How Do Religious People Become Atheists? Applying a Grounded Theory Approach to Propose a Model of Deconversion

The influence of organized religion is decreasing in the West, and trends show a growing number of people abandoning their religious beliefs, or deconverting. However, this phenomenon has received relatively little attention in the psychology of religion. The current study asks “How do religious people become atheists?” and aims to further the understanding of the process of religious deconversion by offering a proposed model of deconversion. The main findings within the literature are examined, and consideration is given to the concept of deconversion itself and to biases within the psychology of religion. Employing an inductive grounded theory approach based on Strauss and Corbin’s guidelines, we investigated the process of deconversion among a sample of atheist individuals who previously identified as religious. The data consists of 30 testimonies obtained from former clergypersons and six semi-structured interviews with atheist participants recruited through an advocacy group. The resulting model of deconversion is comprised of three core categories: reason and enquiry, criticism and discontent, and personal development. Despite being closely interlinked, these categories were clearly distinct and represent an intellectual impetus, moral and ethical judgments of religion, and overcoming personal issues, respectively. For all participants deconversion developed gradually within the close context of family and local community and the wider cultural context of society at large. Findings are discussed in relation to previous research and psychological theory.

But what phenomenon is behind the rise of non-religion, at an individual and psychological level? Before considering that, it is important to observe that the increasing number of atheists and other non-religious groups cannot be attributed to demographic patterns, as non-religious people tend to have fewer children than religious individuals (Zuckerman 2011). In addition to this, Galen (2014) also indicates that, when compared to conservative religious people, the non-religious tend to marry (or cohabit) later in life. Furthermore, and with the exception of atheists from highly secular societies (Zuckerman 2008), most of those who are atheists today in traditionally religious countries have experienced some form of religious upbringing (Cragun 2017;Zuckerman 2011). Taken together, it appears that the observed increase in atheism is best explained by more and more people abandoning their religious beliefs, or deconverting (Hunsberger and Altemeyer 2006).
In contrast to the considerable amount of psychological research on religiosity/spirituality in the last 30 years, little attention has been given to deconversion (Paloutzian et al. 2013). There are a number of reasons why investigating the phenomenon of deconversion is important. First, studying deconversion complements our existing knowledge on the psychology of religion. Second, as it relates directly to atheism, studying deconversion can help to legitimize a worldview that is often neglected Pérez, S and Vallières, F. 2019. How Do Religious People Become Atheists? Applying a Grounded Theory Approach to Propose a Model of Deconversion. Secularism and Nonreligion,8: 3, Trinity College Dublin, IE Corresponding author: Sergio Pérez (perezs@tcd.ie)

Sergio Pérez and Frédérique Vallières
The influence of organized religion is decreasing in the West, and trends show a growing number of people abandoning their religious beliefs, or deconverting. However, this phenomenon has received relatively little attention in the psychology of religion. The current study asks "How do religious people become atheists?" and aims to further the understanding of the process of religious deconversion by offering a proposed model of deconversion. The main findings within the literature are examined, and consideration is given to the concept of deconversion itself and to biases within the psychology of religion. Employing an inductive grounded theory approach based on Strauss and Corbin's guidelines, we investigated the process of deconversion among a sample of atheist individuals who previously identified as religious. The data consists of 30 testimonies obtained from former clergypersons and six semi-structured interviews with atheist participants recruited through an advocacy group. The resulting model of deconversion is comprised of three core categories: reason and enquiry, criticism and discontent, and personal development. Despite being closely interlinked, these categories were clearly distinct and represent an intellectual impetus, moral and ethical judgments of religion, and overcoming personal issues, respectively. For all participants deconversion developed gradually within the close context of family and local community and the wider cultural context of society at large. Findings are discussed in relation to previous research and psychological theory. (Beit-Hallahmi 2007;Brewster et al. 2014). Lastly, given its existentialist nature, deconversion offers insight into how people interpret their lives, how they make sense of the world, and how they create meaning and identity (Barbour 1994;Schnell and Keenan 2011;Smith, 2011). In this paper we argue that people who abandon religious belief share certain psychological aspects that play a fundamental role in driving the process of deconversion.
Next, we offer a series of definitions of the terms used in this article, followed by some considerations about the biases in the psychology of religion, and the main findings in deconversion research. Religion, religiosity, spirituality and atheism Religion has proven to be a hard concept to define, as shown by the plethora of definitions found in the literature (Zinnbauer and Pargament 2005). Depending on what aspect of religion is addressed, these definitions can be grouped into three main approaches: substantive, functional, and prototypical. Substantive definitions concentrate on the content or essence of religious phenomena, usually making reference to supernatural entities. Conversely, functional definitions describe what role religion might have for its followers, such as answering questions of life meaning (Schilbrack 2013). Lastly, prototypical approaches offer instead a series of "religion-making" features (e.g. God, sacred objects, rituals, and moral codes) that might be present in a religion to a greater or lesser extent (Cohn and Klausner 1962;Alston 1967).

Definition of terms
A proper definition, as de Muckadell explains (2014), must describe the essential features of religion, that is, the necessary and sufficient conditions that make a system of beliefs a religion. Considering this, substantive definitions do a better job than the alternatives, since by nature they describe religion's fundamental characteristics. Although religion is malleable and changes overtime, substantive definitions can still address this aspect by being sufficiently vague as to permit a range of phenomena. Functional definitions, on the other hand, tend to be too vague and excessively inclusive, while prototypical definitions fail to distinguish the centrality of each feature.
For these reasons we have taken a substantive approach, adopting Argyle and Beit-Hallahmi's definition of religion as "a system of beliefs in a divine or superhuman power, and practices of worship or other rituals directed towards such a power" (1975,1). This definition covers what Argyle and Beit-Hallahmi (2014) observe as the common denominators of religions worldwide: the presence of a supernatural being or force and the relationship that human beings have with such entity. The statement "a system of beliefs in a divine or superhuman power" is sufficiently vague (i.e. it does not restrict itself to the God of the Bible, for example) and yet characteristic of religion.
Religiousness or religiosity, as the quality or state of being religious, can be commonly understood as a personal belief in God/gods/a higher power, institutional practices (e.g. Sunday mass), and commitment to organized religion. Similarly, the construct of spirituality is also related to beliefs, but adopts a more subjective relationship with a higher power or transcendental reality, without the need for organizational involvement (Zinnbauer et al. 1997). By contrast, atheism is defined as the absence of belief in the existence of God or the belief that God does not exist (Martin 2007). Whilst secular, non-religious individuals show ranging degrees of disbelief, more than half identify as atheist (Galen 2009).

The concept of deconversion
Deconversion is a relatively new term within the psychology of religion (Streib et al. 2009). As a phenomenon it has been conceptualized under a number of different definitions, making it difficult to identify a clear description (Bromley 1988). Barbour offers a general definition of deconversion as "a loss or deprivation of religious faith" (1994,2), as suggested by the prefix "de", denoting freeing from, removal, or opposition. In other words, deconversion is a process that implies the diminution of religiosity over time resulting in the adoption of a non-religious identity (in the form of atheism or, at the very least, agnosticism).
However, deconversion has also been conceptualized in ways that are at odds with Barbour's definition. For example, Streib et al. interpreted deconversion as "migration in the religious field" (2009, 28), a definition that includes adopting a different religion, becoming privately religious, or abandoning religion for a secular worldview. The first case, however, represents an instance of religious switching, and the second case represents only a disaffiliation from the organized element of religion. Therefore, only the third case would qualify as deconversion (McAdams 2010).
For Cragun and Hammer (2011) the use of the term deconversion is unwarranted and conversion should be used instead. The authors argue that "everyone who adopts a religious identity, everyone who switches religions, and everyone who exits a religious identity is a convert (emphasis added)", because they experience a "change in religious identity" (2011,159). The problem with this conceptualization, however, is that it overlooks the direction and the result of this change. To say that a person who exits a religion has converted falls short as a description and does not do justice to the type of transformation. Those who become atheists have shredded the religious aspect of their lives and as a result they no longer hold a religious identity. In fact, the formation of an atheist identity resides in the rejection of theism and faith (Smith 2011). Therefore, it could be argued that although an individual who becomes an atheist experiences a change in religious identity, the distinctive outcome of this change is, more precisely, the abandonment of a religious identity and the adoption of a non-religious one.
The term deconversion is necessary because the narratives and elements involved in this process are qualitatively distinct from those present in conversion stories (Barbour, 1994). For example, Fazzino (2014) compares conversion and deconversion journeys, showing that, although they share basic structural aspects (ideological shifts, emotional responses, self-transformations), the nature of these aspects is different. Namely, conversion stories often involve feelings of security, assurance, and certainty, whereas in deconversion individuals tend to feel guilt, alienation, and eventually freedom. Moreover, a "deconversion perspective" emphasizes a reaction against the hegemony of a religious worldview that is often taken as the default ideology (Fazzino 2014).
Other frequently used terms applied to the phenomenon of religious deconversion are apostasy and disaffiliation. Caplovitz and Sherrow argue that apostasy "indicates not only loss of religious faith, but rejection of a particular ascriptive community as a basis for self-identification" (1977,31). Similarly, Beit-Hallahmi refers to apostasy as "disaffection, defection, alienation, disengagement, and disaffiliation from a religious group" (2007,302). These definitions are not incorrect from a descriptive point of view, but, as Cragun and Hammer (2011) argue, apostasy is a term used pejoratively by those who remain religious. Disaffiliation has been defined by Gooren as "(the process of) detaching one's involvement in an organized religious group" (2010,4). This definition focuses on detachment from organized religion, but it does not say anything about loss of faith, indeed including "an inactive member who still self-identifies as a believer" (2010,49). Therefore, apostasy and disaffiliation are not better suited than deconversion to describe the phenomena addressed in the present study.

Biases within the psychology of religion
As with most disciplines, the field of psychology of religion has not been free of certain biases taking the form of assumptions that have in turn influenced the research, results, and type of knowledge produced (Belzen 2010). Three of these biases are explained here. First, religion is assumed to be universal, and every person is expected to be religious in one way or another by default. The base of this assumption lies in equating religion and religiosity with "ultimate concerns" (Tillich 1957), a concept directly borrowed from theology. Ultimate concerns refer to concerns that deal with questions of life meaning, rendering all other considerations secondary. In other words, as long as a person is occupied with these existential matters (which are, of course, legitimate) he or she can be considered religious. However, although religion deals with ultimate concerns, it is not in itself the same as ultimate concerns. As a form of analogy, this would be similar to confusing a type of food with the feeling of hunger; the existence of the need itself must not be confused with the means whereby the need is satisfied. As discussed previously, religion has at its core a supernatural element as well as the human relationship with it, and this is the way most people understand it. Therefore, the likening of religion to the over-encompassing concept of ultimate concerns departs from conventional ways of thinking about religiosity, and in turn, makes non-religious perspectives difficult to conceive. In this fashion, non-religion is defined out of existence (Campbell 1971).
A second assumption stems from the first, where a similar tendency for universalization is applied to the concept of faith. As with religion, faith is extended beyond the faithful and is incorporated as part of everyone's life, regardless of people's religiosity or lack thereof. The work of theologian James Fowler on faith development has proved influential in this respect, generating theory as well as methods of research in the form of faith development interviews (Fowler et al. 2004;McDargh 2001). Notably, the use of faith development interviews to gather qualitative data on all aspects of a person's faith is problematic when applied to non-religious people. For example, a cross-cultural study on deconversion conducted by Streib et al. (2009) found that participants who opted for non-religious worldviews were assigned even higher stages of faith than religious people. Given that atheism is characterized by the absence of faith (Altemeyer and Hunsberger 1997), it appears counterintuitive to state that those who identified as non-religious had a sense of faith more developed than religious persons themselves. It certainly seems that Streib et al. (2009) have conflated theology with social science bringing theological concepts into research that is arguably psychological in nature (Gooren 2011).
Thirdly, existing theories within the psychology of religion suggest a positive link between higher levels of religiosity (and spirituality) and physical and mental health ). However, most of the research that has been considered to propose this link suffers from the same caveat: a failure to differentiate those with low religiosity from the non-religious. Rather than addressing the full spectrum from non-belief to strong belief, studies typically use measures such as church attendance, thus confounding religiosity with other variables like social capital, support, and group participation (Galen 2014). Positive mental health outcomes are better explained by the certainty and commitment to one's worldview rather than the specific content of the beliefs themselves. Conversely, lower levels of wellbeing correspond to uncertainty and doubts about cherished beliefs (Donelson 1999;Galen and Kloet 2011;Ross 1990). Furthermore, Kier and Davenport (2004) argue that uncritically accepting that religiosity promotes mental health, as a number of researchers have claimed (e.g. Hill and Pargament 2003;Miller and Thoresen 2003), or that spirituality is a necessary part of the human condition (Sue et al. 1999), falsely implies that atheists experience poor mental health.
Perhaps one way to counter these biases is to promote the study of deconversion not only within the psychology of religion but also, as Lee (2012) suggests, within "nonreligion" studies. Non-religion studies is a relatively new field that focuses on phenomena such as atheism, agnosticism, secularism and their relationship with other areas of research. Ultimately, this could lead to an all-inclusive area of study, encompassing both the religious and the nonreligious, under an umbrella term sometimes referred to as the "psychology of worldviews" (Koltko-Rivera 2004; Nilsson 2013).

Main findings in deconversion research
The literature on deconversion suggests that deconversion involves a very gradual change, which tends to happen during adolescence and early adulthood (Altemeyer and Hunsberger 1997;Brewster 2014;Hunsberger and Altemeyer 2006;Streib et al. 2009;Zuckerman 2011). In his examination of autobiographical accounts by several writers and intellectuals who experienced deconversion, Barbour (1994) argues that there are four dimensions involved in the process: intellectual doubt regarding the system of beliefs; moral criticism towards a way of life; emotional stress and suffering; and the repudiation of the individual's former community. These four aspects of deconversion are commonly driven by discrepancies between a) what one would expect from the world if religious tenets and beliefs were true and b) what one sees and experiences in life.
Furthermore, Barbour (1994) indicates that the abandonment of religious belief is also linked to striving for personal independence and identity in the immediate context of a religious family. In a similar vein, Brewster (2014) states that deconversion in the United States, where the majority is religious, is related to a process of individualization such as leaving the family home to attend college. Also relevant is Smith's (2011) research on how the atheist identity is developed. Based on an American sample, Smith argues that atheists form their identity based on the rejection of theism. Although the process of self-definition involves establishing "who we are" and "who we are not", the atheist identity relies heavily on the concept of the "not-self". That is, atheists create their identity by emphasizing the behaviors they do not engage in and the beliefs they do not hold. Furthermore, Smith (2011) argues that compared to the person of faith, whose behavior is dictated by internalized religious doctrine and norms, the atheist's identity does not fall within the socially defined roles of theism, but is grounded instead in personally created meaning and life events. Therefore, it is mostly idiosyncratic or "biographical" in nature, what Smith describes as an "achieved" identity.
A recurrent theme in the literature is the undermining effect of an intellectual orientation on religiosity (Altemeyer and Hunsberger 1997;Barbour 1994;Caplovitz and Sherrow 1977;Fazzino 2014;Galen 2014;Hunsberger and Altemeyer 2006;Hunsberger and Brown 1984;Zuckerman 2011). For example, drawing on data from interviews with students who became atheists after a strong religious upbringing, Altemeyer and Hunsberger (1997) affirm that these students had considerably more questions and doubts compared to their religious counterparts. According to the authors, these questions were characterized by belonging to the realm of ideas rather than emotional matters. Many students had issues with the veracity of stories in the Bible, their incompatibility with science, and the likelihood of God's existence. In trying to resolve these issues some students asked for help from other people but were typically unsatisfied with the answers received. Altemeyer and Hunsberger reported that these were "very bright" academically successful students who could not resort to faith to find answers either, since they did not view faith as something virtuous, as most religious people would, but as something that severed them from rationality. Moreover, reading about history, science, social issues, and religion has been identified in research old (Vetter and Green 1932) and new (Brewster 2014;Hunsberger and Altemeyer 2006) as a catalyst for the loss of faith.
In search for an explanation as to why deconverts seem to be so compelled to find answers to their concerns, Altemeyer and Hunsberger (Altemeyer and Hunsberger 1997;Hunsberger and Altemeyer 2006) present an unexpected idea: this "commitment to truth and integrity" stems from religious training itself. This means that successful religious training would instill students with a drive to seek truth and integrity to such an extent that they would also assess the truthfulness of their own religion, should they detect any issues within their belief system. Barbour (1994) also considered concern with the truth as an important element in deconversion, but argued that this concern comes from a duty towards intellectual honesty and the ethics of belief, stating that every person has a moral responsibility to justify his/her beliefs with enough evidence.
In addition, the literature shows that deconverts were influenced by experiencing conflicts between religious views and how they felt towards issues related to abortion, sex, gender inequality, women's-and LGBTQrights, and the questionable behavior of some religious figures (Barbour 1994;Altemeyer and Hunsberger 1997;Hunsberger and Altemeyer 2006;Zuckerman 2011;Brewster 2014;Fazzino 2014;Galen 2014). Describing how deconverts dealt with these issues, Altemeyer and Hunsberger (1997) noted that individuals prioritized freedom of thought and, having abandoned their previous beliefs, they expressed feelings of confidence, openmindedness, and being true to themselves. Reflecting on her research participants' troubled relationship with religion, Brewster (2014) states that none of them left their faith because of being "angry at God" or as a rebellious act; rather, these deconverts abandoned a belief system that was never personally chosen but handed down from their parents during childhood. This reaction against the teachings and traditions received during upbringing is similar to what Hunsberger (1980) noted in his observation that doubt and disagreement with parental religious doctrine predicted future deconversion.
It is also worth mentioning the work of Dennett and LaScola (2015), who interviewed (ex) members of the clergy as well as seminary students and professors who had lost their religious beliefs. This piece of research is of interest because it deals with drastic cases of deconversion: a faithful priest who becomes an atheist, going from one end to the other in the spectrum of religious belief. Naturally, individuals involved in the ministry would have their own particular challenges when abandoning belief, but ultimately these cases still illustrate clear examples of strongly committed religious persons who experience religious deconversion. Dennett and LaScola found that their participants' experiences in the seminary, where the Bible and religion were taught through historical and critical approaches, had an unexpected negative impact on their faith. The "demythologizing" of scripture and academic approach to religious belief led participants to not only start to have questions and doubts but also feel disappointment in a place where they were hoping to find safety and structure. The preparation they received did not equip them with the necessary skills to do the preaching their congregations were expecting. Moreover, as their faith eroded, they started to feel hypocritical about the work they were doing.
Research on deconversion is still in its infancy and to date has been mostly explorative, descriptive, and atheoretical in nature. This study seeks to further the understanding of this particular phenomenon, by proposing a model of deconversion that could be used as a possible starting point for future theoretical developments. In light of these aims, the present study asks, "how do religious people become atheists?" to elucidate the deconversion process, that is, the process by which an initially religious person abandons his/her beliefs and adopts an atheist identity.

Methodological approach
The present qualitative study was conducted using a grounded theory (GT) approach. GT was developed as a set of procedures to collect and analyze data in a systematic and cyclical fashion in order to allow concepts to emerge and generate theories grounded in data (Glaser and Strauss 1967). This methodology was fundamental to the development of qualitative research in the social sciences (Kenny and Fourie 2014), and today three main versions of GT can be recognized (Willig 2013): the original, often referred to as "Classical" or "Glaserian"; the approach proposed by Strauss and Corbin (1998) or "Straussian"; and a version put forward by Charmaz (2006) based on a constructivist paradigm.
We have chosen GT because it offers the opportunity to rigorously generate theory from data in those areas where theory is lacking or poorly developed (Henwood and Pidgeon 1992). Specifically, we used the Straussian version of GT (Strauss and Corbin 1998) due to its coding techniques. Coding involves applying labels or "codes" to small chunks of data to identify "categories of meaning", progressing from a descriptive to an analytical level (Willig 2013). Straussian GT proposes a coding paradigm "that explicitly focuses upon, and thus alerts the researcher to, manifestations of 'processes' and ' change' in the data" (Willig 2013, 73). By definition, deconversion involves change and the process associated with it, and therefore this coding paradigm dovetails with the aims of this study. Furthermore, Straussian GT "is concerned with how participants interpret, act, and interact with the studied phenomena" Fourie 2015, 1283). Applied to the study of deconversion, this perspective can stimulate valuable insights, since it would sensitize the researcher to look at how participants "interpret, act, and interact" with religious belief and deconversion and the consequences of such actions.
A good example of the application of Straussian GT can be found in Pinto and Moleiro (2015). The authors identified a dearth of research on transsexuality within the Portuguese context and conducted a study to understand "how transsexual people recognize, acknowledge, and come to terms with their gender identities" (2015, 12).
Applying Straussian grounded theory, Pinto and Moleiro presented a theoretical model consisting of five stages of gender identity development and transition triggers between stages.

Data collection
Data was collected from two sources: testimonies from The Clergy Project and interviews with participants recruited through Atheist Ireland. The Clergy Project (hereafter CP) is a non-profit international organization that provides support for former clergy members who no longer have religious beliefs. This association, based in the United States, has already attracted the attention of some researchers; Dennett and LaScola (2015) interviewed many of these ex-clergy members as part of their study, as mentioned in the previous section. In the present study, all 30 online public testimonies from the CP website were used (The Clergy Project 2016). Clergy testimonies came from mostly male individuals from the United States (with a few exceptions), who defected from different religious denominations; these were mainly Baptist, Methodist, Evangelical, Pentecostal and Presbyterian. The available demographic information from the testimonies is shown in Table 1. These testimonies were collected and used with the permission of the Board President of the CP. They were compiled into a single document for analysis to be worked on as a whole, since individual testimonies varied in length, averaging 1084 words each.
Participants for the semi-structured interviews were recruited through advertising on the social media page of Atheist Ireland. Atheist Ireland is an Irish atheist advocacy group founded in 2008 that promotes atheism and reason, and an ethical and secular society without special treatment for any religion (Atheist Ireland 2016). A total of six voluntary participants were purposely sampled based on relevance to the research question (i.e. they were formerly religious and currently identified as atheists). All participants came from Christian denominations, with a strong or fairly strong religious upbringing, involving religious practices and/or education. With the exception of two participants, their deconversion happened in the Republic of Ireland. Participant B's story of deconversion unfolded in Northern Ireland, where she is from. In the case of participant C, her story of deconversion occurred in South Africa (participants' names are replaced by letters and personal information is avoided to maintain anonymity). Their general demographic information is presented in Table 2. Interviews were arranged with each participant at a convenient location; two were conducted via Skype. The interviews were semi-structured and questions were designed to foster conversation and stimulate explanations. Consistent with a grounded theory approach, questions were modifiable according to the data analysis from previous interviews. This resulted in three slightly different versions of the interview questionnaire throughout the course of the study. Examples of interview questions are "What were the factors or events that played a role in your deconversion and how are they related?" and "How did you start losing your faith or belief?" Interviews were conducted in English, lasted between 30-60 minutes, were audio recorded and subsequently transcribed verbatim.
Although the two datasets came from different sources, their combination is justified since all data depicted the views and experiences of individuals who at one point were strongly religious (whether they had been in the priesthood/ministry or not) and subsequently deconverted. For the purposes of the study, what is important is that both testimonies and interviews show the mechanisms whereby a religious system of belief is replaced by an atheistic worldview. Moreover, most of the individuals in these samples came from two countries (the United  States and Ireland) that share a common trend: an accelerated secularization from a Christian tradition and a weakened relationship between national identity and religion (Cragun 2017). Therefore, the analysis of the stories and experiences of deconversion of these 36 individuals serves to address the research question.

Data analysis
In line with a GT approach, the study was conducted systematically and consisted of a series of alternated stages of data gathering and data analysis (Strauss and Corbin 1998). Once the data from the CP testimonies was compiled, the first stage of analysis comprised the following steps. First, open coding forced a line-by-line in depth analysis to discover concepts in the data. Next, axial coding generated categories, which are concepts at a higher level of abstraction representing a problem, event or phenomena that seems to be significant to the process under study. Coding consists of analytical strategies such as asking questions to the data (e.g. how, what, when, etc), comparing concepts and possible meanings, hypothesizing, the "flip-flop technique" (thinking about opposite concepts to elicit new insights) and "waving the red flag" (paying close attention to words that might indicate bias). After this the main categories (and subcategories) were selected and articulated as contexts, conditions, interactions, and consequences of the central phenomena to yield an initial deconversion explanatory framework. The technique of constant comparison, involving the examination of similarities and differences between instances of a phenomenon, codes, and categories, was applied during every stage of the analysis. Constant comparison allows for the identification of the common thread within a category as well as the differences between its component elements or subcategories (Willig 2013). There were three more stages of data analysis, this time using the data gathered from two interviews in each stage. Open coding, axial coding, and the selection of main categories were applied as before. Subsequently each participant was presented with the analysis of their individual contribution to obtain validation and feedback. They were asked whether the analysis truthfully captured what they tried to express, and if anything had been missed, overemphasized or misinterpreted. All participants responded, expressing high levels of agreement, positive feedback and some more valuable comments. After analyzing the explanatory framework derived from the CP testimonies in the first stage and the two interviews of the second stage, a general theory started to emerge. The emergent theory evolved twice more during the third and fourth stages of analysis, to finally reach the current proposed model of deconversion.
Memo writing, an important element of GT, was carried out throughout the process. This was useful to keep a record of the analysis and theory development, capturing ideas, reflections, questions, and alternative explanations. Triangulation between interviews and the CP testimonies, representing individuals from different backgrounds, offered converging categories adding more reliability to the findings.

Reflexivity
Reflexivity is a necessary element of qualitative research, where the researcher must consider how his/her beliefs, experiences and assumptions influence the investigation at hand (Watt 2007;Willig 2013). As the main author I turn now to first-person voice to address this section.
I grew up moderately religious, however, during my mid-twenties I started to reflect on my religious views. This, coupled with a growing interest in mythology, philosophy and science, led me to abandon my beliefs and become an atheist. Naturally, I had my own pre-assumptions and ideas on the topic, but I made a conscious effort to put them aside from the beginning of the project. I tried to maintain a neutral and naive attitude throughout, making sure my findings were grounded in the data and being aware of my own confirmation bias. From an epistemological standpoint, although I favour the view that social reality is mainly socially constructed, I believe that there is an objective reality behind it that can be accessed to some degree, and I avoid notions of extreme relativism and postmodernism. From the outset of this study I have speculated that there would be some factors that, independently of each particular individual, would be common to most people going through deconversion.

Results
The full analysis generated a proposed model of deconversion (See Figure 1). The three emerging core categories of this model include: reason and enquiry; criticism and discontent; and personal development. Each category was further divided into more specific subcategories. The core categories, while strongly interlinked, were still identified as separate emerging themes. While certain categories were more influential than others depending on the individual, all three were generally present in the process of deconversion. Taken together, our findings reflect the process of deconversion as a gradual progression characterized by a cumulative effect and by significant events or milestones that hold special meaning for individuals. This combination of factors occurs within the individual's close context represented by family, friends, and local community, which, in turn, is embedded in the wider context of society and culture at large. The following sections further detail each of the main features of the model of deconversion.

Reason and enquiry
A strong cognitive element, the core category of reason and enquiry is characterized by an intellectual impetus, the use of reason, and the act of questioning. Doubt was pervasive and it emerged during the early stages of the analysis. Most of the time doubt prompted nagging questions that felt unavoidable and needed to be asked, sometimes starting as early as childhood. Participant B recounted a memory from one Sunday in church: "the first question out of my mouth was 'why are the dinosaurs not in the Bible?' {I was} very quickly put to the back of the room and told 'be quiet and don't ask questions'". In other cases doubts were ignored and suppressed until they accumulated and became too numerous and impossible to evade.
Participants also recalled how faith-based answers to these questions were particularly unsatisfying, since they felt that answers grounded in logic and everyday experience were more appropriate. Reflecting on the factors that played a role in her deconversion, Participant C commented: Every time I tried to ask something I would always be told I just had to have belief and have faith, like no one could answer any questions that I had, you know, it was oh well you just believe just have faith, faith, faith, and you know, I wanted answers and I couldn't get them, I just had to believe… A former minister from the CP articulated clearly the role of reason and logical explanations in his experience: It was the logic of science that killed my belief in a god. I discovered that astrophysics has a more provable story of how the cosmos came to exist; and that biological evolution has a more testable account of how humans were created.
The development of critical thinking was, for many, the key to find the answers they were looking for and to make sense of the world. Sometimes this was aided by having attended higher-level education, by talking to different people who would encourage participants to start wondering about different issues, and by reading books and other materi-als. This analytical approach led deconverts to reflect on a variety of topics pertaining to religion and belief including: the origins of religious dogma; contradictions and historicity of the Bible; incompatibility with science; inefficacy of theological arguments; and the "problem of evil". Critical thinking was also the means by which previous religious experiences were reinterpreted in non-supernatural terms, as this extract from the CP testimonies describes: Then I started thinking, in all my years of prayer, how many times had I actually heard the voice of God? How many times had I seen His presence work to answer my request? My memories were full of emotional outbursts, personal intuitions, and positive thinking. What if my God had never intervened for me? What if I had wanted to believe so much that I had given credit to divinity for what humanity had accomplished?… As can be observed from the previous quote, it was common for deconverts to show skepticism, examine previous assumptions, and consider alternative explanations while avoiding a reliance on feelings and emotions.

Criticism and discontent
The second core category, criticism and discontent, is defined by a disapproval of religious institutions, ideas, and behaviors on moral and ethical grounds which, in turn, generates feelings of detachment and rejection. For some, both of these feelings served as the starting point for the questioning of religious belief. For others, these feelings intensified any existing doubt and questioning, as described above. This theme was originally conceptualized as disappointment (with God, with the Church, with religious people); however, further analysis found that criticism and discontent was a more appropriate description of the data. This change was justified on the basis that disappointment does not necessarily rule out disbelief; for instance, a person might be disappointed with the clergy but still believe in God.
As an institution, the Church was frequently condemned by participants for its stance on social issues. For example, when asked about their process of deconversion Participant F answered: I think the thing that started me questioning was, would have been around the time when HIV, when the AIDS scare was really at its peak, and the Church's stance on that, also at that time I knew friends who were coming out as gay, and I remember you know, the Church's stance calling AIDS the 'gay plague', and it just, the lack of compassion, didn't add up to the people in my life that I love… Discontent with the Church's stance on other issues, including contraception, abortion, mistreatment of women, and child abuse, were also present during interviews. A concern with fairness and justice was also apparent, with some participants and former clergy complaining about lack of support, feeling they were unfairly judged or that other religious people were unsympathetic to their needs. They also witnessed the same kind of treatment towards others.
Hypocrisy and bigotry from some religious individuals and leaders also fostered discontentment with religion and undermined belief. Once example was religious influence on politics and incitement of intolerance towards other groups. Double standards were easily spotted by some participants and had a lasting negative effect on the participants' image of religion and on the perception of the religious individuals performing those behaviors. Participant D described a group of people that he personally knew who were excited about going to an important religious event; however, as he describes, they did not always adhere to religious standards: But I knew these guys quite well, and three of them had girlfriends as well as wives, they were, many of them were drinking too much, two of them probably were gambling addicted or playing the horses quite a bit… Other issues raised in this category included the malleability of canon law by men, how barbaric episodes from the Old Testament could serve as teachings in the modern day, and why a loving God would send people to hell. These concerns were not merely critiques, but were also indicative of a deeply troubled relationship with belief.

Personal development
The final core category, personal development, is characterized by a process resulting in the discarding of religious belief and the desire for a greater sense of freedom, empowerment, honesty, and an interest in self-knowledge. Here, differences emerged between the experiences outlined in the CP testimonies and those from the interviews. When former pastors and ministers started to have doubts about their beliefs they had a greater interest than interviewees in maintaining and strengthening their faith. Arguably, they experienced their waning faith more painfully, feeling ignored by God at the beginning of their deconversion. Former clergy members had to entirely change their way of life, leaving a congregation, finding other means to make a living, and even facing divorce (clerical marriage is allowed within protestant denominations). They also faced cognitive dissonance when preaching something they no longer believed. For some, this experience was difficult and traumatic. However, others saw this as a positive experience that led to knowing themselves better.
Emerging from both CP testimonies and interviews were feelings of low self-esteem and unhappiness associated with a time of crisis in religiosity. In this case, depression was not seen as a cause nor consequence of unbelief, but rather as a state that once resolved, had a direct impact on the belief system of the individual. This is best exemplified by Participant B when she explained: The life I had up until about 25 was pretty abysmal and I think that breakdown and that sort of 'why me?' you know, sort of, that's why I always describe it as 'the cold light of day', when I had the breakdown everything was stripped away and I was, you were left with reality, and then because I was building on the confidence and self-esteem, along with depression you know that comes from the childhood, and all of that sort of accumulated into the, I don't need this extra addition, this fantasy that someone is punishing me… In some cases, deconversion was intertwined with a process of facing and overcoming fears, guilt, internal conflicts, depression or family issues. For instance, thinking about how she lost her belief, Participant E explained: It was really difficult I think it was, kind of scary, I'm questioning too much I shouldn't be doing this, I'm going to hell and these kind of issues… and then… to be honest I was a complete mess for a few years, but then I think what help kind of shift to the final letting it all go and being ok with it was becoming involved with other people who had been through the same thing… Reading was a recurring code throughout the data, as many interview participants and ex-clergy members discussed the profound effect of reading books on their process of deconversion. Reading about alternative ideas on the Internet and widening their social circle by meeting new people were both cited as contributing factors to the realization that having doubts was more common than they thought, that they were not alone, and that nothing bad would befall them should they abandon their faith.

Context and gradual change in deconversion
In all cases, deconversion occurred within a dynamic environment whereby different factors from the individual's close and wider contexts acted as inhibitors or facilitators to the process. Here, the close context is defined as the influence of family -and the religious socialization and expectations of conformity that can come from it -as well as the local community, which can reinforce or relax religious practice. Within the wider context several factors can be found: links between national identity and religion, educational settings (e.g. Catholic or non-denominational schools), pressure to conform, the relationship between gender and atheism, and the minority status that atheists still have in many places. Some of these contextual influences were discussed by Participant A: I did two years of therapy, because I wanted to get all that crap that I've taking on from my parents and what not, that they indoctrinated me with and you know the whole culture really indoctrinates it, you know, everybody has a little enabling part of it and I wanted to get rid of that, {…} and you know a big part of that was having the strength to go counter culture and become atheist, because your shooing your whole culture doing it, and when you do that you are on your own, because {…} we are growing but we are not a majority you know… Importantly, the process of deconversion usually stretched over long periods of time, from several months to many years. Thus, the strength of an individual's belief system typically weakened before adopting an atheist identity, going through stages of spiritualism, theism, or agnosticism. This gradual change can clearly be seen in this extract from CP testimonies: So my journey from a committed Baptist with Sunday School attendance pins down to my navel to atheism was a slow but progressive process. Nothing sudden. Just one hint after another that a god as understood in theistic religions made no more sense to me.
Lastly, in most cases, individuals articulated their stories of deconversion around personal episodes that were specific and loaded with meaning, making their journeys akin to others, but at the same time particularly unique.

Discussion
The proposed model of deconversion presented in this study is compatible with previous findings in the research on deconversion and to broader concepts in psychology. Namely, the three core categories of the model are largely consistent with Barbour's (1994) dimensions of "intellectual doubt" (reason and enquiry), "moral criticism" (criticism and discontent) and "emotional suffering" (some aspects of personal development). Barbour's fourth dimension of "disaffiliation from community" was less emergent in the data collected as part of this study, although it is still related to the general feeling of detachment and some aspects of the core categories (e.g. hypocrisy and bigotry from religious leaders/congregation and widening social circles). Furthermore, participants and CP testimonies' experiences related to learning and education, meeting new people with different ideas, resolving conflicting beliefs, and dealing with religion-inspired fear closely resemble the narratives found previously (Brewster 2014;Zuckerman 2011).
Reason and enquiry recalls the intellectual orientation stressed by Altemeyer and Hunsberger (Altemeyer and Hunsberger 1997;Hunsberger and Altemeyer 2006), and a connection could be made also with the effect of "wide reading" described by Vetter and Green (1932) as key to facilitating deconversion. Caplovitz and Sherrow (1977) had also counted the influence of intellectualism among their causes of abandoning religion. The types of questions that participants and many former clergy had asked themselves, and the kind of answers they were looking for, are akin to those expressed by Altemeyer and Hunsberger's (1997) participants, belonging to the realm of ideas and demanding logical explanations. Not surprisingly, the inability to take faith as an answer to quench growing doubts was another similarity. Altemeyer and Hunsberger's (1997) hypothesis of successful religious training planting the seed of its own demise by promoting truth seeking and integrity is, however, hard to corroborate within this study. Our results showed that a commitment to truth was generally underpinned by an inescapable need to make sense of both the world and personal experiences, while requiring evidence-based rather than faith-based answers. It could be argued that this sentiment is more strongly related to Barbour's (1994) notions of intellectual honesty. Moreover, instead of a passion to seek the truth, the religious training and upbringing of CP members and interview participants brought, at times, fear, guilt or confusion, mirroring previous findings (e.g. Dennett and LaScola 2015;Zuckerman 2011).
In criticism and discontent, the ethical and moral issues that emerged as factors weakening religious belief and detaching individuals from institutionalized religion also echoed some of the concerns expressed by atheists in previous studies (Altemeyer and Hunsberger 1997;Barbour 1994;Hunsberger and Altemeyer 2006;Vetter and Green 1932;Zuckerman 2011). Examples of these are the Church's position in relation to homosexuality, gender, and human rights issues. Here, detachment from religion resulting from these appraisals was neither due to disappointment with religion (as explained in the change of category name in the results section) nor to being "angry at God", as Brewster (2014) and some interview participants pointed out. Instead, and more consistent with social identity theory (Tajfel and Turner 1979), religious beliefs were no longer internalized as part of the selfconcept and the social identity that religious membership afforded became inadequate and negative, driving individuals to disaffiliate from the religious community.
In regards to the core category of personal development, it seems plausible to draw some connections with selfactualization if we consider how some participants had to overcome internal conflict and fear to achieve higher levels of well-being. Schnell and Keenan (2011) argue that for highly committed atheists, self-actualizing behaviors are part of the sources used to find life meaning. In Maslow's (1954) study on motivation and personality, few of the self-actualized people he described were religious. Some characteristics of self-actualization (Maslow 1954) seem to be present in the process of deconversion here described: namely, self-acceptance (contrary to lack of self-esteem related to a former religious self), autonomy (breaking from religious norms and socialization) and strong ethical standards (as evidenced in their discontent with religion's stance on certain topics). However, this is not to imply that these characteristics cannot be present within highly religious people; as previously mentioned in the introduction, well-being is associated with a high commitment to one's worldview, whether that worldview is religious or atheistic (Galen and Kloet 2011).
From a cognitive standpoint, dual process theories of cognitive styles have been proposed to explain how people differ in their propensity to have religious and supernatural beliefs. Although dual process theories have been described in different ways, they all propose two different kinds of processing that people engage in. System one (or heuristic) is usually characterized as fast, unconscious, effortless, intuitive, and automatic, whereas system two (or analytic) is slow, conscious, reflective, effortful, and deliberate (Evans 2008). A substantial body of research indicates that those individuals with a tendency for an analytical cognitive style are less likely to hold religious beliefs (Aarnio and Lindeman 2005;Aarnio and Lindeman 2007;Lindeman and Aarnio 2006;Lindeman and Aarnio 2007;Gervais and Norenzayan 2012;Shenhav et al. 2012;Pennycook et al. 2012;Pennycook et al. 2013;Pennycook et al. 2014a;Pennycook et al. 2014b;Bahçekapili and Yilmaz 2017;Saribay and Yilmaz 2017). Moreover, research shows that the negative relationship between a proclivity for analytic reasoning and religiosity holds even after controlling for cognitive ability, personality factors, and demographics (Shenhav et al. 2012;Pennycook et al. 2012;Pennycook et al. 2013). Conversely, relying on an intuitive and heuristic cognitive style more likely leads individuals to be religious and to believe with more confidence (Lindeman and Aarnio 2006;Aarnio and Lindeman 2007;Shenhav et al. 2012;Svedholm and Lindeman 2013;Saribay and Yilmaz 2017). One of the cognitive mechanisms by which analytical thinking weakens religiosity is conflict detection (Pennycook et al. 2014a). Many religious beliefs are counterintuitive and violate the expectations set by our experience of the natural world and an analytical style would detect inconsistencies and conflicting information, thus undermining belief. As has been stated in the current study and in previous research (Altemeyer and Hunsberger 1997;Brewster 2014;Hunsberger and Altemeyer 2006), detecting and questioning inconsistencies, not only in religious doctrine but also in religious behavior is a common occurrence during the deconversion process. Other cognitive mechanisms characteristic of analytic cognitive style, as shown by experiments involving tasks in deductive reasoning, are a better predisposition to avoid belief bias, and a tendency to spend more time in problem solving (Pennycook et al. 2013). An inclination towards an analytical style among deconverts is the cognitive counterpart of the intellectual orientation observed by Altemeyer and Hunsberger (1997), and is represented within the core category of reason and enquiry of the present theoretical model.
At this point some clarifications are in order. First, it should be acknowledged that, arguably, ex-clergy members' experiences and consequences of deconversion may be especially strong given their particular predicament. For instance, as expressed in the core category of personal development, CP members had to face greater challenges finding new occupations. The presence of CP members' testimonies as part of the sample might have given more intensity to the findings. However, all the elements of the theoretical model (the core categories, the contexts, and the gradual character of change) were present, at a greater or lesser extent, in the experiences of CP members and interview participants alike.
Second, it is important to note that the process of deconversion does not necessarily exclude an agnostic outcome. Moreover, it is entirely possible to experience a similar process and still remain religious. The proposed model of deconversion put forward in this study accounts for the experiences of the study's participants and the CP members who shared their stories online. Although the implications of the findings potentially relate to factors that affect more people, both individually and in society at large (e.g. cognitive styles, self-esteem, organized religion), generalizations should be approached with caution. Furthermore, GT approaches are always provisional, the product of which is open for expansion, revision or modification (Willig 2013). More importantly, and as Zuckerman (2011) pointed out, deconversion, the same as religious socialization, is always culture-bound and it would be inappropriate to attempt to formulate a theory that aims to be applicable to every time and place.
This study is not without limitations. It would have been beneficial to interview religious people who went through a similar process but remained religious in order to compare responses and enrich conclusions. Also, the data only includes deconverts from Catholicism and from Protestant denominations, therefore leaving open the question of how relevant these findings could be to deconversion from other major religions (e.g. Judaism, Islam, Hinduism).

Conclusion
This study aimed to consolidate current psychological knowledge about deconversion through theory building. The resulting theoretical model of deconversion depicts this phenomenon as a process, a gradual change, in which the degree of belief wanes over time and is marked by significant events. This process reveals an interaction of three interrelated factors: reason and enquiry, criticism and discontent, and personal development. Deconversion therefore appears to be driven by an intellectual impetus, by moral appraisals of beliefs and institutionalized religion, and by overcoming internal conflicts. Furthermore, this process does not occur in isolation but rather within the context of family, friends and community, and also the wider context of society as a whole. These findings are consistent with previous insights from the literature on deconversion.
Future research would benefit from a number of recommendations. First, diverse psychological theories should be integrated in order to understand each aspect of the phenomenon in more detail; for example, apart from the areas already mentioned, deconversion could also be combined with existential issues using terror management theory and theories of moral reasoning. Second, issues of faith should not be approached from a theological perspective, but instead from a psychological perspective; for instance, faith could be analyzed as a discursive practice with qualitative methods. Third, deconversion research would benefit from a multidisciplinary dialogue with other social sciences, incorporating ideas from sociology, anthropology and philosophy.