may not be able to identify other desistance pathways, especially for immigrant participants, who had a very low rate of follow-up participation. And third, the follow-up period may be too limited; thus, it could be extended to acquire more evidence that participants who have not achieved the markers of adult life can still achieve the conventional adult roles found among those who did achieve financial independence and family lives (Dufour et al., 2015).

Implications
The main theoretical implication of the research is that, as suggested by Cullen (1994), strain-social support theory should be emphasized in research and practice for giving a prevalent explanation of the process of desistance among imprisoned populations. Social support is not only a relevant protection mechanism in the framework of strain theory (moderating the strain that originates from not achieving a pro-social adult role) but also a potential theoretical explanation of desistance in itself, as social support received by imprisoned people produces a feeling of reciprocity and a desire of compensation that explains their motivation to initiate and maintain change and that stimulates the person to seek hooks for change. This research confirms the value of the transition into adult roles (economic independence and family formation) and the resulting new commitments as relevant factors and processes in understanding desistance (Laub & Sampson, 2003) but suggests that ongoing social support may explain the stability of the desistance trajectories despite the adverse social circumstances that impede the achievement of conventional adult roles. This finding should be considered to support the suggestion by Ohio Lifecourse Study researchers (Giordano et al., 2002;Schroeder et al., 2010) that in societies in which stable work and family formation are difficult to achieve for people who have accumulated social disadvantage, other sources of desistance may be more relevant. Finally, the research supports the idea from cognitive transformation theory regarding the offender's active role in the desistance process, but it suggests that the context of support may provide offenders with the motivation to take advantage of opportunities to change (treatment and training programs, job offers) that ultimately contribute to their feeling of self-efficacy regarding the success of the desistance process.
The results of the present research may be considered to provide further evidence of the idea, underlined by other researchers (Visher & O'Connell, 2012), that support during imprisonment may explain the emergence of cognitive transformations. The likely novelty of our research is the finding that some mechanisms that explain desistance on release from prison are related not to the achievement of markers of adult life but to ongoing support. If the findings of the present research could be replicated in other contexts, the implication for reentry policies should be to guarantee that during and after imprisonment, every imprisoned person has conventional support-state, community, or family-based-that could activate and sustain desistance. Most persister participants in our study lacked conventional family or partner support, suggesting that other types of personal and instrumental support should be provided in those cases. Our findings support a reintegration program based on mentoring for personal support and strong social policies aimed at providing instrumental support for those prisoners who lack conventional family support. This study generally implies that the more prisoners have access to hooks for change, the more possibilities of desistance will appear.

Acknowledgments
Aina Ibáñez has done very valuable work as a research assistant on the project, carrying out most of the interviews. The research has also benefited from the work of Eugenia Albani, Anna Meléndez, and Ferran Restrepo. The Catalan Prison Administration (and in particular Miquel A. Esteban) has been extremely helpful during the project. We thank the anonymous reviewers for their helpful comments on previous versions of this article that have allowed us to deepen in the analysis. We thank professor Bottoms for his continuous availability to discuss the theoretical interpretation of our findings.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article:  (42) Control/Support Support -E92 (43) Control/Support Support -Late-onset participants E360 (31) b Desister (Support) Desister (Support) Support E255 (33) Control/Support Support -Note. Control = social control theory; Support = Strain-social support theory; Cognitive = cognitive transformation theory. a In these participants, we identify an initiation of a process of desistance based on the theory indicated in brackets; however, this process has not been maintained, and at the first interview, they have a persistence narrative. The mechanisms that explain persistence in these cases are the lack of control mechanisms (E69) or the lack of hooks for change (E186 and E202). b These participants have started and maintained desistance during early release (based on mechanisms of the theory indicated in brackets) but have failed afterwards.  Cullen (1994), social support theory has two dimensions in relation to desistance. First, it may be seen as a development of strain theory, providing a factorsocial support-that may avoid the emergence of strain or the production of its effects. Conversely, social support theory offers psycho-social processes-such as the process of compensation to those who provide support-that are autonomous with respect to strain theory. To demonstrate that social support theory emerges from strain theory but incorporates some autonomous mechanisms, we use the expression strain-social support theory. 2. Unless otherwise stated, the references to control theory in the present article refer to the "Age-graded theory of informal social control" developed by Laub and Sampson (Laub & Sampson, 2003;Sampson & Laub, 1993). 3. In a further contribution to cognitive transformation theory, Giordano and colleagues distinguish between openness to change that is produced in the context of new adult relationships (turning points, in the terminology of Sampson and Laub's control theory) and openness to change that is not linked to the development of these new relationships.
Although the second process is also social and explainable from an interactionist perspective, the authors stated that "… we do not have access to the myriad of social situations that may have literally produced these types of transformations" (Giordano, Schroeder, & Cernkovich, 2007, p. 1627). 4. Two previous articles have been published on this research. The first one is based on the first wave sample, using interviews that occurred during the finals months of the prison sentence, and analyzes the differences between desistance and persistence narratives and their origins (Cid & Martí, 2012). The second, which considers the two samples of data, is specifically focused on prison visitation by relatives and its effects on recidivism (Martí & Cid, 2015). 5. One point of concern raised by two anonymous reviewers is the extent to which the underrepresented non-recidivist are those with better background and more social bonds that would have made them easier to be contacted and reinterviewed. This fact would have biased the results of the analysis. To address this issue, we have compared the subgroups of non-recidivists in the first and second waves. Results indicate that although the difference regarding nationality is important (45% of non-recidivists are foreigners in the first wave vs. 24% in the second wave), the subgroups remain similar in relation to other characteristics, that is, mean age at release (36.2 vs. 35.6 years), early release (72% vs. 71%), drug consumption (11% vs. 14%), having contact with parents during imprisonment (55% vs. 57%) and/or with a romantic partner (38% vs. 43%), mean age at first imprisonment (28.2 vs. 26.6 years), and proportion of adult life in prison (34% vs. 33% above mean). 6. As suggested by one anonymous reviewer, this quote may reflect the long-term impact of probation work emphasized in the research of Farrall, Hunter, Sharpe, and Calverley (2014). 7. Using reincarceration as a measure of desistance may be problematic if some of the participants who were not reincarcerated committed crimes but were able to avoid incarceration. This situation characterized two participants at the time of the second interview, but these two participants who reported reoffending were later incarcerated. Regarding the desisters, we found two situations: Most of them reported no criminal offenses since their sentences had expired; other participants reported some criminal offenses (or other kinds of illegal behavior), but compared with their previous criminal careers, these offenses were less serious and, in principle, not imprisonable.