1From a Western standpoint, negatively perceived portrayals about the entanglement of religion and politics overshadow the perception of Turkish Muslims and the topics of fundamentalism and Islamism seem to dominate international scientific studies (Delibas 2015; Demmrich, Hanel 2023). Yet, it is well-known that Turkish society is religiously and politically divided (Kehl-Bodrogi 2003; Mardin 1962), thus, the topic of the secularization of Turkish society is of great interest within the social sciences (Çokgezen 2022; Ertit 2018; Gezici Research Institute 2021; Hecker 2022; Konda 2018; Mardin 2005). Turkey has long been, and still is, a context in which Islam takes various forms, and religious and spiritual life has always been a dynamic issue (Ardınç 2008; Mardin 1991, 1993; Özay 2007; Ünal 2013). This is, generally speaking, because Islam itself is differentiated depending on specific historical, cultural, and political developments in Turkey (see e.g., Ağılkaya-Şahin et al. 2021).
- 1 This occurred between the 8th and 10th centuries AD before the migration of Turks from Central Asia (...)
2It should be noted, that Turks were introduced to Islam primarily through Sufism1 (Ağılkaya-Şahin 2015), “a mystical Islamic belief and practice in which Muslims seek to find the truth of divine love and knowledge through direct personal experience of God” (Schimmel 2019), which ideologically and practically supported a liberal interreligious and interdenominational perspective (e.g., Aslan 2020). However, the various religions and interpretations of Islam have become increasingly officially restricted since the foundation of the modern Turkish Republic in 1923 – and the effect of these restrictions is still being felt today: many non-Muslims, especially Christians, were displaced from Turkish land (Gürsoy 2008) and only one Islamic interpretation, which was considered compatible with the state doctrine of laicism, became the officially accepted Islam under the control of the Presidency of Religious Affairs (Diyanet) and the various faculties of theology (Korkut 2016; Öcal 1986).
3With regard to everyday life, the first fifty years of the Republican era were marked by state-supported laicist policies, which created great tension and left traces of trauma in the memory of the dominant, religious conservative citizens (Sevinç et al. 2017). Eventually, the Islamist opposition, which first launched its agenda with a military coup in 1980, came to power with democratic elections at the beginning of the new millennium and is still in power today (Şen 2010). One of the most salient characteristics of this current period in Turkish history has been the alleged gradual desecularization of Turkish society through the direct and indirect promotion of religious groups, the influence of which is now being felt by Turkish Muslims (Demmrich, Hanel 2023). Once again, it seems that Turkish people today are building a new tension around the over-religionization of public life (Hecker 2022; Okumuş 2021). As a result, there seems to be growing discontent with the traditional religious and with the Islamist characterization of the government and society.
4Such recent societal tensions are also mirrored in individuals’ religious self-assessments. The present study examines this aspect through established categories (religious, believing, non-believing/atheistic) as well as new categories (deistic, religiously indecisive) of religious self-assessment. For this reason, we analyze a dataset of Muslims in Turkey collected previously (see Demmrich, Hanel 2023), which asked for these categories. Our primary focus is on determining the frequency of these self-assessments and understanding their religious and sociodemographic profiles. Based on these profiles, we, finally, aim to identify clusters of religious self-assessment among Muslim Turks.
- 2 Similarly, a decrease from 13 % to 10 % was observed for the category ‘sofu’ (pious) in both studie (...)
- 3 See also Gezici Research Institute (2021), e.g., 28.5 % are religiously not believing.
5A widely cited panel study by the Think Tank Konda (2018) compared data from 2018 with data from 2008 and found a slight decrease in the number of self-assessments as orthodox-religious (dindar, from 55 to 51 %). At the same time, there were slight increases in the categories of (unorthodox) believing (inançlı, from 31 to 34 %)2, atheistic (ateist, from 1 to 3 %), and non-believing (inançsız, from 1 to 2 %). This trend is accompanied by a decrease in religious practices, such as fasting (from 77 to 65 %) and not veiling (from 34 to 37 %). Similar results were reported by a more recent study (Konda 2022). This decline in different forms of religious self-assessment and religious practices seems to be more pronounced in the youngest generation3, who were allegedly raised in a more conservative and religious environment than earlier generations (e.g., Babacan et al. 2021; Çokgezen 2022). However, it is also likely that this is the influence of globalization, especially via new media, and that this is an indication of individualization also occurring on a religious level.
- 4 See e.g., Konda (2018), for 3 % atheists and 2 % non-believers.
6In addition to these few panel studies, emerging developments such as these raise a lot of public interest in the (social) media yet remain, until today, unsystematic. For many Turks, one of the most remarkable manifestations of this religion-based tension was the #10YearChallenge trend on social media (Kasapoğlu 2019; Schlotzer 2019) during which many young women removed their headscarves, one of the pivotal symbols of Islamic movements for several decades. Social media also testifies to a general disappointment with religion and the plan of the ruling AKP to Islamize the youngest generation (see Yalnız Yürümeyeceksin 2022). A quite recent suicide of a university student (Ayçin 2023), who expressed his atheistic identity and feelings of being increasingly overwhelmed by the religious demands of his family and the wider environment, created a nationwide controversy about religion, religious education, and the religious organizations that run student dorms and houses. As a result, it seems that the self-assessment of being explicitly atheistic or non-believing may create very strong barriers within one’s environment and might therefore be avoided by many4. Thus, a relatively new category of religious self-assessment, which seems more compatible with Turkish mainstream society as it concedes to the lowest common denominator – the belief in one God – is emerging: deism.
7For approximately three decades, Western social sciences have been hotly debating spirituality (e.g., Demmrich, Huber 2019; Vincett, Woodhead 2016; Zinnbauer et al. 1997) and, more recently, about new self-assessment categories such as ‘spiritual but not religious’ (e.g., Beyer 2015; Hastings 2016; Parsons 2018). It seems that this is not a well-established self-assessment category among Muslims in general or in Turkey specifically (Gür 2020), however, alternative categories of the understanding of one’s own religiosity may well have emerged.
- 5 These schools are specially designated for religious education.
8Everyday stories, such as the one about the female school student, who went to an imam-hatip school5 and revealed to her mother after school one day that she had become a deist, are flooding Turkish media and sparking controversial debates in the Turkish public (Saruhan 2018). One such debate centers on the definition of deism: some religious public figures and preachers use the word ‘deism’ interchangeably with the term atheism (e.g., Şenlikoğlu 2021; Vanlıoğlu 2022). However, Turkish people who consider themselves ‘deistic’ usually mean that they believe in God without necessarily being religious in a normative, orthodox way or without practicing a religion. It is considered a category that is prevalent and growing especially among the younger generations (e.g., Diyanet İşleri Başkanlığı 2014; Gündoğar,Yürgüç 2019), and some are even predicting a cultural-religious revolution in huge parts of the MENA region similar to the European Enlightenment, which seems to be behind the deistic self-understanding (Doko 2021). However, most academic contributions on deism remain limited to a theoretical-normative level (e.g., Aydın 2019; Aytepe 2017; Başçı 2018; Bozkurt, Kuyucu 2020; Dorman 2021; Erdoğan 2018; Gündoğar 2020; Kaplan 2017; Peker 2018; Topaloğlu 2017).
- 6 Unfortunately, the method lacks qualitative standards so that neither the sample nor the data colle (...)
- 7 Turkish Fen liseleri, are high schools that emphasize science education in the STEM fields.
9A small number of academic papers have begun the process of empirically investigating deism. For example, an initial study by Günaydın (2017) collected data from 11th- and 12th-grade students at imam-hatip high schools6 and argues that imam-hatip school students are under the influence of antireligious modern movements, notably atheism and deism. Two years later, Gündoğar and Yürgüç (2019) conducted a much wider study of N = 1,150 students from different high schools regarding their perception of faith. This survey showed that for the statement 'Belief in God is a necessity' 4.7 % disagreed and 6.6 % were indifferent. In general, it was noted that there is a tendency for up to 20 % of students to lean towards deistic views and that this tendency is also present in up to 8.3% of imam-hatip school students. Atheistic tendencies were found among science high school7 students with a rate of 9.7 %. Similar to other investigations (e.g., Menküç 2019), this study concluded that a tendency toward deism exists but is not widely socially spread at the moment and that it might, therefore, be exaggerated in the media and public.
10In addition to this empirical evidence that the phenomenon exists, some qualitative studies have attempted to shape a deistic self-assessment in religious terms. Firstly, Menküç (2019) conducted semi-structured interviews with N = 34 high school teachers of religious culture and ethics courses regarding the deistic tendencies and related attitudes among their students. Her findings showed that students who tend towards deism hold a very critical attitude towards Turkey’s religious culture, including ”Religious Culture and Morality” courses, the Diyanet, and religious groups and leaders. Secondly, Dikbaş (2021) also interviewed parents in addition to (social sciences) teachers. Her study emphasized the impact of popular culture on young people and the developmental characteristics of the adolescence period. She reported that specific accounts from parents and teachers indicated that some high school students are experiencing cognition-based doubts and questioning religious issues. Her research findings suggest that youths in high school undergo the typical adolescent developmental processes, influencing their level of religiosity. Additional factors contribute to the diminishing interest in religious matters among some of these adolescents, including family dynamics, pervasive social media exposure, insufficient religious education, and the absence of relatable role models. Furthermore, her study highlights a noteworthy trend where certain individuals are gravitating towards atheism and deism as alternative (although marginal) identity choices.
11Three recent contributions focus on deism from a more social-scientific perspective. For one sample from generation Z (N = 701), Çalışkan (2021) showed that 10.4 % of the participants tend toward deism. These results also indicated that higher family income levels and higher parental education are the best indicators of deism tendencies. Ermiş (2019) surveyed N = 400 adolescents and found that the acceptance of deism is positively related to higher education, a higher parental income level, demands for self-determination or detachment from religious instruction, peer pressure towards deism, and the developmentally typical, ambiguous adolescent thoughts and feelings.
12Finally, Koç (2022) interviewed 20 teachers about the topic of deism among their adolescent students and conducted a questionnaire study of N = 338 school students. The results showed that students doubt God’s existence and, therefore, also question religious rituals, revelation and prophecy, and struggle with the question of theodicy. The influencing factors that can lead to deism were noted as the relationship of the students to their parents and the resulting religious upbringing in the parental home, media and film consumption, use of social media, and the public ostracization of religious representatives and institutions due to their misconduct.
13Nevertheless, all these studies focus on the youngest generation, namely adolescents and even minors. Therefore, none of the samples used in these studies can be considered representative of Turkey. In addition, religious indifference or religious indecisiveness, which is primarily reflected in highly secularized contexts (e.g., Demmrich 2016; Demmrich, Blume 2018), was also found to be a socially relevant category in the studies by Gündoğar and Yürgüç (2019) and Dikbaş (2021). No study has, as yet, systematically investigated religious indecisiveness in Turkey.
14Our study aims to present a balanced view: neither Islamization nor secularization is the sole focus of this study but rather the diversity of categories of religious self-assessment among Muslims in contemporary Turkey. Based on a representative quota sample, we first show the rates of established (orthodox religious, believing, non-believing/atheistic) and new (deistic and religiously indecisive) categories of self-assessments. In order to present a religious and sociodemographic profile of Muslims with a certain religious self-understanding, we relate these self-assessment categories to religious (normative, popular, fundamentalism, sense of connectedness, west-compatible Islam, secularization) and sociodemographic variables. Finally, we aim to detect distinct clusters of religious-self-assessment based on the simultaneous inclusion of sociodemographic and religious variables through a multivariate method.
- 8 See supplementary materials for more details.
15We are drawing on a dataset of Muslims in Turkey collected previously (see Demmrich, Hanel, 2023). It used a quota sample of the Turkish population representative of age, gender (both TÜIK 2020), and education level (OECD, 2014) and was recruited with the support of Qualtrics. In addition, the sample is representative of ethnicity, urbanicity (for both, see Doğan 2018), and voting preference (Metropoll 2021)8. Qualtrics’s use of captcha, Geoblocking, and IP addresses ensured that the participants, who completed an online questionnaire between January and the beginning of February 2021, were all real people living in Turkey who participated only once. Filter questions at the beginning of the survey included religious affiliation and age. Any participants under the age of 18 years were not allowed to complete the questionnaire for legal reasons. A median completion time of 20 minutes was indicated by a pre-test with 10 % of the sample in which participants who finished in less than half of the median time were automatically excluded. A further 16 participants were manually excluded due to dubious response patterns (e.g., straight-liners).
16The final sample included 736 participants (18-24 years: 14.9 %, 25-34 years: 23.7 %, 35-44 years: 22.7 %, 45-54 years: 18.1 %, 55-70 years: 20.54 %, 4 missing; 50.3 % female, 0.1 % diverse). While 47.0 % stated that they have no schooling or are currently in school or continuing their education, 18.1 % reported that their highest level of schooling is a primary or secondary school diploma. 33.8 % reported an upper secondary school diploma/A-level as their highest level of education. 88.9 % indicated a Turkish identity, 14.5% a Kurdish identity, and 2.2% another ethnic identity (multiple responses were possible). 73 live in a village, 85 in a town, 204 in a city, and 374 in a metropolitan city. 35.8% expressed support for one of the two parties in government (AKP and MHP) while the rest supported another or no party.
17No power analysis had been performed before data collection but a sensitivity analysis indicated that 736 participants would be sufficient to demonstrate a small effect size of r = .12 with a power of .95.
18Religious self-assessment was measured by the question “How would you describe yourself religiously?” with a single choice answer format of orthodox religious (dindar), believing (inançlı), deistic (deist), non-believing/atheistic (inançsız/ateist), indecisive (kararsızım), and ‘other’ with the option to specify what was meant by this answer.
19Sociodemographic variables included questions about age (in years), gender (1: men, 2: women, NA: diverse), ethnicity (1 = Turk [majority], 2 = Kurd and others [minority]; multiple answers possible), and urbanicity (1 = village, 2 = town, 3 = city, 4 = metropolitan city). Highest educational qualification (based on the International Standard Classification of Education of UNESCO, 1997, ranging from 0 = no degree [yet]; to 3 = upper secondary degree/A-level), and unemployment, which incorporated ‘job seeking’ and ‘not employed for other reasons’ (occupational status: 1 = full-time, 2 = part-time employed, 3 = [university] student, 4 = in vocational training/apprenticeship or in retraining, 5 = job seeking, 6 = retiree, 7 = housewife/househusband, 8 = not employed for other reasons) were also measured (for the sociodemographic representativeness of the sample, see table S1 in the supplementary material.).
20Normative and popular religiosity was assessed using the Normative-Popular Religiosity Scale originally developed by Çoştu (2009) in the Turkish language. Normative religiosity is defined as general religious beliefs and practices based on the Quran and Hadith and is measured using 30 items (e.g., “I believe in the Hereafter [Paradise and Hell] after death.”). Popular religiosity is defined as religious beliefs and practices that are traced back to folk religiosity rather than the Quran or Hadith and is measured using seven items (e.g., “I believe in possession”). Both were measured on a four-point-answer scale from 1 = strongly disagree to 4 = strongly agree and the reliabilities were rated good to excellent (α = .95 for normative religiosity, α = .74 for popular religiosity).
21Religious fundamentalism was measured with a four-item scale (Pollack et al. forthcoming). Example items include “There is only one true religion” and “Only Islam is able to solve the problems of our time” which were answered on a four-point scale from 1 = strongly disagree to 4 = strongly agree (α = .85).
22Religious practices included the frequency of mosque attendance (“Outside the pandemic: How often do you go to mosques or similar places of worship?” from 1 = never to 7 = every week or more often), the frequency of obligatory prayer/namaz (“How often do you pray the obligatory prayer?” from 1 = never to 8 = several times a day), and the frequency of personal prayer/du’a (“How often do you pray personal prayers?” from 1= never to 8 = several times a day).
23Sense of religious connectedness was measured using the questions “To what extent do you feel connected to the Muslim community (ummah)?” and “To what extent do you feel connected to your local mosque community?” and both were answered separately on a four-point answer scale from 1 = not connected at all to 4 = very closely connected.
24West-compatible Islam was assessed using the item “The religion of Islam does not contradict Western values” which was answered on a four-point scale from 1 = strongly disagree to 4 = strongly agree.
25Secularization was examined through a belief dimension (“I used to have a religious belief, now I am no longer a believer”) and a religious practice dimension (“Outside the pandemic: I used to go to mosques or similar places of worship, I don't anymore.”). Both were separately answered on a four-point scale from 1 = strongly disagree to 4 = strongly agree.
26Table 1 shows the frequencies of the single answer categories of religious self-assessment. Interestingly, less than about a third of the sample stated that they are orthodox-religious and the majority regard themselves as unorthodox believers. Another 0.8 % defined themselves as non-believers or atheists and less than 3 % as deists. Among these new categories of religious self-assessment, religious indecisiveness was highest at 3.5 %. One percent of the sample indicated ‘other’, but none of them further specified what this meant to them. As it could be expected that this category represents a very small portion of the sample seeking to avoid answering the question, we excluded this category from further analyses.
Table 1. Total and relative frequencies of the categories of religious self-assessment.
27Table 2 shows the descriptive statistics and results of the ANOVAs and the significant results of the nonparametric Games-Howell post-hoc tests (used because of their robustness against disproportional group sizes). With regard to the sociodemographic variables (upper part of table 2), the groups differ in age, urbanicity, education, and unemployment, but not in gender and ethnicity. The vast majority of the significant differences were detected between the orthodox-religious group and the other groups. In detail, the orthodox-religious group is significantly older than all other groups, more rural than the believing, non-believing/atheistic, and religiously indecisive group, and less educated than the believing group. The orthodox-religious group exhibited, like the believing group, a higher rate of unemployment than the non-believing/atheistic group. However, the sample size of the non-believing/atheistic group was small and all of them were employed. Here, the effect size was the smallest for all significant effects in the ANOVAs. For these reasons, no further interpretation of this effect is presented. As age has been discussed as a fundamental variable when it comes to religious-self-assessment, especially in its new forms, a cross-table between age groups and religious self-assessment is provided as table S2 in the supplementary material.
28With regard to the lower part of table 2, all the religious variables differed between the self-assessment categories. The orthodox religious group scored significantly highest and believing individuals second highest – with significant differences from almost all other groups – on normative and popular religiosity, fundamentalism, religious practices, sense of connectedness to the umma and local religious community. The orthodox religious group scored lowest and believing individuals scored second lowest on both secularization variables and differed significantly from each other.
29Additionally, nonbelieving/atheistic individuals constituted their own group with the lowest scores for normative religiosity and frequency of personal prayer. They also differed significantly from deistic and religiously indecisive individuals regarding normative religiosity. Similarly, non-believing/atheistic individuals showed a significant difference from religiously indecisive individuals regarding the frequency of mosque attendance.
30Interestingly, the understanding that Islam is compatible with Western values is high among orthodox religious and believing individuals (above the scale mean) and low among the other three groups (below the scale mean), with deists constituting the group with the lowest scores that differ significantly from the orthodox religious and believing groups. In addition, deists scored lowest on the belief dimension of secularization and differed significantly from not only the orthodox religious and believing individuals but also from the religiously indecisive group.
Table 2. Results of the ANOVAs and post-hoc tests for the religious self-assessment categories regarding sociodemographic and religious variables.
31Note. * p < .05, ** p < .01, *** p < .001.
32As the variables reported in table 2 can be confounded (e.g., less educated might be older, popular religiosity might be more widespread in rural versus urban areas) and as some categories of religious self-assessment seem to have a similar sociodemographic and religious profiles, we ran a two-step cluster analysis in order to detect any distinct clusters of religious self-categorization. Therefore, we included all ordinal and binary variables presented in table 2 in the cluster analysis and ran an ANOVA with Games-Howell post-hoc tests and the clusters as factors.
33Table 3 shows the results of the analysis, resulting in k = 3 clusters. Cluster 3 (n = 218 ≙ 29.8 %) represents the orthodox religious cluster, which is highest in age, the most rural, and has the lowest levels of education and unemployment. All this cluster’s religiosity variables are highest and both secularization variables lowest. Cluster 2 is the biggest (n = 421 ≙ 57.5 %) and represents the vast majority (91.6 %) of those individuals who identified themselves as believing. Together with Cluster 1, they are the youngest in age, are more likely to live in urban areas, and have higher levels of education and unemployment than Cluster 3. With regard to religiosity, this ‘believing’ cluster shows a medium level of normative religiosity, fundamentalism, frequency of religious practices, sense of connectedness to umma and the local religious community, and the belief dimension of secularization. In addition, cluster-2-individuals score as high as cluster-3-individuals on popular religiosity and West-compatible Islam. However, on the practice dimension of secularization, cluster 2 scores as high as Cluster 1. This smallest cluster, Cluster 1 (n = 93 ≙ 12.7%), represents deists, non-believers/atheists, religiously indecisive individuals, and a small share of individuals who categorized themselves ‘believing’ (8.4 %). In addition, almost all individuals in this cluster identified ethnically as Turks (with slightly significantly lower percentages in Clusters 2 and 3). Cluster 1 also scored lowest, by a significant margin, on all religious variables and highest on secularization variables.
Table 3. Results of the two-step cluster analysis.
34Note. * p < .05, ** p < .01, *** p < .001.
35The variety of religiosities among Muslims in Turkey is an interesting topic and international contributions to this field raise great interest in both scholarly and public circles (e.g., Çokgezen 2022; Hecker 2022; Gezici Research Institute 2021; Konda 2018). However, while earlier studies have been based on representative samples but focused on the extremes of Islamization or secularization (e.g., Demmrich, Hanel 2023; Gezici Research Institute 2021) or have neglected newer categories of religious self-assessment (Konda 2018, 2022), other studies have been restricted to specific religious self-assessment categories, such as deism, among non-representative samples, such as school students in specific districts or cities (e.g. Günaydın 2017; Çalışkan 2021). The present study is the first investigation into both established (orthodox religious, believing, non-believing/atheistic) and new categories of religious self-assessment (deistic, religious indecisive) among a representative quota sample.
- 9 Especially in relation to deism, e.g., Aydın (2019), Doko (2021), Dorman (2021), Kaplan (2017).
36With this focus on the diversity of religious self-assessment categories, the first aim of the study was to show the proportions of these categories. With regard to the established religious categories, a majority of 62 % declared themselves as ‘believing’ (inançlı) and almost 30 % identified themselves as orthodox religious (dindar). Less than 1 % characterized themselves as non-believing or atheistic (inançsız/ateist). This data can be hardly compared to earlier studies. For example, Konda (2018, 2022) found that only one-third declared themselves as believing, half as orthodox religious, and 5-7% as non-believing or atheistic. However, the trend goes in a similar direction as our study, given the portion of believing individuals seems to have grown and the portion of orthodox-religious individuals seems to have shrunk during the last ten years in the Konda study. However, Konda (2018, 2022) neglected the new categories of religious self-assessment, such as deism and religious indecisiveness. In summary, one strength of our study is that participants were given self-assessment choices of religiosities which not only reflect the current status of academic debates in Turkey9, it also represents the developments in individual religiosity that are being discussed publicly (e.g., Kasapoğlu 2019; Saruhan 2018).
37With regard to these new categories, our study found that about 3 % consider themselves deistic which, again, can be hardly compared to earlier studies. Previous studies found higher rates (e.g., 8-20 %, Çalışkan 2021; Gündoğar, Yürgüç 2019) but were only conducted among the youngest generation and mostly using convenience sampling at one location (e.g., a certain school or town). This initial figure for the number of deists among Turkish Muslims in the general public confirms the earlier findings that there is a tendency toward deism but that it is not yet very widely spread at the moment and might, therefore, be exaggerated in the Turkish media and other public forums (Gündoğar, Yürgüç 2019; Menküç 2019). In addition, our study could uncover an initial figure for the proportion of adults who regard themselves as religiously indecisive, which is, with 3.5 %, even higher than the portion identifying deists. Although Gündoğar and Yürgüç (2019) and Dikbaş (2021) indicated a tendency toward religious indecisiveness and disinterest (even among imam-hatip school students) they did not report how many people identified with this religious self-assessment category.
- 10 Muslims make up 90-99 % of the Turkish population, see Konda (2022); Sevinç et al. (2017). Therefor (...)
- 11 Prominent among the youngest generation still under 18 years, see Gezici Research Institute (2021).
38Finally, a limitation of our study must be acknowledged at this point. Due to the scarcity of representative data in the field of religiosity in Turkey, we drew on an earlier data set of a quota sample representative for age, gender, education-level ethnicities, and urbanicity of adults in Turkey. Although Muslims seem to constitute the vast majority of the Turkish society,10 a study among Turks that was not restricted to those with a Muslim affiliation might have returned higher rates in deistic, non-believing/atheistic, and religious indecisive self-assessments, especially in the context of growing secularization and with it, potentially decreasing Muslim affiliation. However, the study was conducted among adults (≥ 18 years) for legal reasons and the quota of the age groups ends with the age group of > 55 years (because of the online format of the questionnaire). This might have influenced the low number of deistic and non-believing/atheistic individuals11 and orthodox-religious self-assessments (prominent among the older generations). Future studies could use different questionnaire methods (e.g., face-to-face with parental consent) and thus be able to include younger and more elderly members of Turkish society.
39Despite the low total numbers of individuals in the subsamples of deism, unbelieving/atheism, and religious indecisiveness, we found significant differences between the religious groups with regard to their sociodemographic and religious profiles. In the context of our second study aim, orthodox-religious (dindar) individuals constitute the oldest subsample, live mainly in rural areas, and tend to have lower than primary school educations. Unsurprisingly, they score highest on all religiosity variables (normative and popular religiosity, fundamentalism, religious practices, and sense of connectedness to the umma and their local religious community, West-compatible Islam) and lowest on both secularization items.
40Believing (inançlı) individuals were found mostly in the younger age groups of 25-34 years and 25-44 years, who are urban citizens with a middle school education. With regard to their religious profile, they show high popular religiosity and score second highest (after the orthodox religious group) on normative religiosity, fundamentalism, religious practices, and both feelings of connectedness. Interestingly, they do not perceive a struggle between Islam and Western values but indicate higher rates than the orthodox religious group on both secularization dimensions (although at a level that is still below the scale mean). This is a new finding indicating that secularization might begin with the self-assessment as ‘believing’ instead of ‘orthodox religious’.
- 12 These results seem to reflect deism as ‘belonging without believing’, see McIntosh 2015.
41Deistic individuals are mostly found in the younger age groups of 25-34 years and 35-44 years but, interestingly, more seldom in the youngest age group (18-24 years) as we could expect based on the public discussion and previous studies of deism focused on adolescents (e.g., Gündoğar, Yürgüç 2019; Koç 2022; Menküç 2019). Deism might, therefore, not simply be an adolescent expression of doubt or detachment from family or their religious school (cf. Dikbaş 2021; Ermiş 2019; Koç 2022). However, we are limited on this point given our aforementioned exclusion of an even younger age group (younger than 18 years) from our study. In our study, deists were not higher educated than all other groups except the religious-orthodox group (cf. Çalışkan 2021; Ermiş 2019). Deists scored lower than orthodox religious and believing Muslims on all forms of religiosity and were characterized by secularization in the belief dimension12. Moreover, they indicated the highest struggle between Islam and Western values among all religious groups. This aspect should be investigated in further studies, which could test whether deism is an outcome of Turkey as a “prototypical torn country” (Huntington 1997: 42) between Islam and the West – especially in the context of increasing globalization.
- 13 But again beyond adolescence, cf. Dikbaş (2021).
42Non-believing/atheistic individuals constituted the smallest religious group. They tend to be younger13 and to live in more urban environments (see also Sevinç, Mehmedoğlu 2016). However, they do not differ from the other four religious groups in their sociodemographic profile. Unsurprisingly, they scored lowest on all forms of religiosity. They indicated that they had stronger religious beliefs in the past but that their frequency of mosque attendance, which is very low, had remained consistent over the course of their lives.
43Finally, religious indecisive individuals were again quite young (none of them were in the > 55 years category) and tend to live in more urban environments. They are characterized by a medium level of normative religiosity but a low level of popular religiosity. In this regard, and with regard to the other forms of religiosity, they scored low and while they resemble non-believers/atheists, they are closest to deists. In contrast to deists, religious indecisive individuals do not struggle overly with a contradiction between Islam and Western values and show significantly lower secularization of their beliefs than deists. These results could stimulate further research on the different reasons for identifying with deistic versus indecisive attitudes towards religion that may evolve over an individual’s lifespan.
44In summary, the strongest sociostructural differences that we uncovered were found between orthodox-religious and all other groups. The orthodox-religious are older, live in more rural environments, and are less educated. With regard to the religious profiles, the sharpest drop was found from orthodox-religious (highest values throughout) and believing individuals to deistic, non-believing/atheistic, and religiously indecisive individuals (lowest values). Additionally, nuances were found for some, but not all, religiosity and secularization variables within the orthodox religious and the believing group respectively. Given this overlap, the last aim of this study was to identify any distinct clusters of religious self-assessment based on the simultaneous inclusion of sociodemographic and religious variables. An orthodox-religious cluster, encompassing slightly less than a third of the whole sample, confirmed the previous univariate analyses with regard to both the sociodemographic and the religious profile.
45A second cluster, the believing cluster, constituted more than half of the sample. Their popular religiosity and West-compatible version of Islam are high while their normative religiosity, fundamentalism, and all forms of religious practice are on a medium level. Moreover, the believing cluster is characterized by a medium level of secularization of belief and a high level of secularization of practice. It is possible that ‘believing’ is, especially in contrast to the negative context of the public discussion of deism and atheism (e.g., Vanlıoğlu 2022), a more diplomatic expression signaling the beginning of an individual secularization process and might resemble the Western category of ‘spiritual (but not religious)’ (see Gür, 2020; Zinnbauer et al. 1997). Our results show that a self-assessment of ‘believing’ is indeed religiously located between religious-orthodox and the more secularized categories of deism, non-believing/atheism, and religious indecisiveness. Nevertheless, the Turkish expression of ‘having a belief system’ (inançlı) can be applied to different religious attitudes – such as being a believing Muslim with a laicist and anti-Islamist standpoint14 (see Hecker 2022), holding certain values as sacred, believing in natural religion, and many more. As a result, qualitative studies are urgently needed in order to start uncovering what lies behind this self-assessment and the potential subgroups within this category.
46This should be also seen in the context of the third cluster, which consists of all deists, non-believing/atheists, and religiously indecisive but also some ‘believing’ individuals. As all forms of religiosity are significantly lower in this cluster and both forms of secularization at their highest, this cluster can be considered secularized. Interestingly, this last cluster does not differ from the ‘believing’-cluster (Cluster 2) in its sociodemographic profile. Future studies could focus on determining the differences between believing and more secularized individuals not only from a sociodemographic lens, but also from the perspective of personality differences (e.g., believing individuals could be more agreeable, less open-minded, or more concerned about self-enhancement) and socialization agents (family, peers, school) as some studies have already started to investigate in relation to the single categories of deism (e.g., Ermiş 2019; Koç 2022) and atheism (e.g. Sevinç, Mehmedoğlu 2016). In addition, longitudinal or trend studies could further clarify whether the believing self-assessment and all the more secularized categories are indicative of a generational (Gezici Research Institute 2021; Konda 2018, 2022) or life cycle effect (Dikbaş 2021; Ermiş 2019; Koç 2022).
47This study presented a view on the contemporary variety of religious self-assessments occurring among Turkish Muslims. Such a variety of individual religiosities seems to exist despite political efforts to enforce certain forms of Islam and religiosity. Additionally, it was also found that the vast majority, i.e., all but the orthodox religious group, display some kind of secularization during their lifetime, either in the belief or in the practice dimension. These results can have crucial implications for religious educators and clergy in contemporary Turkey since they should accept and even welcome such changes of the understanding of the own religiosity, especially among the younger generations. By updating their vision, mission, and/or curriculum such religious authorities can take this situation as an opportunity to encourage religious growth and maturation based on a personal reflected faith and independent of political agendas (e.g., Völker 2018).
48Future studies should include more variables that potentially differentiate between the religious groups (e.g., personality, socialization, attitudes towards the Holy Quran or religious authorities) and should also use qualitative methods in order to shed more light on what people mean when they define themselves as ‘having a belief system’, as ‘deistic’, or as ‘religiously indecisive’. Given more subcategories could be detected through deeper investigations using these methods, an even greater variety of religiosities could be uncovered (e.g., agnostic, i.e., clear confession of unknowingness about the existence of God) – this, again, could challenge the prevailing public view of Turkish Muslims as thoroughly pious and/or Islamist (see İsmail 2004; Kehl-Bodrogi 2003).
49The data is available at
https://osf.io/pnm74/?view_only=478d0c24337d45a3af811f4df81037eb
Table S1: Representative distribution of the sociodemographic variables (middle column) in comparison to our quota sample.
Table S2: Cross table between age group and religious self-assessment.
50Note. Valid n is displayed in the cells.