|Home | About | Journals | Submit | Contact Us | Français|
Cultural sociologists and other social scientists have increasingly used the concept of narrative as a theoretical tool to understand how individuals make sense of the links between their past, present, and future, how individuals construct social identities from cultural building blocks, and how culture shapes social action and individual behavior. Despite its richness, we contend that the narratives literature has yet to grapple with narrative change and stability when structural constraints or barriers challenge personal narratives and narrative identities. Particularly for marginalized groups, the potential incompatibility of personal narratives with daily experiences raises questions about the capacity of narratives to influence behavior and decision-making. In this study we draw on prospective longitudinal data on the reentry narratives and narrative identities of former prisoners to understand how narratives do and not change when confronted with contradictory experiences and structural constraints. We identify and describe the processes generating narrative change and stability among our subjects. These findings inform a framework for studying narrative change and stability based on four factors: the content of the narrative itself, the structural circumstances experienced by the individual, the institutional contexts in which the individual is embedded, and the social networks in which the individual is embedded.
Cultural sociologists and other social scientists have increasingly used the concept of narrative as a theoretical tool to understand how individuals make sense of the links between their past, present, and future; how individuals construct social identities from cultural building blocks; and how culture shapes social action and individual behavior. For example, Moon (2012) shows that different narratives of collective selfhood among stigmatized groups contain different implicit assumptions about the nature of the self, the boundaries between self and other, and the sources of personal change, and that these differences lead to distinct preferences for group action. Based on a study of stories of resistance to legal authority, Ewick and Silbey (2003) argue that stories about resistance reveal how social structures work to distribute power and advantage, and provide openings to extend that resistance. Maruna (2001) analyzes the desistance narratives of criminals and finds that those whose life stories contained an exaggerated sense of agency and conception of a true-self as emerging from past misdeeds were more likely to desist from crime. Across the disparate research topics to which the narrative concept has been applied, there appears to be remarkable consistency in the core elements of a narrative, the role of “narrative templates” (Ewick and Silbey 2003) or “meta-narratives” (Alkon and Traugot 2008) in the construction of individual narratives, and the importance of narrative identity for “guiding” (Somers 1994), constraining, or enabling action (Polleta et al 2011).
Despite its richness, we contend that the narratives literature has yet to grapple with narrative change and stability when structural constraints or barriers challenge personal narratives and narrative identities. Although how individuals change their personal narratives is implicitly or explicitly a part of many literatures, from religious conversion (e.g. Yang and Abel 2014) to political mobilization (e.g. Davis 2002) to the experience of illness (Hyden 1997; Bell 2000) and mental health and substance use recovery (e.g., Estroff et al. 1991; McIntosh and McKeganey 2000a, 2001), the stability of narratives is rarely explicitly addressed, especially the question of what happens when structural constraints generate experiences that might be interpreted as counter to one’s narrative. Particularly for marginalized groups with limited access to the means with which to realize narrative prescriptions and who face numerous structural constraints, the potential incompatibility of personal narratives with daily experiences raises questions about the capacity of narratives to influence behavior and decision-making. More specifically, if narratives are thought to guide, enable, or constrain action, when do narratives as templates for action remain stable despite contradictory experiences or circumstances? What happens, if anything, to an individual’s narrative or narrative identity when experiences do not conform to the expectations encoded in the narrative?
The answers to such questions are far from obvious. On the one hand, new information that challenges our cognitive categorizations can, under the right circumstances, lead us to change or abandon seemingly incongruent ideas. For instance, the availability of multiple narratives in one’s cultural and institutional environment might provide ample opportunity to employ new narratives when old ones do not seem to fit. On the other hand, the human mind has considerable capacity to interpret new information in ways that are consistent with existing schema and is also capable of holding multiple contradictory ideas. Moreover, some have argued that much of individual behavior is governed by unconscious motivations that are culturally influenced (Vaisey 2009, Martin 2010). However, narratives exist in conscious thought, memory, and expression and may simply be post-hoc justifications for behavior, and therefore highly unstable. Finally, some narratives may be so vague or incoherent that they resist challenges posed by new information or experiences (Polleta et al. 2011).
In this study we draw on prospective longitudinal data on the reentry narratives and narrative identities of former prisoners both before and after release in order to understand how narratives do and do not change when confronted with contradictory experiences and structural constraints. Former prisoners provide a particularly compelling population in which to study narrative change and stability in the face of structural constraints because they are a racially and economically marginalized population that grapples with reconstituting their social and moral identities, has considerable time for self-reflection and narrative construction in an institutional context in which change is openly discussed and normatively encouraged, and faces stigma and other structural challenges to their new identities and change narratives after release. The reentry narratives we describe are attempts to wrestle with and explain past criminal behaviors and substance abuse, identify a motivation for change, and provide a script for maintaining the changed self in the future. They represent the interplay between structure and agency in a highly marginalized population.
We examine our subjects’ reentry narratives before release and in the period immediately following release, and then track changes in these narratives over time as our subjects experience the difficulties of prisoner reentry and reintegration into free society and frequent challenges to their redemption claims and new narrative identities. Previous research on narratives in the context of desistance from crime has emphasized the importance of cognitive transformation or “hooks for change” (Giordano, Cernkovish, and Rudoph 2002), and features of narratives common among desisters as compared to non-desisters (Maruna 2001). However, these studies employ retrospective life-story narratives that cannot address narrative change in the early periods of attempted desistance. Put differently, the construction of new narratives and the development of new identities is likely a circuitous process marked by fits and starts, a process that would be masked in purely retrospective accounts. Moreover, the primary account of the role of narratives in desistance from crime focuses almost exclusively on the content of the narrative and its implications for identity transformation (Maruna 2001). In contrast, our analysis will suggest that, in addition to narrative content, social structures, institutions, and networks condition whether and how narratives relate to changes in identity and behavior.
Our results indicate that narrative stability is relatively common, even in the face of significant structural, social, and institutional challenges, but that some elements of a narrative are more stable than others, particularly those elements that help people to make sense of the past. We inductively identify seven processes that generate narrative change or stability. We emphasize that these processes tended to act in concert rather than independently. We draw on these findings to develop a conceptual framework for understanding narrative change and stability that focuses on four factors that determine narrative change and stability: the content of the narrative, an individual’s structural circumstances, the institutional contexts in which the individual is embedded, and the social networks in which the individual is embedded. That each of these four factors is involved in both narrative change and stability suggests that these are highly contingent and interactive processes.
We begin by reviewing prior theoretical and empirical work on the nature of narratives, narrative identity, and the role of narratives in shaping behavior. We then discuss prior work on narratives and reentry among former prisoners, and argue that this population provides fruitful analytical leverage for investigating narrative change and stability. Then we describe our own data on former prisoners and our methods of analysis. Our results section analyzes how and why our subjects’ narratives do or do not change over time after release. In the conclusion, we develop the implications of our analysis for conceptualizing narrative change and stability, and we discuss the implications of our work for the future study of narrative identity and the links between narratives and behavior. We also discuss implications for the burgeoning literature on prisoner reentry and for racially and economically marginalized populations more generally.
Although scholars bring diverse perspectives and empirical interests to the study of narratives, narratives tend to have three interrelated core characteristics. The first is that narratives are inherently temporal; stories unfold over time as a series of key events (Franzosi 2010). Thus, narratives can serve as a cognitive filter, highlighting some events, scenes, “themes” or “motifs” as particularly salient or relevant in processes of “selective appropriation” (Ewick and Silbey 2003, Somers 1994). Indeed, narratives do not simply describe what happened in the past but are active in shaping our perceptions of the “facts” of past events (Orbush 1997). The second is that we use narratives to make sense of the past, present, and future, and the links between them (see also Mische 2009, Frye 2011 on the related notion of imagined futures). As Somers (1994) writes, “experience is constituted through narratives.” Stories provide individuals with ways of coping with stressful or emotionally charged events by giving a “sense of control and understanding of the environment” (Orbuch 1997: 459). The third is that a narrative has a plot. It not only describes the events that occur but also provides an interpretation of the causal links in the story. This idea has been variously described as “emplotment,” “relationality of parts,” “narrative causality,” or “story structure” (Ewick and Silbey 2003, Somers 1994, Polleta et al 2011).
Individuals do not develop narratives in isolation, but rather they construct narratives from culturally available “narrative templates,” “public narratives” or “meta-narratives” (Somers 1994, Ewick and Silbey 2003, Alkon and Traugot 2008) and interactively in conversation with others (Ewick and Silbey 2003). As Polleta et al (2011) argue, the plots of narratives are familiar, as the wider culture provides individuals with a “cultural stock of plots.”1 This suggests that individuals have at their disposal more narratives than they deploy at any particular point in time. However, individuals do not simply mimic existing stories. They revise and reinterpret them to suit their own needs and experiences and those of their intended audiences, another instance of “selective appropriation.” Narratives may or may not conform to the experiences or facts as interpreted by others (Maruna 2001), but nevertheless are thought to affect action through their capacity to structure meaning and understanding, as we further discuss below. Narratives may also be challenged by structural realities, as the reform narratives of former prisoners often are.
Finally, institutions play an important role in the preservation and distribution of such narrative templates. For example, addiction recovery programs are an important source of “self-help” narratives among former prisoners (Leverentz 2010, 2014), which are linked to broader cultural narratives of therapeutic “self-transformation” (Silva 2014) or “self-blame” (Sharone (2013) and provide “cognitive blueprints” for future action (Giordano et al 2002).2 Schools, religious institutions, political parties and many other formal and informal social groups promulgate, evaluate, and legitimate various narrative conventions. Polleta et al. (2011) suggest that institutional conventions shape narratives because storytelling is a way that organizational logics are communicated and because individuals within institutions (employees, clients, patients, prisoners) are evaluated based on whether the stories they tell are consistent with these logics (see also Jacobs and Sobieraj 2007). Institutions – whether they are welfare bureaucracies, courts of law, or workplaces – may impose particular narratives and “discipline” those who fail to conform (Foucault 1977; Fox 1999; Rose 2000).
Narratives provide individuals with a framework in which to make sense of their identities and social positions. Narratives “constitute our social identities” (Somers 1994: 606) by relating the teller of the narrative to others (Ewick and Silbey 2003).3 In telling ourselves and others who we are and in laying claim to particular experiences and group memberships (who we are like and unlike), narratives define, assign and reinforce identities. Narratives also embed identities in a particular time and space. For example, Abelman (2008) argues that women’s mobility narratives in South Korea are constitutive of personality. Her subjects drew from an available “archeology” of narratives and personalities that were specific to the particular moment and culture. The role of institutions in narrative construction provides a link between institutions and identities, as institutions shape narratives and narratives constitute identities.
Narratives have been theorized to play an important role in linking culture to individual action, because their culturally-informed interpretations of the past shape expectations for the future. As Somers (1994) explains, “People are guided to act in certain ways and not others on the basis of the projections, expectations, and memories derived from a multiplicity but ultimately limited repertoire of available social, public, and cultural narratives” (614). For example, Moon (2012) delineates four alternative narratives of collective selfhood among stigmatized groups and shows how different narratives lead to distinct preferred causes of action. Differential access to distinct narratives based on the institutions and cultural environments with which they interact (Ewick and Silbey 2003) may help to explain differences in social action.
Previous research on narratives has identified a number of dimensions on which the content of narratives may vary, a crucial consideration as we take up the question of which narratives change or remain stable. Narratives may vary in the emotions that they prompt (Orbuch 1997), the degree of agency or constraint that they imply (Ewick and Silbey 2003, Polletta et al 2011, Leverentz 2010, Lempert 2004), in their coherence or specificity (Leverentz 2010, Polleta et al. 2011), and in their normative implications or moral components (Ewick and Silbey 2003, Polleta et al. 2011). Different narratives may also assign different categories of blame or moral redemption to their actors. Each of these narrative components may have implications for narrative resilience. For instance, it may be that more specific or coherent narratives are more susceptible to challenge, whereas less coherent or specific narratives are more resilient, as they can be adapted to new experiences, new information, or alternative narratives of the same events (Polleta et al. 2011). Alternatively, more detailed narratives may have stronger implications for future action or behavior (Giordano et al 2002), making them more useful and less likely to be abandoned.
Although this prior work suggests potential hypotheses about narrative change and stability, we have been unable to locate any prior research that explicitly examines or theorizes processes of change or stability in personal or self-narratives. In other words, while the content of most narratives is about change, the question of how narratives themselves change or remain stable has received little attention. Moreover, although prior research has addressed the question of how marginalized groups understand their experiences with structural constraints and disadvantages, that literature has been primarily cross-sectional rather than focused on change and stability in these understandings over time (e.g. Young 2004, Edin and Kefalas 1995, Edin and Nelson 2013).
Related literatures provide some guidance on processes of identity change, however. The literature on religious conversion, for example, shows the importance of key moments of crisis for identity reconstruction (e.g. Snow and Machalek 1984, Maruna, Wilson, and Curran 2006, Yang and Abel 2014), the role of social relationships in conversion (e.g. Lofland and Stark 1965, Snow and Machalek 1984), the role of institutional templates in guiding the conversion process (Snow and Machalek 1984), and the “biographical reconstruction” that accompanies religious identity change (Snow and Machalek 1983). Likewise, medical sociologists interested in the experience of symptoms and disability have shown how illness—particularly chronic illness—can present as a “biographical disruption” requiring individuals to grapple with changes in identity and engage in narrative reconstruction (Bury 1982, 1991, Charmaz 1983, Williams 1984, Corbin and Strauss 1991), negotiating their lived experience with, for example, biomedical understandings of their condition (Mishler 1984, Kleinman 1988, Frank 1995, 1996), stigma (1996), stigma (Schneider and Conrad 1980, Weitz 1991, Jacoby 1994), and collective illness identities and experiences (Zola 1982, Barker 2008, 2010, Brown et al. 2010). Similar themes can be found in research that explores recovery processes for those with histories of addiction (Waldorf 1983, Biernacki 1986, McIntosh and McKeganey 2000a, 2001, Reith and Dobbie 2012, Andersen 2015) and/or mental illness (Estroff et al. 1991, Karp 1996, Ridge and Ziebland 2006, Cardano 2010, Thompson 2012) — individuals who also commonly confront stigma and biomedical and institutionalized interpretations of their conditions as they engage in identity reconstruction. We will see parallels to these ideas in our analysis of change and stability in reentry narratives.
Narrative can be a useful conceptual tool for understanding the cognitive and identity changes that may accompany incarceration and reentry. Prison is often a time for reflection, as prisoners try to make sense of the course their lives have taken, the consequences of their incarceration, and how to change their lives (Comfort 2012). Incarceration can be a catalyst for identity reconstruction, as “prisoners … face a crisis of self-narrative” due to the social and psychological trauma of separation from others and the prison environment (Maruna et al 2006: 168). Individuals exiting prison and re-entering free society likewise grapple with their past, present and future. Assuming they intend to avoid criminal activity and return to prison, as most do, they are also faced with whether and how to develop a narrative account that specifies how and why their life trajectories will change (Maruna 2001).
They also face constructing new identities for themselves that supplant their “spoiled identity” (Goffman 1963) as a felon or prisoner and its accompanying stigma (Pager 2007). Because prisoners are cast as immoral, their reentry narratives may make claims to moral change. One way to do this is develop a narrative that conforms to narrative conventions or templates of redemption or reform. The audience for such narratives is not just the former prisoner him or herself but others, such as family, potential employers, or agents of the justice system whose decisions hold sway over post-prison life. Further, former prisoners re-entering free society face frequent challenges to their change narratives as a result of involvement with institutions of the criminal justice system and social service bureaucracies, which provide competing narratives that essentialize them as permanently morally suspect.
Moreover, stigma, structural constraints such as lack of job opportunities, and other negative experiences during reentry may challenge the newly developed narratives of former prisoners. Barriers returning prisoners face to finding stable sources of employment, public assistance, and social support (Holzer, Raphael, and Stoll 2004, 2007, Pager 2003, 2007, Pager, Western, and Bonikowski 2009) as well as the disadvantages that characterize this population, including low levels of human capital and a high prevalence of mental health problems and substance use (Visher and Travis 2003), all make stability and security a significant challenge. Incarceration may erode human capital, as skills decline, a gap in the work record is established, diseases and psychological disorders are exacerbated, and behaviors learned for survival in prison conflict with workforce norms (Bushway, Stoll, and Weiman 2007). Even if employment is established, it may be difficult for ex-offenders to maintain (Pettit and Lyons 2007, Sabol 2007, Tyler and Kling 2007). These challenges can be understood as consequences of a broader shift in institutional logics toward punishment, regulation, and social control of marginalized populations such as poor people of color (Garland 2001, Simon 2007, Wacquant 2008, 2009). Moreover, high rates of incarceration over the last four decades have been implicated in the recent rise in inequality and persistent racial inequalities (Western 2006). For these reasons, the process of prisoner reentry and reintegration is a fruitful site to study narrative change and stability among marginalized populations.
One theory of the process by which such cognitive changes occur and their role in desistance is developed by Giordano et al (2002) based on a retrospective qualitative study of former juvenile delinquents in Ohio interviewed in adulthood. Giordano et al argue that cognitive transformation is a four step process. (1) Openness to change leads to (2) exposure and receptivity to “hooks for change” provided in the environment, such as prison or treatment, religion, employment, parenthood, or marriage. “Latching onto” these hooks for change leads to (3) identity transformation which in turn leads to (4) a change in the meaning and desirability of deviance and crime. Possible hooks for change vary in their “transformative potential” based on their capacity to provide a “clear pro-social cognitive blueprint,” a new identity, a template for future behavior, a link to “positively valued themes,” and access to non-deviant others who can reinforce the new identity, such as peers, romantic partners, or family members. Although not specifically cast in terms of the narrative concept we describe above, Giordano et al.’s (2002) model contains many of the same elements and ideas, including a focus on cognitive models of change, expectations for future behavior, institutionally or culturally provided templates or blueprints, and new identities. From the perspective of the narratives literature, it adds the possible role of structural positions and significant others in reinforcing or challenging cognitive models. We build on this work by empirically examining the development of narratives over time in the early stages of possible desistance, particularly in the face of structural constraints and other challenges to reentry narratives.
In Making Good, Maruna (2001) compares “life-story narratives” (McAdams 2001) of desisters and non-desisters collected a single point in time to understand which narratives lead to desistance from crime. He argues that the development of a “coherent, pro-social identity” is required for desistance and that such identities must be based on a life narrative that “provides unity, purpose, and meaning” (Maruna 2001: 7–8). These life narratives have two key features, an account of the past and an account of change, particularly how the problems of the past led to the current reformed self. Maruna’s psychologically-motivated analysis aims to “identify the common psychosocial structure underlying these self-stories” (8). He finds that narratives of successful desisters typically featured an “exaggerated sense of control over the future” and a “missionary zeal or purpose,” that “recast” their criminal pasts as the necessary prelude to their current lives and emergence of the “true self” (9). Maruna emphasizes that the narratives that precipitated desistance were overly optimistic given the reality of offenders’ lives, the result of a “process of willful cognitive distortion” (9).4 Maruna’s work suggests that some narratives may be insulated from challenge because the content of the narrative downplays barriers and emphasizes agency. In this paper we build upon this work by prospectively analyzing change and stability of ex-offenders’ narratives and in so-doing incorporate the role of structural constraints into the literature on reentry narratives and narrative identities. As we will see below, while the content of a narrative is indeed crucial, so too is one’s capacity and opportunity to realize the plans it sets forth.
In sum, when considered together, the wider literatures on narratives and related forms of identity change as well as the literature on cognitive change among ex-offenders suggest a number of processes that might influence change or stability of reentry narratives. Variation across narratives in their specificity or ambiguity and in how well they conform to existing narrative templates and institutional logics suggests that the content of a narrative will play a key role in its susceptibility to challenge and its resilience in the face of challenge. The role of institutions in providing narrative templates as well as schema for evaluating them suggests that the institutions in which an individual is embedded will influence narrative change or stability. Because narratives are constructed in dialogue with others, often with the goal of justification to others of one’s actions, the social networks in which an individual is embedded may impact which narratives are reinforced or undermined. Finally, because narratives are used to understand one’s experiences and social positions, experiences with structural constraints and roles that do or do not conform to a narrative’s expectations may help to explain narrative change or stability. In the next section, we describe the data we use to examine such processes and the methods we use to analyze those data.
We take an inductive approach, relying upon qualitative methods as those best suited to uncover diverse and complex social processes. In qualitative interviews, the researcher can begin to reveal the subject’s understanding of his or her experiences, gather data on the details of those experiences, and explore if and how the processes suggested in the literature square with the subject’s experiences and conceptualizations (Lofland & Lofland 1995). Our data come from in-depth longitudinal qualitative interviews that probe the social, economic, and cultural processes related to prisoner reentry and criminal desistance. The sample of 15 male and seven female interview subjects was selected from Michigan Department of Corrections’ (MDOC) administrative records based on their expected release date (those who would be released within two months of the baseline interview and had already been granted parole or who would reach the end of their sentences) and release county (four counties in Southeast Michigan).
We intentionally chose to study a small number of subjects intensively over a relatively long period of time for three reasons. First, a longitudinal design is necessary in a study of released prisoners due to the rapidly changing nature of their lives. Reentry is a period of significant flux, and ex-offenders’ experiences immediately after release may be very different from their experiences months and years later. Second, a longer follow-up allows for the observation of change or stability in narratives that takes time to develop. Third, frequent interviews are required to capture the processes driving narrative change or stability over time. Fourth, frequent interviews can help to increase subject retention in this hard to study population. Fifth, frequent interviews are important for building and maintaining rapport, and to facilitate the discussion of unconventional, deviant, or illegal behaviors.
Interviews covered a diverse array of topics, both researcher and subject driven, but focused on the subject’s life histories, community context, family roles and relationships, criminal activities and experiences, labor market experiences, life in prison, service use, and health and well-being, including drug and alcohol abuse. Interviews were unstructured, meaning that we prepared a detailed interview protocol with a lengthy list of questions and follow-up probes on the aforementioned topics but let the conversation follow the interests and experiences of the subject. Initial in-prison interviews were roughly 90 minutes, while follow-up interviews usually lasted 1–2 hours. Our research design captures subjects both directly before release, allowing for investigation of subject’s pre-release expectations, and during the first 2–3 years after release, a critical period for desistance (National Research Council of the National Academies 2007).
As other qualitative studies of former prisoners have found (e.g. Leverentz 2014), drug and alcohol addiction was also common in our sample, and these issues played an important role in their attempts to understand their past experiences and develop future plans. Six of our 15 male subjects characterized themselves as alcoholics, five as both drug abusers and alcoholics, three as drug abusers solely, and only one reported no addiction to drugs or alcohol. Of seven female subjects, four characterized themselves as drug addicted, one as drug and alcohol addicted, and one as formerly drug addicted. Only one woman did not describe a serious current or past problem with drugs or alcohol. The prevalence of addiction is significant for our analysis of reentry narratives because, as the literature on addiction recovery shows, the embodied experience of addiction – in which someone's social roles, daily activities, and identity are disrupted – often leads to a loss of self and necessitates engagement in a process of biographical reconstruction (Biernacki 1986, McIntosh and McKeganey 2000a, Gibson et al. 2004, Reith and Dobbie 2012). Although the severity of their addictions varies, many of our subjects are faced with such biographical reconstruction as they develop their reentry narratives in preparation for release from prison.
Because prior research provides little guidance on the nature or causes of narrative stability, our data analysis was largely inductive. Our analysis alternated between two parallel forms, a subject-based mode, which considers the details of each case, and cross-case mode, which looks for patterns across individuals or narratives. We conducted our analysis in three stages.6 In the first stage, we coded, synthesized and summarized each subject’s initial narrative or narratives that emerged from the pre-release interview and post-prison interviews at one and two months after release. We define a narrative according to three components discussed in the narratives literature, (a) an account of past, present, and future events that (b) links these events both temporally and causally, providing (c) an interpretation of the links between past, present, and future. Applying this conceptualization to “reentry” narratives, we focus on three core elements, the subject’s interpretation of (1) the sources of past problems that led to criminal behavior and/or incarceration, (2) the impetus for change that would put the subject on a new life trajectory, and (3) the script for moving forward with a new life after prison.7 We also examined the narrative identity that the subject linked with each narrative. Such “reentry narratives” are distinct from “life-history” narratives conceptualized by McAdams (2001), in that they are more temporally focused self-stories about key events that contain both a script for moving forward and a new identity.
Although most prior studies conceive of narratives as implicitly projecting a future, our conceptualization departs somewhat from prior work by explicitly including scripts for moving forward, which are comprised of not only future goals but, more importantly, the steps that our subjects imagine taking in order to accomplish those goals. This approach to narratives is consistent with recent work by scholars such as Mische (2009, 2014) and Gibson (2011a,b), who focus on future-oriented narratives as a tool for understanding diverse social phenomenon. Indeed, Mische (2014) argues that narrative analysis is one of the primary methods through which sociologists can study imagined or projected futures. Our inclusion of scripts for moving forward in our analysis also reflects the inherent future orientation of reentry narratives, as an account detailing a break with the past and the establishment of a new identity. This conception of narrative is consistent with prior research on the narratives of those involved in the criminal justice system (Maruna 2001, Giordano et al 2002). For example, one can see a similar set of narrative elements in Maruna et al.’s (2006) analysis of prison conversion narratives, which include an account of past experiences and behaviors, an explanation of the moment of change, and plans for the future. Similarly, Comfort (2012) describes how young men recently released from prison interpret their prison experiences in a forward-looking way, as preparing them for their future lives.
In the first stage of the analysis, we identified the core elements of the initial reentry narratives through subjects’ descriptions of their pre-prison criminal behavior, criminal justice involvement, and substance use. For instance, the interview protocol contained questions about how subjects ended up in prison, what their lives were like growing up, how they imagined their lives would be different after release, how this release from prison would be similar or different from past releases (for those who had been to prison multiple times), how likely they were to return to prison and why, what subjects most worried about regarding their reentry and reintegration, and what they saw themselves doing at specific time intervals after release. These questions were asked repeatedly throughout the study to see whether and how subjects’ narratives changed over time after release. Unsurprisingly, all subjects viewed their criminal justice involvement as a problem they wished to avoid in the future, and thus, narratives frequently emerged through descriptions of how they ended up in prison and their strategies for not returning. However, we also identified narrative elements through discussions prompted by interview questions in specific domains when such domains were central to the subject’s narrative (e.g., questions about social support, employment, and mental health, depending on the particular subject). We also combed through the initial interviews for linguistic devices subjects used to signal the narrative elements of interest. For instance, language such as, “It all started when…” indicated subjects’ beliefs regarding the source of the problem or, “Things are going to be different this time because…” signaled an impetus for change.
In the second stage of the analysis, we used subsequent interviews to examine change and stability in individual subjects’ initial narratives. For each initial narrative for each subject, we identified the central domains involved in the narrative (e.g. substance use, peers, employment) and social roles or identities central to the narrative (e.g. patient, parent) and used these to develop a set of subject-specific codes. Coding of individual subject trajectories involved documenting continuity and change in how the subjects talked about these domains and roles or identities and their actions, thoughts, and decision-making related to these domains, roles and identities. Any interview data related to the domains of interest for a given subject were coded, with coded data summarized for each interview. When new narrative elements developed over the course of the interviews, these narratives were also coded, summarized, and tracked in future interviews. Two research assistants were trained in the meanings of the codes and rules of their application to assist in coding. Research assistants recoded transcripts that had been previously coded by one of the authors until a high degree of agreement occurred. All coding by research assistants was also reviewed by at least two of the authors.
Coding was then summarized for each subject across interviews in terms of stability and change of the initial narratives. The summaries focused on change and stability in each of the three narrative components (problem definition, impetus for change, script for moving forward), as well as all behaviors and experiences related to the individual’s narrative(s). For example, key changes in social support across interviews were summarized for subjects whose past narratives focused on the role of negative influences from others as the causes of their trouble, in addition to tracking changes in the negative influences narrative (which could remain stable despite changes in actual experiences of social support). Similarly, changes in narratives were tracked, for instance, when subjects switched from a script in which assuming personal responsibility and gaining employment was critical for moving forward to a script that involved relying instead on public aid, which happened when a criminal record came to be viewed as an impediment to stable employment. Throughout this process, we were careful to attend to contradictory beliefs or statements as well as to distinctions between subjects’ understandings and interpretations of their problems and their reported behaviors and experiences, tracking both separately in our coding. For instance, some subjects reported regular substance use, but did not attribute their problems to such substance use or develop an addiction narrative.
In the third stage of analysis, we sought to understand the processes that led to either narrative change or narrative stability for each subject. We drew on the completed summaries to examine critical moments either when we saw narrative change or when such change might be expected given behaviors or experiences that might contradict subjects’ narratives. We returned to the coded trajectories to identify subjects’ descriptions and interpretations of the experiences, contexts, and social ties that alternately reinforced or challenged their initial narratives. When trying to understand narrative change, we first sought to understand how the subject herself accounted for a change in narrative, but we also examined changes in experiences, contexts, and institutions related to both the new and old narratives. When trying to understand narrative stability, we first had to identify moments of possible challenge to the narrative and then examine salient behaviors and experiences related to those moments and to the narrative.
From these analyses, we developed a typology characterizing the processes underlying narrative change and narrative stability across research subjects. Assignments of research subjects to types were made by all authors separately and then cross-checked together. Our main goal in using such an approach was to characterize the sources of narrative change and stability across research subjects and over time. This synthesis of the data is what is elaborated in the results section below. While identifying and describing narratives and narrative change and stability requires considerable simplification and interpretation of complex patterns, the narratives we describe nonetheless broadly capture significant distinctions and larger patterns in our longitudinal data.
Because of the inductive nature of our analysis, this third stage might be susceptible to falsely attributing stability to what we have called stability processes or falsely attributing change to what we have called change processes, simply because subjects who experienced stability processes also experienced narrative stability and subjects who experienced change processes also experienced narrative change. We guarded against this possibility by taking a “process-tracing” approach (Mahoney 1999, 2000) to our analysis of each subject’s experiences and narratives (see also Tavory and Timmermans  on “a continuous stream of events in sequences,” and Lichterman and Reed  on “chains of action”). In process tracing, the researcher uses the in-depth information about the case or subject to understand how causes and effects are linked. Here we implemented the process tracing approach with two interrelated strategies. First, we focused on moments of potential challenge to narratives and narrative identities, moments of particular salience to the subject (as evidenced, for example, by their focus on them during interviews) related to the key domains of their narratives that they struggled to understand and interpret. Second, the longitudinal nature of our data allowed us to look at change or stability of key processes and examine how they corresponded in time to change or stability in narratives. Combining these two strategies allowed us to examine how salient experiences did or did not lead to narrative change.
Finally, we know that the stories people tell are affected by the contexts in which they tell them (Holstein and Gubrium 2000). Indeed, one conclusion of our analysis is that narratives can be heavily shaped by narrators’ positioning in institutions, networks, and social structures. In this vein, a concern for this analysis is the extent to which the interview setting and/or the interviewer may have affected the way our interview subjects talked about – and even thought about – their lives. This might be particularly true with regard to stigmatized criminal and substance use behavior, but may have been true of other domains as well.
We dealt with this potential pitfall in multiple ways. First, because data collection with each subject occurred across multiple settings over time, we could observe whether changes in the interviewer-interviewee relationship or interview setting shifted how subjects told narratives. Although this did not fundamentally alter the interview dynamic, it created opportunities for subjects to change self-narratives away from those that might resonate with the projected preferences of the interviewer. Yet we did not observe such changes occurring systematically in relation to interview rapport or setting. In some cases, subjects discussed stigmatized behavior such as substance use only in later interviews, but they did not change their narratives related to this behavior. Second, because interviews were unstructured and subject-driven, the subjects played a key role in determining how the interview unfolded and what issues were particularly salient for their narratives. Third, examining the narratives of 22 former prisoners allowed us to examine the diversity and difference in narratives told within the interview context. These narratives did not uniformly conform to conventional understandings of substance use and criminal justice involvement. Ultimately, we argue that our subjects exercised interpretive discretion in how they drew on their experiences and the vocabularies through which they were told “all the while constrained, but not completely controlled, by the working conditions of the moment” (Holstein and Gubrium 2000, p155).
Our objective in this section is to identify, characterize, and analyze the processes that drive change and stability in narratives and narrative identity over time among our subjects. Using the inductive process described in the methods section above, we track narrative change and stability and then identify and describe seven processes that generate change and stability. Table 1 briefly summarizes and describes these seven processes, details whether they contribute to change or stability, and documents the source of their effects (structural circumstances, institutional embeddedness, informal social networks, or narrative content). We organize the presentation of our results below according to these four sources of effects, discussing how each source can lead to change and then how it can lead to stability. Possible reasons for the absence of cases in which one’s structural position led to narrative stability are addressed in the conclusion.
Before we describe and analyze the stability and change processes we have identified, we note two more general features of narrative change and stability that we observe among our subjects. The first is that our subjects exhibited less change in their narratives than we – perhaps naively – expected given the challenges they face in realizing the futures they imagined at release. Thirteen of 21 subjects exhibited no change in any of the three components of their initial narratives, even though most of them experienced what could have been interpreted as significant challenges to these narratives, including failure to secure employment, re-incarceration, drug relapse, and separation from loved ones and other social supports.
Second, when narrative change did occur, most change occurred in the script for moving forward rather than in the subject’s understanding of the sources of their past problems that led to incarceration and the accompanying impetus for change. Of the eight subjects who exhibited change in their narratives, only three changed their interpretation of the source of the problems that led to their incarceration. Although changes in a narrative’s problem definition were accompanied by a change in the script for moving forward, the opposite was not always the case, as new scripts for moving forward were understood as compatible with the original problem definition. This finding indicates the importance of considering the elements of a narrative separately for the purposes of analysis and suggests that links between elements are not always as strong as they initially appear. As we will see below, it also suggests that narrative identities may be linked more tightly to some narrative elements than others.
We also note at the outset that the importance of “narrative” templates or the “cultural stock of plots” is readily evident in the examples we provide below, particularly the institutionalized practices of the prison and of substance abuse treatment programs (see also Leverentz 2014, and Watkins-Hayes, Pittman-Gay, and Beaman 2012 on institutional frames). The role of these institutionalized practices and common stock of plots is likely strong here because almost everyone in our sample had to be approved for release by the parole board, which may look for evidence of particular narratives of individual guilt and change prior to granting parole. It is thus not surprising that most of the initial narratives at prison release downplay the importance of structural barriers and emphasize the role of individual agency and culpability. In other words, our subjects’ narratives were likely well-rehearsed and highly institutionalized accounts.8
One process by which we observed narratives changing over time is what we term Structural Incompatibility, which occurs when the individual’s structural circumstances are so constraining that the script for moving forward cannot be realized, and the subject questions or re-evaluates the initial narrative as a result. Structural circumstances include both material conditions (e.g. poverty, homelessness), as well as the structural positions subjects occupied in wider society that influence these material circumstances. For our subjects, these positions included their class status, racial identities, and their membership in a stigmatized group--former felons and prisoners. We categorized narrative change as attributable to structural incompatibility only when the individual recognized the structural constraints and directly attributed to them their failure to realize the script. This process was far less common than we expected it would be when we began this project. The majority of our subjects appeared to us to experience structural constraints (e.g. high local unemployment rates, low wages and high turnover in the secondary labor market, felony stigma, inadequate public transportation systems, lack of access to mental health treatment), but only three directly attributed their problems realizing the scripts in their narratives to these constraints, focusing instead on what they understood to be personal failures such as addiction, inability to secure support from family members, or failure to effectively take advantage of social services.9
An example of narrative change through structural incompatibly comes from Jada, a 31-year old African American mother of two. Prior to her incarceration she had worked for 11 years as a home-health aide while raising her children on her own. She served a two-year term in prison for smuggling drugs to her incarcerated boyfriend. When caught, she had already been on probation for drug dealing and firearm possession, although it was her boyfriend who had been responsible, “I didn’t want my baby daddy to go back to jail. He was on parole. So, I went to prison for that.” When Jada reflected on how she had landed in prison, she attributed it to the bad choices she had made in the past to make money and preserve the lifestyle she had enjoyed with her boyfriend, a case of economic motivation. Thus, within her narrative, avoiding prison in the future simply required her to make better choices. “Can't nobody make me do nothing I don't wanna do… I'm not weak minded. If I'm gonna do it, I'm gonna do it 'cause I wanna do it.” Framed in this way, her criminal involvement reflected nothing of her core identity, and reform required no significant personal change. Yet, over the months following her release, this narrative began to fray as she increasingly recognized how constrained her choices had become.
Prior to prison she had found employment fairly easily, but because home health positions were closed to felons, she was forced to pursue other options. Over the months following her release, Jada’s search for a job became increasingly desperate. Seven months after release and frustrated with the interviewer’s questions, she snapped, “I mean it’s not nothing to talk about. I can’t find no job because I’m a felon, it’s the same old stuff. If I do find a job it ain’t going to be worth nothing. It’s just ridiculous how these people just label people with felons that you can’t get no job. It’s stupid.” Later in the interview the normally sharp-edged Jada broke down in tears explaining “I’m just a little frustrated. I want a job so bad, you know? … a steady job. And it’s hard. I didn’t know it was going to be this hard. I’m used to working.”
Jada’s narrative began to change as she began to attribute her criminal actions to structural circumstances. Eleven months after release she expressed disgust at sentencing schemes that punish drug offenders more harshly than violent offenders, and explained that ex-offenders turn to drug selling out of desperation:
And then it’s hard for people to get a job and they expect people to come out here and not do the same thing. If that’s what you want somebody to come out here and get a job why would [Governor] Granholm make a law like the felons can’t get a job, that’s stupid. You all, I feel like y’all setting people up to go for failure, to go back. Can’t get no job, what’s the point?
Indeed, it appears that Jada herself has gone this route. During the interview a young man knocked on the door and put a wad of cash in Jada’s hand, leaving with no explanation. Jada mistakenly handed the money to the interviewer before taking it back with a laugh. While Jada maintains some elements of her narrative throughout the study period—for instance criminal behaviors never become part of her core identity—her initial narrative about crime as a bad choice, made independent of circumstances, recedes and is partially trumped by an alternative narrative as she experiences the substantial economic challenges of reentry.
Another process leading to narrative change is one we term Institutional Incompatibility, in which the institutional cultures, identities, or roles in which the individual is embedded conflict with some aspect of the narrative, which leads to narrative change.10 Institutions have their own cultural logics that may or may not be compatible with a particular subject’s narrative. One way of resolving these incompatibilities is to change one’s narrative to conform more closely to the cultural logic of the institution in which one is embedded. For our subjects, the three most common institutional settings were workplaces, treatment programs, and social service non-profits.11 An example of narrative change through institutional incompatibly comes from Lenora.
Lenora is an African-American woman who returned from prison to Detroit at the age of 51. She has two children, some college education, and a work history in the low-wage service sector. She has been to prison eight times, mostly for large-scale shoplifting or “retail fraud,” though she also admits to being involved in drug dealing. Lenora’s initial narrative attributed her criminal activities to her long-time addiction to alcohol, cocaine and other drugs, which started at age 14. This initial narrative is relatively simple: she is a “different person” when high on drugs, and someone who makes “bad choices.” Although she experienced trauma, including the death of her father during adolescence and an alcoholic mother, her initial narrative does not directly link these experiences to her drug and alcohol abuse or her criminal behavior. Indeed, she specifically differentiates herself from other drug and alcohol users, citing her level of education, her view that she was a good mother despite her addiction, and a strong desire to work and reform. Her impetus for change stems largely from a desire to improve her relationship with her sons and to contribute to society through working, key aspects of the new identity she hopes to construct. Moving forward, to remain sober she will avoid friends and family who are users as well as avoid romantic relationships in order to “focus on myself.” She plans to get a job, go back to college, and use her financial aid money to buy one or more houses, renovate them, and sell them. Despite participating in an intensive residential drug treatment program in prison, Lenora has only selectively adopted the language and logic of that institutional context. She does not view herself as fundamentally an addict, and does not plan to continue drug and alcohol treatment after release. She does, however, see the need to avoid people and places associated with substance use and to nurture a pro-social identity.
Lenora’s narrative began to change following a relapse and ninety-day placement in a substance abuse treatment facility. She explained that the relapse had been caused by disappointment over her inability to find a full-time job after her 90-day transitional job ended with no full-time job and her part-time employment had become infrequent at best. Yet her narrative change was also tied to her role as a patient in the treatment facility, as she explicitly voiced an alternative explanation for the relapse, one provided by her therapist. “What I’m going to do is I don’t want to overwhelm myself again. That’s what my therapist said I did. …I overwhelmed myself and I didn’t even know it, you know…So she knew that.” Lenora voices both the institutionally-prescribed explanation for relapse, while also maintaining her own explanation. Her therapist also helps her to define a new addiction-origin story, which locates her drug abuse in trauma:
“So she—they try to find your issues and mine was when my father passed. …‘Cuz see I had a beautiful childhood. Okay. My father passed when I was 14. …And that’s what I learned…when my dad passed, I got with a guy and I placed him as the void of my father. …Because he had money, he was a big dope dealer. …So my therapist was telling me I used him as the void to—as a father figure, you know?
Scholars of addiction recovery programs have emphasized the role of drug treatment workers and the inpatient treatment setting more generally in the construction of new narratives (e.g., McIntosh and McKeganey 2000a, Anderson 2015). Similarly, Lenora comes to a new understanding of the origin of her addiction, guided by an institutional logic that places great emphasis on the language and logic of addiction recovery programs. Six months later, after months spent residing in a substance abuse treatment facility, her explanation for the cause of her offending has also changed. She identifies cognitive patterns – divorced from structural circumstances related to challenges finding employment – as her pathway to relapse and re-offending, “If I drink and drink too much I get depressed because I ain’t got no money, and I will go into Macy’s, I will, because it’s…stored in my thinking.” Drinking alcohol remains the pathway through which she turns to stealing, but Lenora has moved from an understanding of herself as a “different person” when she steals, to an explanation of a cognitive process that guides this behavior, language that comes directly from treatment. Although she never entirely abandons her initial script for change, continuing to pursue work as a strategy throughout, her understanding of both the root of the problem and the pathway through which re-offense may occur are increasingly shaped by her institutional context.
Institutional embeddedness not only can lead to change but also be a source of narrative stability. Through a process we term Institutional Reinforcement, institutional roles, identities, or cultural schemas in the institutional environment may align with and reinforce the initial narrative. Institutional reinforcement is the narrative stability counterpart to change due to institutional incompatibility. For our subjects coming out of prison who were able to attain professional or middle-class status, the institutional roles or identities they took on served as a signal consistent with their narratives of redemption and social renewal. Their narrative identities were reinforced by these positions, providing a “stamp of approval” and legitimacy to replace a previously spoiled identity and a signal of the personal transformation their narratives implied. An example of institutional reinforcement is provided by Leon.
Leon is a 37-year old married African-American man from a middle class family in Detroit. He started petty theft and shoplifting in elementary school. He dropped out of college after one year and served in the Army. He served ten years in prison for a series of armed robberies and related firearms crimes, and was eager to return home and mend ties with his wife and 10-year old son. Leon attributes his descent into serious and frequent criminal activity to a need for additional income to support his family. He was working a low-wage telemarketing job and faced debts he could not pay. Once he started robbing, he explained, he found it easy to do again. Reflecting back, Leon sees these actions as narcissistic and selfish, and prison as a terrible waste of time. He vowed never to do anything that could send him back. Yet as he transitions out of prison, he continues to worry about the barriers he will face, particularly the state of the economy and the felony stigma that will make finding a job even more difficult. “You can only self-help yourself so much. They set it up where…they don’t really set you up to fail, because you fail yourself. …But they make it hard transitioning back to society.” Before leaving prison, Leon described a plan that involved focusing on finding work so he could support himself and his son. He chose to live with his father and stepmother, whom he believed would be supportive of this focus on employment. In short, Leon’s initial narrative was that of economic motivation, as he attributed his criminal past to poor decisions he had made when struggling to support his family. Making better decisions about how to earn a living was his script for moving forward, though he worried about the stigma and barrier his felony and prison term would pose.
Leon did indeed parole to live with his father and stepmother in a middle class neighborhood near Detroit’s waterfront. Once his time on an electronic monitoring “ankle bracelet” was complete, Leon set out on his own. After a few months of bouncing between different halfway houses and short-term stays with friends and family, Leon landed at a long-term residence with his sister and her girlfriend. His uncle helped him to get a job at a social service agency for the homeless, where many of the clients were also former prisoners. Although initially the job paid just above minimum wage, carried no benefits, required him to work at night, and required a 2-hour commute each way on multiple buses, the job nonetheless reinforced his desired narrative identity as a responsible middle class person. For instance, his wages, meager as they were, allowed him to pay $300 a month in informal child support. Although these child support payments were about a third of his monthly take-home pay and made it impossible for him to pay down his parole fees and restitution, he viewed it as a symbol of his responsibility rather than as an unfair burden:
That's one-third of my monthly pay, basically. I don't complain about that too much. Because you do have to pay the bills of where [my son] lives, you still have to take more responsibility in yourself and do something… She's living on his Social Security check, her Bridge card, and my child support. That's it, nothing else.
Leon’s job also reinforced his new narrative identity by drawing a symbolic distinction between himself, a successful former prisoner, and the clients he served, mostly former prisoners who were struggling with addiction and homelessness. Leon described how he dealt with an irate client who felt he was not getting the services he needed from Leon’s organization by recounting his own story of overcoming obstacles to be successful after prison:
I had told him my story just to help him out because some of the people that work there have the same problems that a lot of the homeless community has as well with substance abuse and stuff like that, and people try to share. Why you don’t let me help you get you into treatment?
Leon also started a romantic relationship with a coworker and the couple moved in together after Leon got a promotion that significantly increased his pay and provided benefits. The new position further reinforced his narrative identity as a professional middle-class person, and reinforced his pride in what he had accomplished.
It was a tough year, walking a lot, you know struggling to you know stay up with my job, try to move ahead and get a better position and stuff like that, but you know that’s how life is. You’ve got to work hard to get something and everything’s not easy… I wouldn't be the same person today if I didn't have that job, actually. 'Cause I love the work. I don't mind working at all. It defines me a little bit as far as just being useful to society in some type of way.
A third factor that affects narrative change and stability is the informal social networks in which an individual is embedded (networks not linked to the individual through a formal institution). Networks prompted narrative change among our subjects through a process we term Social Network Incompatibility, in which members of one’s informal social network reject some or all of the initial narrative. For our subjects, important informal social relations were typically with family members or romantic partners. The frequent, intense, and repeated interactions characteristic of such relationships make the views and actions of these individuals particularly salient. Such relationships also often come with specific social roles (e.g. parent, fiancé, or boyfriend) that may alternately reinforce or challenge one’s narrative identity, conception of the source of past problems, or scripts for future behavior (see also Charmaz (1983), Biernacki (1986), and Williams (1984) on the disruption of social roles and the construction of new narratives and identities and McIntosh and McKeganey (2000b) on the importance of reconstructing social networks as recovery “strategies” for those with addictions).
An example of narrative change through social network incompatibility comes from Geoffrey, a 45-year-old white male who began using cocaine when he was a teenager but nonetheless also had a fairly successful career as a plumber and home remodeler. When we met him, he was completing his fourth spell in prison, this time for cashing a stolen check. Geoffrey’s initial reentry narrative was one of economic motivation. Although he acknowledged his cocaine addiction, which he attributed to a combination of negative peers and a difficult childhood, he initially viewed his criminal past as stemming primarily from bad business transactions. For example, he explained that he stole the check that sent him to prison most recently because he was owed the money from a job, but the client had refused to pay. Geoffrey left prison eager to begin plumbing work again and confident that his drug-using days were behind him. Moving forward, he planned to rely on his self-proclaimed high skill level and network of colleagues who could hire him or help him find work. He believed that, by focusing on work, avoiding people or situations that might get him into trouble, and being a good provider for his girlfriend and his five children (none of whom were living with him), he could change his life trajectory. At release, he went to live with his girlfriend in a white working class Detroit suburb. Despite his high expectations, he struggled to find stable employment and quickly turned to working informally for friends and acquaintances at relatively low wages. He attributed these struggles to a general lack of work for plumbers and contractors due to the state of the economy.
Geoffrey never managed to connect with his children, and his relationship with his girlfriend soured after he started making romantic advances toward her daughter, who also lived with them. He moved out and experienced a series of temporary living arrangements that were mostly in-kind compensation for plumbing and remodeling work. A hand injury sustained while working and a month-long hospitalization due to an infection and hepatitis C put him in an even more precarious situation, and he soon had to disconnect his phone and sell his tools to afford food. With no relationships to cushion him, either materially or psychologically, he feared he would become homeless and even contemplated suicide. And absent anyone to “provide for,” and grim prospects for effectively fulfilling that role anyway, Geoffrey’s narrative identity as worker and provider came under challenge. “I’ve never had to live this way in my life. I’ve always made good money. I been taking care of myself since I was 14 years old.”
To be sure, Geoffrey’s economic problems undermined his narrative identity as a worker and provider, another example of structural incompatibility. But his narrative change was also precipitated by a change in his social network, from a romantic partner who reinforced his identity as a provider and needed his support to friends who were involved in crime and drugs, and as a result he experienced social network incompatibility. As his situation worsened, Geoffrey moved into the basement of an old friend with whom he used to work. He soon discovered that the friend was involved in drug selling and insurance fraud, and eventually Geoffrey helped this friend with one of his fraud schemes (the two were never caught) and resumed substance abuse. During the period after the major shift in his social network, Geoffrey’s narrative shifted from that of economic motivation to one that attributed his problems to the criminal justice system. He failed to report to his parole officer for a month, but blamed racist black parole officers for sending him to jail for a week and mandatory drug treatment in response. Rather than endure 120 days of treatment, Geoffrey absconded, believing that the Department of Corrections had neither the resources nor the interest to pursue him and send him back to prison. Lacking an identity as a provider to family, Geoffrey eventually concluded that he needed to “work the system” and apply for disability benefits, claiming that he cannot work due to his injury and other health problems. Geoffrey’s experiences illustrate how narratives can change through a social network shift from those who reinforce to those who challenge the initial narrative.
We also observed informal social networks contributing to narrative stability among our subjects through a process we term Symbolic Social Support, in which social relationships, such as those with family and friends, reinforce initial narratives by providing social roles consistent with those narratives, feelings of legitimacy (though without an institutional imprint), and cultural reinforcement (just as a religious convert may receive social support from other members of the new religious group). Symbolic social support is the stability counterpart to social network incompatibility. For our subjects, family relationships like mother or husband provided symbolic social support. Similar to Sampson and Laub’s (1995) argument about the importance of marriage for desistance from crime, meeting role expectations helped subjects to persevere with their narratives’ scripts for moving forward. Further, when family members validated their narratives, these identities were legitimized and reinforced. An example of narrative stability from symbolic social support comes from Paul, a 48-year old white male imprisoned for armed robbery.
Paul grew up in a middle class family, served in the military, and graduated from college. He married and had a 20-year career in counseling social work. Paul quit his job to care for his mother for two years when she was terminally ill with lung cancer. The stress and isolation of providing round-the-clock care followed by the loss of his mother led him to be treated for anxiety and depression with medication. Paul says he received insufficient follow-up care, and began drinking heavily after coming off of those medications. He first got two drunken driving convictions, and then got into a motorcycle accident, which lead to a prescription painkiller addiction. He robbed a video store and a pharmacy to support his habit and ended up in prison. Paul exited prison voicing an initial narrative similar in many ways to Lenora’s initial narrative, in which he acknowledged the role of substance abuse in his past problems but did not construct an addict identity. He attributed his incarceration to his addiction to alcohol and painkillers. With the clarity of thought that came in prison after the fog of substance abuse had lifted, he came to view this addiction as a temporary reaction to his mother’s cancer and eventual death – a fall from grace – rather than as a fundamental part of his identity or personality. For example, although he attended AA meetings in order to meet parole requirements and get his driver’s license reinstated, he did not feel that he belonged there, and purposely made no connections. He also felt that he could safely drink again once he had completed parole and was no longer subject to regular substance abuse tests. His script for moving forward was based on a desire to return to his former middle class social and economic status. Even before his release, his wife helped him to do the paperwork to re-enroll in a master’s program in criminal justice at a local university. He resumed class days after his release, and after one semester landed a position as a graduate teaching assistant and eventually as a full-time lecturer.
This upward trajectory and his narrative of return to middle class status did not come without challenge, however. Paul’s new professional identity was repeatedly contested. His own brother, a former police officer working in private security with whom Paul had a falling out even before Paul went to prison, viewed him as a criminal and an addict and would have nothing to do with him. His parole officer denied his requests to travel out of state to attend professional conferences and required him to work a minimum wage job during the summer break in order to meet parole requirements. Most importantly, administrators at the university initially denied Paul the lecturer position after they learned of his criminal record. He was able to overcome these challenges and reinforce his narrative through the social support of his wife, middle class friends, and professional colleagues. His wife’s income provided Paul with a middle class lifestyle even while he was still a graduate student, but more importantly, his marriage was an important indicator to Paul of his social position. In interviews, he talked frequently about the resilience of his marriage and the strength of his marital bond. Friends and former professional colleagues from before his incarceration welcomed him back into their social circle after his release, providing critical legitimacy to his narrative identity. When he was denied the lecturer position, faculty colleagues and students rallied to his defense and convinced the administration to reverse its decision not to hire him, an important social and institutional reinforcement of his new identity. Paul’s case illustrates the role of social ties, roles, and institutional legitimacy in narrative stability, as these factors allowed him to weather challenges to his initial narrative and the narrative identity he developed for himself. Although Paul experienced both institutional and social network challenges to his new narrative identity, because strong symbolic social support and institutional reinforcement came at key moments (e.g. the challenge to his appointment as a lecturer), his narrative weathered these challenges and indeed was reinforced by his ability to overcome them.
Finally, we also observed the content of the narratives themselves contributing to narrative change and stability. In a process we term Narrative Ambiguity, the content of the initial narrative was thin (contained few detail as to how or why past events were linked together causally), incoherent (contained logical inconsistencies), or vague in terms of a prescribed script for moving forward. While we did not observe narrative ambiguity itself leading to narrative change, it made some subjects’ narratives more susceptible to the above processes. In other words, ambiguity works in concert with other processes to facilitate narrative change (however, we emphasize that not all cases of narrative change involved narrative ambiguity; subjects with very coherent narratives also experienced narrative change). An example of narrative change due in part to narrative ambiguity comes from 48-year-old Jane, who has a lengthy history of both drug addiction and criminal behavior. Her most recent period of imprisonment was her fourth. Before going to prison, Jane held a series of unconventional and criminal jobs, from topless dancing and running an escort service to drug selling and prostitution.
Yet leaving prison, this time as a recently married woman and a new step-mother, Jane explained that this time was going to be different “because I have someone to go back to.” Her initial narrative located her past problems in the sexual abuse she experienced from an uncle, which led to a drug addiction and the criminal activity necessary to support it, a conventional and institutionally prescribed addiction narrative that located the sources of her past problems in traumatic experiences and her use of drugs and alcohol to deal with them. Her impetus for change was her new family role, and her script involved focusing on her addiction recovery by avoiding stress. Yet, in Jane’s case the script for moving forward was thinner than that of many of our other subjects with similar addiction recovery narratives, providing little in the way of a clear plan or concrete actions to take upon encountering triggers or stressors. In her in-prison interview, she explained that each time she had been released from prison in the past she had tried a new approach to staying clean and sober. Her plan this time was to avoid returning to work and simply stay at home, “I want to get comfortable just being in my own skin at home. And that was a big thing for me, just being comfortable, just being, doing nothing and not feeling like this overwhelming desire that I have to be something.” She would “make home my sanctuary” and only later pursue a job or further education. Yet aside from work avoidance, she had no clear plans as to how to avoid relapse, believing that returning to family would be sufficient.
Jane soon found that her home environment was not a sanctuary, and having little to do only made her addiction harder to cope with. “The first day, that wasn't so bad. It was like, not with those feelings, but it was probably like maybe a week or a few days after when it started getting like a little more…I started to get edgy…I don't have no money, I don't have a car…” She began to feel trapped. Further, the family she returned to presented problems of its own. Her husband married her with full knowledge of her drug addiction, had his own substance abuse problems and maintained a side-business selling drugs. He was also controlling and at times verbally abusive. Thus, rather than motivating her sobriety, this relationship presented both opportunity and justification for relapse. Social network incompatibility was a key part of her narrative changes, as her husband was unsupportive of her sobriety goals, and both her husband and step-daughter created substantial stress, which triggered her desire to use. Ironically, she explained that drugs had provided an escape from the stressors of home, “when I use it’s like an escape and then I don’t care, I don’t care that the kitchen’s a mess, I don’t care that [my stepdaughter] stole my stuff, I don’t care, I don’t care…” With both her impetus for change (her role within her family) and her script for moving forward (home as sanctuary) challenged by social network incompatibility, and with an ambiguous narrative about what she should do in response, these components of her reentry narrative faded over time. Her drug use became more serious, and during the remainder of our study Jane never experienced a period of sobriety long-enough for her to develop a new reentry narrative.
The specific content of initial narratives also produced narrative stability among our subjects through processes we refer to as Narrative Resilience. In such cases, the content of the narrative decreases the chances it will be abandoned either through (a) the anticipation and incorporation of experiences potentially challenging to the narrative, or (b) shaping interpretation of experiences so that they are understood as conforming to the narrative. Incorporation of potentially challenging experiences is more likely when the narrative content has a strong developmental component in which change is presumed to unfold over time, possibly unevenly. For example, this process is visible in many (but not all) of our subjects’ addiction narratives based on self-help and 12-step programs, which view the individual as a lifelong addict who must learn to manage and control the addiction, rather than addiction as a phase that is entered and exited definitively (as Paul, above, believed). Relapse into drug or alcohol use is part of the narrative rather than inconsistent with it, as most addicts are expected to relapse at some point. An example of this form of narrative resilience comes from Christopher, a 38-year old white man with a long history of crime and incarceration linked to the drug and alcohol addictions he developed as a teenager. When we met him, he was finishing 16 months in prison for attempted kidnapping.12
Christopher’s initial narrative attributes his drug and alcohol addiction to trauma he experienced as a child, namely sexual abuse at the hands of a neighbor. He began drinking and using drugs in his early teenage years, facilitated by the behaviors of his peer group. Christopher views himself as an alcoholic and addict, and easily parrots many of the ideas central to addiction recovery programs. He takes full responsibility for his choices and actions and emphasizes the importance of focusing on his sobriety. Christopher’s impetus for change is the support he expects to receive, both from family and from various treatment and social service programs. He paroled to an addiction treatment program. His script for moving forward, typical of self-help and 12-step programs, is to attend AA meetings faithfully multiple times per week, avoid people and places that might trigger a relapse, and focus on finding and maintaining employment. Yet before he is released, Christopher is quick to acknowledge the possibility of relapse and re-incarceration given his troubled past, saying he will be “ecstatic” if he could remain free in the community for just one year. His reentry experience was a rocky one, as he faced homelessness, a transitional job that did not become permanent, a series of transitions in and out of sobriety, a month in jail for open container violations and failure to report to his parole officer, and a crack binge in Detroit during which he claims to have smoked over $3700 worth of crack in a week. In contrast, during this time he also married, became a father and began running his own business with his wife. Nonetheless, three years after his release, Christopher was returned to prison for robbing a bank while drunk, his own bank where he had more money in his account than he had demanded from the teller.
Incarcerated again, Christopher still makes sense of his returns to drinking according to the logic of 12-programs. He recounts how the binge that sent him back to prison began: “And it’s the weirdest shit though… How did I just walk into the store one day and like, “Let me get a bottle of vodka?” Where does that shit come from? … There was no depression there. I had a job, everything was going good.” This quote subtlety conforms to the institutionalized addiction narrative, as Christopher explains that there was no external trigger for his return to drinking; something internal and reflective of his “true self,” which he still struggles to understand, took over and prompted him to buy the vodka. Sitting in prison awaiting his next parole, Christopher continues to voice the textbook addiction narrative, rehashing his relapses and resulting criminal behavior as part of a process of maturation and personal development that will eventually lead him to long-term sobriety. He also restated the same script for moving forward, emphasizing the importance of AA meetings and keeping busy with employment.
Narrative resilience also occurs when the narrative shapes perception so strongly that experiences or information potentially challenging to the narrative are not interpreted as such. This is consistent with research on cognition, as there are well known “cognitive biases” toward interpretation of information consistent with existing cognitive schemas, and the discounting of information inconsistent with existing schemas (DiMaggio 1997). An example of this form of narrative resilience comes from Morgan.
Morgan is a 33-year old white man from a working class Detroit suburb who has two children with two different mothers. He has considerable work experience as a car salesman and auto wholesaler and made a solidly middle class income in this occupation prior to his incarceration. Morgan was first incarcerated for fraud and larceny after he stole to repay gambling debts, events that he attributes to a gambling addiction he developed in his teens. While he relies on a variant of the addiction narrative to describe his criminal past, he has developed a different narrative to explain his second incarceration and to make sense of his current reentry. Morgan was returned to prison on a parole violation after his ex-wife alleged that Morgan had perpetrated domestic violence and stole from her. He insists that these allegations are untrue, and in fact it was his ex-wife who stole from him. He explained that, because he was on parole, police and parole agents believed her version of the story rather than his. Because Morgan’s narrative attributes his circumstances to external actors who are part of the criminal justice system, it requires no change on his part, either behaviorally or morally. Although he adopts a victim identity in this instance, he nonetheless approaches this identity with a great deal of agency. He believes that if he does a better job of advocating for himself and takes charge of his involvement in the legal system, he can prevent the types of problems he has had in the past. He is representing himself in his divorce, for example, and keeping very detailed records and paperwork related to both his parole and his divorce and child custody cases. In the long term, Morgan also anticipates campaigning for system reform after he has finished his sentence. In short, Morgan’s initial narrative was based on an understanding of himself as a victim of unfair practices in the criminal justice system. A key aspect of this narrative was that little change was required of him, other than being more careful about his interactions with criminal justice actors. Soon after his release he was threatened with a parole violation. In response, he absconded to another state and was eventually returned to prison in Michigan. Perhaps surprisingly, he again interpreted these problems as caused by unfair treatment. He recounted the experience and his interpretation of it almost a year later in an interview in prison.13
About six weeks after his release, Morgan got into an argument with a former girlfriend, and the police were summoned, but Morgan was not arrested. The next day Morgan reported this “police contact” to his parole officer by telephone, as all parolees are required to do, and, apparently unsatisfied with his explanation of events, his parole officer ordered him to report to the parole office in person immediately to explain the incident. When Morgan questioned why he needed to report in person, the parole officer threatened him with a parole violation. Expecting to be returned to prison, he failed to report and moved out of his parents’ house to make it harder for parole officers to find him. Morgan soon started a new relationship with a woman who helped him to procure a false identity and supported him financially, buying him a car in her name and signing a lease for him. Tiring of this relationship, Morgan left the state with the car and eventually began a new life as a car salesman in Tennessee. Morgan believes that the ex-girlfriend reported the car stolen and told authorities where to find him, and he was arrested and returned to Michigan to serve the remainder of his prison sentence. (The basic sequence of events – failure to report, absconding, arrest in Tennessee, and return to prison in Michigan on a technical violation– is also reflected in MDOC records).
Morgan understood these events through the lens of the problem definition in his wronged by the system narrative. Although the experiences of other subjects and conversations with MDOC personnel suggest that he would not have been violated had he reported in person as requested (given that Morgan was currently employed and that the police contact did not result in an arrest), Morgan felt unfairly treated by his parole officer and believed absconding was his only option. He also believed that as a white parolee, he was being treated unfairly by his black parole officer because of his race. Morgan felt he was falsely accused of stealing the car he took with him to Tennessee and was improperly put in a high security prison for parole violators who have also committed new crimes, saying that the girlfriend had given him the car and that he had been making the payments. As evidence, he cited the fact that neither Tennessee nor Michigan authorities chose to prosecute him for automobile theft. In the end, he was only cited by parole authorities for absconding and was moved to a low security prison as a technical violator. None of these events led him to question his wronged by the system narrative, and he continued to believe that his return to prison was the result of unfair actions by law enforcement. In sum, although many interpretations of these events seem possible, the power of Morgan’s narrative of unfair treatment by the criminal justice to structure his interpretation of events helped to preserve that narrative through potentially challenging experiences.
The concept of narrative has become an increasingly popular and productive sociological tool for understanding how individuals make sense of the social forces in their lives, how individuals construct social identities from cultural building blocks, and how culture can shape social action and individual behavior. Yet the literature on narratives has yet to examine how narratives change or remain stable over time in the face of structural barriers or constraints, an issue critical to each of these possible applications of narratives and to our understanding of how marginalized populations understand their own structural situations and navigate the interplay between structure and agency. In this paper we have drawn upon a longitudinal analysis of the reentry narratives of released prisoners to examine narrative stability and change over time as individuals encounter structural circumstances, institutional contexts, and social relationships that may challenge or reinforce their initial narratives and narrative identities. Our inductive analysis of narrative change and stability conceptualized a reentry narrative as containing three core elements: (1) a problem interpretation, (2) an impetus for change, and (3) a script for moving forward.
We analyzed the processes through which each subject’s initial narrative did or did not change over time. This analysis allowed us to develop a more general theoretical framework for understanding narrative change and stability. They suggest that four factors can account for narrative change and stability: (1) the structural circumstances experienced by the individual and the compatibility of those circumstances with the expectations of the narrative; (2) the institutional contexts in which individuals are embedded, particularly the degree to which institutional logics, practices, and cultures conflict or accord with the expectations of the narrative; (3) the informal social network in which the individual is embedded, particularly the degree to which that social network provides opportunities for role success in accordance with the narrative and the degree to which the views and actions of individuals in one’s network reinforce or challenge the narrative, and (4) the content of the narrative itself, particularly its capacity to anticipate or incorporate challenges and its specificity or ambiguity. Although for conceptual purposes it is useful to distinguish these factors, they tend to operate in concert with one another rather than independently, as is readily apparent in the examples presented above.
This appears to be a fairly comprehensive set of factors, so it is reasonable to wonder what did not play a role in narrative change and stability for our subjects. While we observed family members and romantic partners as key social network actors, we did not observe weaker ties as playing any sort of role (unless they were institutional ties). For example, a strand in the literature on prisoner reentry discusses the role of neighborhoods, but we saw no role in narrative change or stability for either neighbors or neighborhood structural conditions (e.g. neighborhood poverty). While we observed structural positions as contributing to narrative change, not all such structural positions seemed to play important roles. Our subjects’ status as former prisoners was critical, as was their social class background, material resources, and racial identities, but we did not observe a role for, nor did our subjects discuss, other forms of identity such gender, age, sexual orientation, or urbanicity. While we observed institutions playing a strong role, it was only the institutions in which they were directly embedded and heavily involved (treatment programs, criminal justice agencies, workplaces) that mattered. Institutions in which they were less intensively involved did not seem to play a role (e.g. health care settings, educational institutions, religious organizations, political or community organizations).
As indicated in Table 1, we did not observe among our subjects any instances of structural positions promoting narrative stability. We believe this is a product of the disadvantaged structural positions of our subjects rather than a theoretical impossibility. Individuals with greater material resources or who occupy positions of authority, social status, or power might be expected to use their material and social resources to improve the chances of realizing the expectations encoded in their narratives and thereby having them reinforced. For example, we would not be surprised to see that corporate CEOs or Wall Street financiers were able to use their authority, status, or material resources to insulate themselves from those who might challenge their narratives. Moreover, one might also expect that authority, status, and power can by themselves lend institutional or social network legitimacy to one’s narrative, also serving to reinforce it or block challenges to it. Research on narrative stability and change in other populations would be needed to further explore these issues.
One noteworthy finding is that our subjects’ narratives were surprisingly stable, given what seemed to be significant challenges to their narratives from the vantage point of an outside observer. We found far more stability than we expected at the outset of the study. Some narratives resist challenge based on the relationship between the content of the narrative and the structural, institutional, and social circumstances of the individual. For example, we found that narratives may be resistant to change when they are reinforced by social ties and institutions, even if the individual faces structural circumstances that make realizing the narrative particularly difficult. When narratives are not reinforced, they are easily challenged by structural constraints and then abandoned or modified. One implication of this finding is that different people have different opportunities for narrative stability due to the institutional, network, and structural positions that they occupy. In other words, different people have different opportunities for agency to overcome structural constraint, and narrative change and stability may account for why and how that happens. Future research might test this hypothesis more explicitly by comparing individuals with similar narratives who occupy different social positions. Another implication of the importance of narrative content for stability and change is that the cultural stock of meta-narratives, public narratives, or narrative templates to which an individual has access may matter greatly for narrative stability and thereby may also affect the individual’s capacity to realize the scripts contained within their narratives. This may also be a function of the institutions in which they are involved. Future research might test these hypotheses more explicitly by comparing individuals with different narratives who occupy similar social positions. Moreover, one might also hypothesize that narratives that correspond more closely with the most commonly held meta-narratives might be less susceptible to narrative change because they align most closely with existing institutions, which both reflect and reinforce such narratives.
Our results also suggest that narrative change may not always involve a complete abandonment of the initial narrative. This is consistent with other literatures. For example, the literature on religious conversion shows that conversion is rarely total, as former practices and beliefs are often incorporated into the new religious views (e.g. Rambo 1993). Moreover, we found that different elements of our subjects’ narratives were differentially susceptible to change. We found considerably more change in how our subjects conceptualized their impetus for change and their scripts for moving forward than in their interpretations of their past problems. The implication is that one can change one element of a narrative without changing others, although this appears to depend on the coherence of the overall narrative. Less coherent narratives may be more susceptible to this sort of partial change than others. One implication of this finding is that narrative change might often be thought of as narrative revision rather than a wholesale abandonment of one narrative for another completely different narrative. Another implication is that stable narrative identities are likely to be tied more closely to some narrative elements than others. In the case of the reentry narratives studied here, problem interpretation was more stable than impetus for change or script for moving forward. A second implication is that future research on narratives, narrative identities, or narrative change and stability should identify, and differentiate between, narrative elements rather than assuming that narratives remain highly coherent over time.
The examples presented above suggest another set of noteworthy findings related to the dimensions of variation across narratives, both within and between individuals. First, we note the high degree to which our subjects’ narratives (both initial and subsequent) were constructed from common cultural templates and were informed by institutional cultures and practices, in this case those surrounding the prison and wider criminal justice system as well as drug and alcohol treatment programs. Second, consistent with prior research, we see variation across narratives in their coherence (the degree to which the interpretation of past problems links directly to an impetus for change and a script for moving forward) and specificity (the amount of detail in the story), their degree of agency vs. structural constraint, and in the moral status attributed to the subject or other key actors. Third, as Maruna (2001) might expect, we see that prisoners poised for release from prison vary in the degree of identity transformation and personal change in their narratives, and the degree to which a new self emerges to replace the “flawed self” who committed past crimes. For example, some addiction narratives tended to contain more detail about the ways that current problems were rooted in past experiences (problem interpretation) and in prescribed future action (scripts for moving forward) than others. Narratives that explain past problems in terms of economic motivation or an unfair criminal justice system cast the subject in more positive moral terms. Understanding the dimensions on which narratives vary will be critical to any future analysis of narrative change and stability or the causal effects of narratives.
We also see variation in narratives along dimensions not specifically addressed in prior research. One is variation in the degree to which subjects come to adopt the institutionally sanctioned logics, in this case the language of the prison and treatment programming discussed above. For example, Christopher draws heavily from the narrative template provided by substance abuse programs while Lenora incorporates only one aspect of it, namely the concept of “triggers” for use. A second new dimension is the degree to which a narrative is tightly linked to a particular narrative identity. While the plot of some narratives leads naturally to a particular identity, the plot of others leaves space for the creation of multiple identities. For example, addiction recovery and economic motivation narratives are more closely tied to particular identities (addict, patient, worker) than are identities based on mistreatment by an unfair system, which leave open a wider range of possible identities. A third new dimension of variation is the ambiguity of a script for moving forward. Some narratives are quite concrete in terms of a script, such as those that adopt wholesale institutionalized narratives about substance abuse treatment and recovery, while other scripts focus primarily on what the subject should not do, such as avoiding triggers for substance use.
A key argument for the conceptual and empirical importance of narratives is that they structure future action, but this claim has received little direct empirical attention, perhaps because it is claim that is very difficult methodologically to test. Although we too do not directly examine the causal effect of narratives on action, our analysis of narrative change and stability has implications for it. Our finding that some narratives resist challenge and remain stable suggests that there is potential for the independent effects of narratives on action. However, narratives that simply change in response to one’s structural, institutional, or social circumstances seem unlikely to impact behavior. Narratives may be resilient simply because their content anticipates and incorporates challenges from an individual’s experiences, interactions, or contexts, as was the case with the textbook addiction narratives.
However, our findings also suggest that the causal power of narratives to affect behavior, particularly in the longer term, is contingent on an “alignment” between the content of the narrative and one or more of the other three factors we have highlighted in our analysis: the structural position of the individual, the institutions in which the individual is embedded, and the local social network in which the individual is embedded. Narratives that “align” are reinforced, thereby also reinforcing identities and potentially influencing future behavior. Put differently, when narratives align with circumstances and opportunities, the individual may be better positioned to overcome obstacles they encounter rather than abandoning or changing their narrative. One key aspect of this alignment is the ambiguity of the narrative. Ambiguous narratives that are not supported may have little power to help individuals make sense of their life circumstances, but those that are supported may be reinforced and further elaborated.
With regard to the literature on prisoner reentry, our findings document some of the more common narratives that prisoners develop to make sense of their reentry expectations and experiences, contributing to our knowledge of how they understand these experiences. We submit that the richness of the narratives presented in the data above indicates that narratives may be an important conceptual tool for studying the cognitive change that is thought to be a critical part of desistance in some theories of crime. Our findings also suggest that criminal justice and related institutions, such as substance abuse treatment and other programming, play a key role in the development of reentry narratives through the provision of “narrative templates” or “meta narratives.” Yet the narrative templates provided by such institutions often do not prepare prisoners for the challenges they will face after release, leading in some circumstances to their abandonment and the construction of new narratives. This highlights the failure of criminal justice and related institutions to provide well-defined and accessible institutional pathways to realize the scripts for moving forward contained within their narrative templates.
Finally, our argument about the importance of narrative alignment with structural, institutional, and social circumstances contrasts markedly with the dominant view of narratives and desistance from crime – Maruna’s (2001) argument that it is primarily the content of the narrative that matters. He argues that ex-offenders whose narratives contain high levels of agency and interpret their past problems as a “necessary prelude” to their current selves tend to desist. Our findings suggest that these effects of narrative content are likely to be contingent on the alignment of the narrative content with individual circumstances.
In closing, we note four important limitations of this study. One is our relatively small sample size. Although necessary for successfully following subjects longitudinally and interviewing them frequently in a period of considerable uncertainty in their lives, our small sample size means that some types of narratives or some processes of narrative change or stability that are relatively rare may not have been present in our sample. Moreover, it is impossible to provide information on the prevalence or frequency of the processes that we identify in the larger population of released prisoners. Our small and unrepresentative sample also means that we cannot be certain that our results will generalize to the larger population of former prisoners or ex-offenders. It also means we cannot analyze which individual characteristics tend to be associated with different types of narratives. Future research might investigate the social patterning of narrative adoption and construction.
Second, our analysis of narrative change and stability among former prisoners may not generalize to other populations. Of course, hundreds of thousands of individuals leave prison every year in the United States (West et al. 2010), and incarceration in prison has become a typical life course event for some segments of the population, particularly low-skill minority men (Western 2006). Yet, former prisoners are unique in potentially important ways. The same characteristics that make them an interesting population for studying narrative change and stability also means their experiences with narratives may not generalize to other populations. They face extreme forms of stigma (Pager 2007) and extreme forms of marginality (Wacquant 2008, 2009), structural and cultural circumstances that few other populations encounter. Indeed, although theoretically we might expect structural advantages to produce narrative stability, we could not examine such processes in our data because none of our subjects possessed such advantages. Furthermore, the prison itself is a total institution (Goffman 1961), suggesting that it may have a particularly strong influence on narratives and narrative identities. Former prisoners also have high levels of involvement in drug and alcohol treatment programs, another robust source of narrative templates that may result in particularly strong institutional influences. Together these two institutions may limit variation in reentry narratives or provide narratives that are particularly susceptible or resistant to change.
Third, our sample is also specific to a particular place and time, Michigan during the onset of the Great Recession. Social and economic conditions as well as criminal justice policies and social welfare generosity vary considerably from state to state. Michigan is characterized by high unemployment, declining opportunities for employment in low skill positions, low investment in public transportation, and high rates of racial and economic residential segregation. This means that the structural challenges our subjects faced and the institutional and social contexts in which they were embedded may have been particularly difficult. On the other hand, during the study period Michigan had relatively generous welfare, housing, and food stamp benefits and few restrictions on access to social welfare benefits for those with a criminal record. Finally, Michigan also has few Latinos or Asians, so our sample includes only blacks and whites.
Finally, we note that any analysis of narratives is a highly subjective and interpretive exercise. An analyst who approaches the same data with a different interpretive frame may come to different conclusions about the nature of the reentry narratives in our data, the degree of change or stability of narratives and narrative identities, or the processes generating change and stability. We have attempted to limit this possibility by involving multiple analysts in this study and only reporting results on which our research team could reach consensus. A related limitation is that establishing causal relationships between processes and narrative change or stability is challenging in an inductive study of a small sample. We have attempted to mitigate this concern by leveraging the longitudinal nature of our data and focusing on experiences most salient to our subjects that might represent challenges to their narratives or narrative identities. Only further analyses of narrative change and stability by other researchers studying other samples of former prisoners or other populations entirely will reveal whether our findings are generalizable, whether the links we identify can be understood as causal, and whether the framework we have developed is conceptually useful.
This research was supported by the Office of the Vice President for Research, Rackham Graduate School, Department of Sociology, Joint PhD Program in Sociology and Public Policy, National Poverty Center, and the Population Studies Center of the University of Michigan, which is supported by the National Institute of Child Health and Human Development under grant R24 HD041028 and by the National Institute on Aging under grant T32 AG000221. We thank Elizabeth Johnston, Andrea Garber, Elena Kaltsas, Madie Lupei, Carmen Gutierrez, Adam Laretz, and Emma Tolman for excellent research assistance. We thank Sandra Smith, Michele Lamont, and Black Hawk Hancock for helpful comments on a previous draft.
Because statistical representativeness across multiple subject characteristics is impossible in a study with a small sample size, we instead pursued a “purposive” sampling strategy common in qualitative research (Kuzel 1992). Our goal in selecting subjects was to ensure racial and gender diversity, diversity of local geographic context, and diversity of services and supervision provided by MDOC. Accordingly, the sample was stratified by gender, race (white vs. black), reentry county (urban vs. suburban), and type of release (receiving services from Michigan Prisoner Reentry Initiative [MPRI], not receiving MPRI services, or being released without parole [i.e., “maxing” out]). The Michigan Prisoner Reentry Initiative is a statewide policy effort to reduce crime and incarceration by providing additional services to parolees and by implementing a regime of “graduated sanctions” for technical parole violations. During our research, MPRI was still being phased in, and not all parolees received services. During subject recruitment, MPRI parolees tended to be those classified as “high risk” or “medium risk” by MDOC. This research is not an evaluation of MPRI.
Within these categories, potential subjects available at the time of recruitment were selected at random. This sampling strategy ensures that theoretically important categories are present and therefore that conclusions drawn are not particular to the largest group of former prisoners in the population (minority males released to central cities). Three male subjects refused to participate in the study and one female subject was discontinued from the study after she was denied parole after the first in-prison interview. These subjects were replaced by additional randomly selected individuals with the same sampling characteristics, resulting in a response rate for in-prison pre-release interviews of 86% (24/28).
Individuals involved in criminal activity, the criminal justice system, and substance abuse are challenging to study. Although we began the study with pre-release interviews with 24 subjects, two subjects, one male and one female, left the study immediately following their prison release. These subjects were younger than our average subject, and both were subsequently convicted of new crimes. All remaining 22 subjects were interviewed once before release, and were interviewed repeatedly during the two years following their release, with follow-up interviews targeted at approximately 1, 2, 6, 9, 12, 18, 24, 30, and 36 months (see Table A1 for detailed interview timing), for a total of 154 interviews. In-prison interviews were conducted in private rooms (often those used by lawyers visiting their clients). MDOC regulations forbid recorders within prisons, so field notes were used to document in-prison interviews. Post-release interviews were primarily conducted in the subjects’ residences, but also occasionally in the researchers’ offices or in a public location. These interviews were recorded and transcribed.
Subjects were matched with interviewer on gender. The same interviewer conducted all interviews and interactions with each subject throughout the study. Subjects revealed more and more information as interviews progressed, including information that was intentionally withheld in earlier interviews. One potential barrier to trust was the recruitment of subjects in prison with the cooperation of MDOC. However, consent forms clearly stated that no individual information from the study would be shared with MDOC or any other law enforcement agency unless there was indication of imminent harm to subjects or others, and we secured a Certificate of Confidentiality prior to the initial interviews to protect study data from law enforcement and the courts.
The number of post-release interviews per subject ranged from two (two subjects who were returned to prison not long after release) to eleven, with a mean of 5.8 and median of six. For some subjects, we intentionally allowed interview timing to vary, depending on their circumstances. In some cases we conducted interviews more frequently when subjects’ lives were particularly unstable. In other cases subjects took up criminal behaviors – using drugs, committing new crimes, and absconding from parole – for a period of months. For these individuals, we were able to complete interviews at a later date that discussed the missed time period, including follow-up interviews in prison with three subjects. For all but one of our subjects who committed new crimes, we were able to interview the subject at some point afterwards. One subject was killed in a shooting during the second year of the study. Another subject achieved stability one year after release and, because his circumstances remained essentially unchanged over the months that followed, we concluded formal interviews with him. We did keep in touch with him informally throughout the study period. We also monitored his parole agent’s case notes for criminal infractions as well as residential moves or changes in work and schooling, of which there were none. This subject reached a “saturation point” (Bowen 2008) after three consecutive interviews revealed little new information about his circumstances, plans, or perspectives, owing to his stable family and employment situation.
We used a number of strategies to maintain contact with subjects in this difficult to track population. The most important was access to parole agents’ records, which allowed us to maintain contact with subjects whenever parole agents knew how to find them. For all but two subjects, we obtained consent to access their MDOC records, allowing us to view their parole agents’ case notes for updated contact information, substance use tests, parole violations, and arrests. This was particularly useful for tracking residential moves, periods of incarceration in jail, and periods of confinement to residential treatment. A second strategy was to elicit from subjects at the initial interview and regularly thereafter the names and contact information of three individuals who would be likely to know how to contact them. A third strategy was to provide incentive payments of $60 per interview. Finally, we believe, though cannot independently verify, that the rapport we were able to develop with subjects over repeated interviews contributed to our lower attrition rate than some earlier studies of former prisoners. In total, we were able to maintain regular contact with 19 of 22 subjects across the two-year period, although this involved less frequent interviews with four imprisoned subjects once they were incarcerated, and a truncated interview schedule with one subject who was killed. Three subjects attritted during the course of the study at 2, 12 and 20 months.
Half of the male sample is white and half black, as Latinos and Asians make up a very small proportion of the Michigan population and of individuals released from Michigan prisons. At the initial interview, men ranged in age from 22 to 71, with most subjects in their late-twenties to early-thirties. Crimes of which male subjects were convicted range from armed robbery to driving under the influence (multiple convictions can lead to imprisonment) to manslaughter. Five male subjects were being released from prison for the first time; all others had experienced multiple prison spells. The female sample is also half white and half black. Women ranged in age from 22 to 52 at the initial interview, with most subjects in their late-thirties or early forties. Women’s crimes ranged from felony firearm possession to retail fraud to drug selling. Three women were leaving prison on their first release; the other four had served previous prison terms. We assigned pseudonyms to all subjects and to any other individuals mentioned by name.
|Follow-up Interview Number|
NOTE: Excludes initial pre-release interview; Numbers in the table indicate the number of months since initial release from prison at which an interview occurred. Text in parentheses explains reasons for no further follow-up interviews during the study period.
R = Interview Refused
D=Dropped out of the study
X=Subject interviewed following release from prison or jail
1This way of thinking about the development and deployment of narratives is consistent with a toolkit model of culture in which culture provides a repertoire of building blocks from which individuals construct interpretations and strategies of action (Swidler 1986, 2002, Hannerz 1969).
2Institutions, in turn, are products and drivers of larger historical sociocultural movements such as neo-liberalism, thus linking individuals’ narratives to macro narratives (e.g. Lamont and Hall 2013).
3McAdams (2001), writing from a psychological perspective that focuses on life-story narratives in particular, adopts a slightly different conceptualization of the relationship between narrative and identity. He sees a life-story narrative as linking together multiple disparate identities by describing the procession from one to the next, i.e. by accounting for changes in identities over time.
4One might question Maruna’s strong claims about the causal effect of characteristics of narratives on desistance on methodological grounds. His interviews are retrospective, having selected desisters and non-desisters and then solicited from them their life narratives. Because narratives justify or explain one’s experiences, it is hard to imagine retrospective narratives from desisters that did not contain a strong sense of control and minimal emphasis on barriers and constraints. Similarly, one might also expect that those who returned to crime would emphasize the barriers they faced in desisting from crime. Although our study uses a prospective design that elicited narratives before our subjects faced the challenges of reentry, our goal in this paper is not to make causal claims about the effects of narratives but rather to examine how narratives change over time.
5See the methodological appendix for additional information on sampling, subject retention, and subject characteristics.
6An alternative approach to the analysis of narratives is quantitative narrative analysis, or QNA, in which the text of each narrative is coded into its base elements, stored in a relational database, and data are extracted from the database in the form of datasets amenable to quantitative analysis, such as sequence analysis or event history analysis (Franzosi 2010). Although there is much to recommend QNA when the focus of analysis is on the nature of events in the narrative and patterns of associations across narratives, Franzosi (2010) explains that QNA is not well-suited to the analysis of perceptions, understandings, or meaning making of the tellers of the narratives, as is our focus here.
7We also note that some subjects developed more than one narrative at the same time. Due to space constraints, we cannot describe all of the specific narratives articulated by each of our subjects, although our results contain many examples.
8This does not mean we believe that the narratives presented to us in the context of our interviews were simply fabricated for the purposes of release and rehashed for us. The depth and emotional intensity of our subjects’ narratives along with their stability after the incentive to fabricate was over (see below) convinced us otherwise. Moreover, almost all of the narratives we observed at release identified a root cause of criminality as something outside of the individual, such as needing money, abuse or neglect as a child, or peer influences, that sociologists would view as at least partially determined by one’s structural position.
9Attribution of problems to individual failures rather than structural constraints is of course a recurring them in the sociological literature on poverty and inequality (e.g. Young 2004, Sharone 2013).
10The term “institution” is used in many different ways (Small, Harding and Lamont 2010). Here we have in mind a micro conception, particular organizations that create, aggregate, or distribute material, social, or cultural resources, such as a particular firm, treatment program, or non-profit organization.
11One might imagine that other institutions like educational institutions, health care institutions, or religious institutions might be important for other populations.
12This kidnapping was essentially a botched carjacking, in which Christopher threatened a woman to get her to give him her car, but then saw that there were children in the back seat. Seeing the kids, he ran away, but his threats nonetheless qualified as an attempted kidnapping.
13As this was an in-prison interview, only handwritten field notes were allowed, and there are no direct quotes from this interview.
David J. Harding, Department of Sociology, University of California at Berkeley, 410 Barrows Hall, Berkeley, CA 94720.
Cheyney C. Dobson, Department of Sociology, University of Michigan, 500 S. State St, Ann Arbor MI 48109.
Jessica J. B. Wyse, Institute for Social Research, University of Michigan, 426 Thompson St, Ann Arbor MI 48106.
Jeffrey D. Morenoff, Department of Sociology, University of Michigan, 500 S. State St, Ann Arbor MI 48109.