Research on nonreligious populations and experience has proliferated in the past decade, documenting both a growing population of nonreligious people and social marginalization these people face as a result of the privileged position of religion in contemporary American society (Cimino & Smith 2014; Cragun, Kosmin, et al. 2012; Edgell et al. 2006; Hammer et al. 2012; Hammer et al. 2013; Kosmin et al. 2009; LeDrew 2013; Smith 2010; Smith 2013a; Wallace et al. 2014; Zuckerman 2009). However, many components of nonreligious experience remain underexplored. Given the overwhelming focus on organized atheist groups and statistical portraits of nonreligious existence, scholars have yet to examine variations within nonreligious communities or the ways others interpret such communities (Smith 2013b; though see Dunn & Creek 2015). Consequently, little is known about the variety of ways people interpret contemporary nonreligion and what such variations may reveal about religious privilege in American society. Further, it is unclear whether there is a singular or multiple (i.e., varied in terms of race, class, gender, and sexualities) conceptualization of the nonreligious in society (Edgell et al. 2006; Dunn & Creek 2015). In addition, little is known about what intersectional frameworks (see Collins 2005; Crenshaw 1996; but see Grollman 2012 for quantitative examples) might reveal about interpretations of the nonreligious due to nonreligious and religious scholars’ almost exclusive focus on nonreligious populations as a whole (Smith 2013b). Such research, especially of the quantitative variety, typically relies on single-identification measures (i.e., what do respondents think of the nonreligious, see Edgell et al. 2006) rather than complex portraits of nonreligious people (i.e., what might respondents think of transgender nonreligious people in comparison to cisgender nonreligious people).

The present article uses data from the first survey instrument to examine people’s attitudes toward intersectional (i.e., varied in social demographics) subjects to investigate variation in conceptualizations of religious and nonreligious people in relation to different locations within other systems of oppression and privilege. Specifically, this article investigates three research questions. First, when people visualize religious and nonreligious people, do they perceive them differently as a result of sexual and gender self-identifications? If so, what forms of gender and sexual diversity are most important for being recognized as religious or nonreligious and as favorable or unfavorable in society? Finally, to what extent do people experience significantly different reactions or worlds as a result of their locations within interlocking religious, sexual, and gender hierarchies?

As we have demonstrated elsewhere (see, e.g., Sumerau & Cragun 2015; Sumerau et al. 2016a), however, there is currently no way to explore sexual and gender intersections in mainstream (or nationally representative) quantitative data sets because such samples only measure cisgender options and almost never include sexual minority components (see also Ikanovich et al. 2013; Westbrook & Saperstein 2015). As a result, we utilize a convenience sample in combination with a survey of our own design for the purposes of exploring such phenomena and illustrating intersectional nuances that may be missed by traditional quantitative data sets. Rather than generalizing these findings to a given broader population, our analysis directs attention to the need for theorizing and systematic data collection beyond cisgender and heterosexual populations and issues, and to the utility of exercising caution when generalizing findings from cisgender and heterosexual based surveys to the broader social world (see also Nowakowski et al. 2016a; Westbrook and Saperstein 2015).


To understand and explore variation in conceptualizations of religious and nonreligious people, we drew on intersectionality theories (Crenshaw 1996). As a theoretical framework, intersectionality calls into consideration the ways people experience and interpret themselves and others as always part of multiple social groups at once (see, e.g., Collins 2005; Grollman 2012; Ward 2008). Thus, no social aspect, experience, or characteristic may be fully understood without attending to the ways it is shaped and influenced by multiple, interlocking systems of oppression and privilege (McQueeney 2009). A nonreligous person, for example, does not merely experience the world as nonreligious, but rather as a nonreligious person situated within specific historical and contemporary race, class, gender, sex, and sexual hierarchies that shape zir experience of nonreligion. At the same time, one does not encounter this subject as simply a person in the concrete world, but rather as a raced, classed, sexed, gendered, sexualized, and/or “religioused” person one must interpret via their own internalized assumptions concerning any or all of these categories individually and collectively (see also Collins 1990). As a result, researchers must recognize that people experience their own lives and interpret the existence of others within and between intersecting systems of thinking, expectation, feeling and knowing created and sustained by the efforts of other social beings occupying various locations within such systems (Foucault 1980). Of particular concern to scholars studying intersectionality are the ways the combination of multiple social locations influence the experience of any given identity claim or community in relation to the broader social world (see, e.g., Grollman 2012). For example, investigations of nonreligious experience focused on an aggregated sample of nonreligious people may mask the ways other demographic locations influence the experience of nonreligion, nonreligious organizations, and religious influence (Dunn & Creek 2015).

As evidenced in recent scholarship, past research into religious oppression and privilege is often limited by scholars’ tendency to aggregate diverse groups into simple measurements (Cragun & Sumerau 2015) and only focus on singular axes of social inequality (Barton 2012; McQueeney 2009; Wilkins 2008). Whereas this recognition has facilitated the emergence of more intersectional approaches to religion (Avishai et al. 2015) and religious inequalities in society (Sumerau et al. 2015) at times (though such analyses are still rare, see Avishai et al. 2015), this development has yet to find expression in studies of nonreligion (Smith 2013b). Variations in the ways people interpret religious and nonreligious people remain unexamined while singular measures of nonreligion or studies of homogenous nonreligious groups take center stage. In addition, as Dunn and Creek (2015) noted, the almost exclusive focus on organized nonreligious movements and beliefs leaves potential variation within nonreligious populations and experience without much mention. Without examining how people make sense of and experience nonreligion in relation to other systems of inequality, however, efforts to combat the subordination of nonreligious people and communities will – like similar efforts to combat racism, sexism, heterosexism, and cissexism without recognition of the ways each of these systems influences the experience of the others – ultimately be limited by their implicit assumption of a unified social location and experience (see Collins 2005).

Given the almost exclusive focus on homogenized nonreligious experiences, many questions remain about variations in the ways people interpret religious and nonreligious others. First, in what ways do sexual and gender identifications influence social evaluation of nonreligious and religious identification? Second, what social factors – religion, sexualities, or gender for example – play more powerful roles in the interpretation of nonreligious and religious people? Finally, in what ways might examining such variation benefit sociological knowledge concerning religion and nonreligion?

The Varieties of Religious and Nonreligious Social Value

While researchers have begun to map the placement of nonreligious people in the landscape of contemporary American society, our utilization of an intersectional framework led us to remain sensitive to variations in the ways people interpret religious and nonreligious others. As such, we followed Dunn and Creek’s (2015) assertion that understanding attitudes toward religious and nonreligious others requires making sense of the social locations of religious and nonreligious people in relation to intersecting systems of oppression and privilege embedded within the larger social world. To this end, we compared attitudes toward people occupying differential religious, sexual, and gender positions within society as well as the overall similarity or difference such variations created between groups. In so doing, we extend previous findings concerning attitudes toward religious and nonreligious people by revealing variations in social acceptance of religion and nonreligion predicated upon sexual and gender identification.

Previous research has suggested that sexual identity – or identification – may be an integral factor in conceptualizations of religious and/or moral standing in society (McQueeney 2009; Wolkomir 2006). Whereas many studies find consistent negative – though softening while remaining negative in recent years (see Cragun et al. 2015) – portrayals of sexual minorities by mainstream religions (Barton 2012; Worthen 2013), researchers have also noted consistent and significant populations of actively religious sexual minorities (see, e.g., McQueeney 2009; Sumerau 2012; Wolkomir 2006). We examined how people conceptualized nonreligious and religious people of varying traditions in relation to their identification as heterosexual, gay/lesbian, or bisexual in search of potential variations in the ways people make sense of religion and nonreligion. In so doing, our analysis demonstrates that people make sense of religious and nonreligious others in varied ways depending on the sexual identification of the subject.

Although rare, past research has also demonstrated significant differences between transgender and cisgender experiences of religion (see Sumerau et al. 2016; Rodriguez and Follins 2012; Sumerau & Cragun 2015). While nonreligious studies have yet to explore such distinctions, one would hypothesize similar findings since nonreligious people often draw their own beliefs and values from science (Smith 2010), which – like its religious counterparts – has historically erased transgender existence from its constructions of the world (see also Nowakowski et al. 2016; Stryker 2008). To this end, we examined the ways people’s perceptions of nonreligious and religious people shifted as a result of cisgender or transgender identification. As a result, we demonstrate significant differences in the perception of people when faced with the possibility of transgender religious and nonreligious people instead of cisgender subjects.

Another source of variation may arise in relation to religious identification itself. Whereas studies have shown significant negative attitudes related to atheism and nonreligion more generally (Cragun, Kosmin, et al. 2012; Edgell et al. 2006; Wallace et al. 2014), studies have rarely compared reactions to varied religious and nonreligious selves simultaneously. Examining the contemporary religious landscape, however, researchers have often noted variation in the ways people value different religious traditions (Cragun, Henry, et al. 2012; Bolce & De Maio 1999; Edgell et al. 2006) as well as differential reactions to atheism versus other nonreligious identifications (Cragun, Kosmin, et al. 2012; Edgell et al. 2016; Gervais & Norenzayan 2012; Sumerau & Cragun 2016; Swan & Heesacker 2012). We sought to outline the ways people interpret differently gendered and sexual beings in relation to varied religious and nonreligious identifications. In so doing, our findings revealed complicated and nuanced distinctions in the ways people conceptualized various religious and nonreligious others.

While we do not in any way mean to suggest that the variations we focus on here (i.e., religion, nonreligion, gender and sexualities) are by any means exhaustive, our analyses begin the process of systematically subjecting religious and nonreligious variation to empirical scrutiny. We suggest our utilization of measurements capturing intersectional subjects, and our demonstration of the variations such measurements produce could provide guidance for scholars seeking further development of intersectional approaches to religion and nonreligion in contemporary society. Further, our incorporation of non-cisgender and non-heterosexual options via the use of a convenience sample allows for theoretical development currently unavailable through the use of most traditional quantitative data sets (Nowakowski et al. 2016a), and sheds light on nuances and variations often hidden in the design of existing social scientific survey projects focused on religion and/or nonreligion (see also Westbrook and Saperstein 2015).


Data for this study come from an online survey. After IRB approval, students in two of the first author’s sociology courses were tasked with recruiting friends and family members to participate in the survey who met two criteria: they had to be over the age of 18 and could not be college students. These criteria were used as many studies rely on the responses of college students, who exhibit limited variation in their characteristics and life experiences and are not representative of the adult population in the US (Henrich et al. 2010). While we make no claims about the representativeness of the sample we used, we do believe that it is more reflective of the adult population in the US than is a sample of college students. Students were required as part of a course project to recruit at least 10 participants, but were given extra credit if they recruited more than 10.

Students provided those they recruited with the URL of the survey and the student’s ID for the project to track the number of participants each student recruited. Students were specifically instructed not to observe participants complete the survey in order to protect participants’ anonymity. A total of 1,584 individuals began the survey. However, close to one third did not complete the survey. Additionally, many of those who completed the survey did not answer at least two thirds of the questions or failed to correctly answer two questions that were included in the survey as attention checks. After data cleaning, 618 cases were retained, though participants were not required to answer all questions, so the number of responses on any given variable may vary slightly.


As scholars have noted (Westbrook & Saperstein 2015), traditional quantitative data sets drastically limit the complexity and variation within existing social categories. Seeking to overcome this limitation, we designed a survey specifically targeting variation within and between categories typically measured as simplified and discrete representations. Because the focus of the survey was on attitudes toward intersecting identities that included minority genders and sexualities, we provided survey participants with definitions and explanations of sexual and gender identities in order to insure that responses were consistent across participants. These definitions are included in Table 1.

Table 1

Definitions of Sexual and Gender Identities for Survey Participants.

Biological Sex:
     male an individual who has been assigned – typically based on genitalia – to the sex category that is generally responsible for producing sperm; most human males have penises and testes
     female an individual who has been assigned – typically based on genitalia – to the sex category that is generally responsible for producing ova (i.e., eggs); most human females have a vagina, ovaries, and a uterus
     intersex individuals whose sex assignment is ambiguous, typically based on genitalia that are ambiguous (i.e., their genitalia do not clearly reflect genitalia that are usually assigned to male or female categories)
     gender the range of characteristics relating to and differentiating between masculinity and femininity
     masculinity the set of qualities or characteristics that are considered appropriate for boys or men
     femininity the set of qualities or characteristics that are considered appropriate for girls or women
Sexual Orientations:
     heterosexual a characteristic of someone who is attracted primarily or exclusively to an individual of a different biological sex or gender (e.g., a male attracted to a female)
     homosexual a characteristic of someone attracted primarily or exclusively to an individual of a same biological sex or gender (e.g., a female attracted to a female)
     bisexual a characteristic of someone attracted to individuals of different and same biological sexes or genders (e.g., a female attracted to both males and females)
Gender Identities:
     transgender a characteristic of someone whose gender identity does not match their assigned biological sex (e.g., someone who was assigned to be male at birth but who feels feminine)
     cisgender a characteristic of someone whose gender identity does match their assigned biological sex (e.g., someone who was assigned to be female at birth and feels feminine)
     transman an individual whose biological sex was assigned as female but who feels and identifies as male
     transwoman an individual whose biological sex was assigned as male but who feels and identifies as female
     cisman an individual whose biological sex was assigned as male who feels and identifies as male
     ciswoman an individual whose biological sex was assigned as female who feels and identifies as female

Immediately following the definitions, participants were asked the following questions about their sex, gender, and sexual identity. First they were asked, “What is your biological sex?” Response options included: male, female, intersex, and other. Respondents were then asked, “What is your gender?” Response options included: cisgender, transgender, and other. Finally, participants were asked, “What is your sexual orientation?” Response options included: heterosexual, homosexual, bisexual, and other.

Given the focus on religion, we asked four questions to capture participants’ religiosity. We asked participants their religious affiliation and provided twelve response options, plus an “other” option. We asked participants their view of God or a higher power. Response options included: “I do not believe there is a god” (atheist), “I do not know if there is a god and I do not believe there is a way to find out if there is a god” (agnostic), “I believe in some form of higher power” (deist), “I believe in god sometimes,” “I believe in god, but I doubt my belief sometimes,” “I am confident god exists” (theist), and “other.” We asked participants their views toward the Bible. Response options included: “The Bible is the actual word of God and is to be taken literally, word for word,” “The Bible is the inspired word of God but not everything in it should be taken literally, word for word,” “The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men,” “The Bible is not part of my religion,” and “other.” Finally, we asked participants about their religious service attendance. Response options ranged from “never” to “several times a week,” with a total of 9 options.

We also asked participants a number of demographic questions, including the year they were born (recoded into age), state or country of residence (recoded into regions), marital/relationship status, race/ethnicity, educational attainment, income, and political views.

Intersectionality Variables

Given our aim in the survey of better understanding how attitudes vary toward intersecting identities, we asked participants to evaluate twenty different identities that intersected religious affiliation with sexual orientation and gender on a 100 point thermometer scale. Specifically, the question asked participants, “On a scale of 0 to 100, where 0 indicates you would feel really uncomfortable being around people with these characteristics and 100 indicates you would feel really comfortable being around people with these characteristics, how comfortable or uncomfortable would you feel being around people with the following characteristics.” We included five (non)religious identities: Christian, Jewish, Muslim, nonreligious, and atheist. These were intersected with three sexual identifications and one gender: heterosexual, homosexual, bisexual, and transgender. This resulted in 20 intersections, such as: Homosexual Atheist Individuals and Transgender Muslim Individuals (the complete list is shown in Figure 1).

Figure 1 

Intersectional identity mean thermometer scores, ranked (all participants).


Our sampling methodology resulted in a fairly diverse sample (see Tables 2 and 3). The sample was disproportionately female (62.8%), and included several individuals who reported their biological sex as intersex (.3%) or other (.2%). There was little variation in gender, with just 1.0% of respondents reporting their gender as transgender (.2%) or other (.8%). We had slightly more variation in sexual identity, though most respondents were heterosexual (92%). Smaller percentages identified as homosexual (3.9%), bisexual (3.1%), and other (1.0%). The mean age for the sample was 39.31 (sd = 16.015), with a range from 18 to 86. The majority of our respondents were from Southern (29.2%) and Northeastern states (43.8%), which reflects the social networks of the students at the university, most of whom come from those regions. However, we have participants from 38 states, 2 US territories, and a small portion from outside the US (8.8%). Nearly one third of participants were single (32.8%) and just over one third were married (38.7%). Our sample was predominantly non-Hispanic White (68%), but included notable percentages of non-Hispanic blacks (6%) and Hispanics (both white and black; 15.0%).

Table 2

Demographic characteristics of survey participants.

N = 618 % (mean; sd)

age (39.31; 16.015)
US region or international*
     South 29.2
     Northeast 43.8
     Midwest 6.8
     Pacific and Mountain 5.3
     US territories 6.2
     international 8.8
marital/relationship status
     single, never married 32.8
     married 38.7
     divorced 10.2
     widowed 1.1
     separated 1.1
     dating exclusively 8.7
     cohabiting 6.5
     other 0.8
     White, non-Hispanic 68.0
     Black, non-Hispanic 6.0
     Hispanic White 14.2
     Hispanic Black 0.8
     Asian 1.5
     Other 3.6
     missing 6.0
     less than high school 0.8
     high school diploma 11.1
     some college 21.0
     Associate’s degree 10.1
     Bachelor’s degree 34.1
     Master’s degree 14.7
     PhD 1.0
     professional degree (MD/JD) 7.2
     under $10,000 4.2
     $10,000–$24,999 6.5
     $25,000–$49,999 15.5
     $50,000–$74,999 18.3
     $75,000–$99,999 17.5
     $100,000–$199,999 21.8
     $200,000–$499,999 12.1
     over $500,000 4.0
political views
     very conservative 2.6
     conservative 16.9
     moderate, but lean conservative 11.8
     moderate 29.6
     moderate, but lean liberal 12.1
     liberal 21.7
     very liberal 5.3

*We did not have survey respondents from every state, but did have respondents from 38 states and 2 US territories. The regions shown in this table included respondents from each of the following states: South = Alabama, Florida, Georgia, Kentucky, Louisiana, North Carolina, South Carolina, Tennessee, Texas; Northeast = Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, Virginia, West Virginia, District of Columbia; Midwest = Illinois, Kansas, Michigan, Minnesota, Missouri, Nebraska, Ohio, Wisconsin; Pacific and Mountain = California, Colorado, Idaho, New Mexico, Oregon, Utah, and Washington; Territories = Puerto Rico and Virgin Islands.

Table 3

Sex, gender, sexual orientation, and religiosity characteristics of survey participants.

N = 618

biological sex
     male 36.7
     female 62.8
     intersex 0.3
     other 0.2
     cisgender 99.0
     transgender 0.2
     other 0.8
sexual identity/orientation
     heterosexual 92.0
     homosexual 3.9
     bisexual 3.1
     other 1.0
religious affiliation
     Catholic 35.6
     nonreligious (includes atheists and agnostics) 22.3
     Southern Baptist 3.1
     Non-denominational Christian 14.5
     Jewish 6.9
     Methodist 3.3
     Presbyterian 2.4
     Lutheran 2.9
     Muslim 1.1
     Hindu 0.3
     Buddhist 0.5
     Protestant (otherwise unspecified) 2.8
     other 4.2
view of god/higher power
     I do not believe there is a god 8.9
     I do not know if there is a god and I do not believe there is a way to find out if there is a god 10.5
     I believe in some form of higher power 16.7
     I believe in god sometimes 4
     I believe in god, but I doubt my belief sometimes 12.9
     I am confident god exists 44.8
     other/choose not to respond 2.1
view of Bible
     The Bible is the actual word of God and is to be taken literally, word for word. 12.7
     The Bible is the inspired word of God but not everything in it should be taken literally, word for word. 51.9
     The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men. 26.6
     other 1.6
     The Bible is not part of my religion. 7.2
religious service attendance
     never 16.9
     less than once a year 13.8
     about once or twice a year 15.3
     several times a year 13.7
     about once a month 7.7
     2–3 times a month 7.2
     nearly every week 7.9
     every week 14.3
     several times a week 3.1

Participants were generally well educated, with 34.1% having Bachelor’s degrees, and more than 1 in 5 (22.9%) having advanced degrees. Participants were also fairly wealthy, with 37.9% making at least $100,000 per year. Politically, our participants ranged widely across a liberal to conservative spectrum, with close to 1/3 leaning conservative, another third leaning liberal, and the last third identifying as political moderates.

Religiously, our sample was disproportionately Catholic relative to the US adult population, with 35.6% identifying as such (around 24% do nationally). Just over 1 in 5 identified as nonreligious, which is proportional to the US adult population. Protestants were under-represented in the sample, but our sample did include some participants from a number of minority religious groups, including Hindus, Muslims, and Buddhists. The other measures of religiosity suggest our sample is slightly less religious than the US adult population more generally, with just 44.8% reporting they were confident god exists; 8.9% indicated no belief in god and 10.5% reported an agnostic position on god’s existence. Likewise, a fundamentalist/literalistic view of the Bible was under-represented in our sample, with just 12.7% reporting that they believed the Bible was the literal word of god. However, religious service attendance was not much lower in our sample than in the US adult population generally, with 32.5% reporting attending religious services two to three times a month or more.

Testing our hypotheses regarding how sexual, gender, and religious identities intersect was quite straightforward. Figure 1 shows the mean scores for each of the intersecting identities ranked from most to least highly rated for all participants in the survey. As Figure 1 shows, attitudes toward heterosexuals, regardless of their (non)religious identity, are the most favorable, while attitudes toward transgender individuals are the least favorable.

However, there are some important shifts that occur as attitudes are intersected. At the top of Figure 1 the intersecting identities follow what might be expected based on representative surveys of attitudes toward religious groups: Christians are the most highly rated, followed in order, by Jewish, nonreligious, Atheist, and Muslim individuals. This ordering suggests that most surveys that ask about Americans’ attitudes toward religious affiliations are really capturing attitudes towards these categories of people if it is assumed that those people are heterosexual and cisgender, which is not, of course, explicitly mentioned in the questions asked (see Sumerau et al. 2016).

Intersecting religious identity with sexual and gender identity, however, immediately complicates people’s views. Moving down Figure 1 to the next most favorable group, homosexuals, results in a slightly different picture. Now, nonreligious homosexual individuals are rated more favorably than are Christian or Jewish homosexual individuals, though Atheist and Muslim homosexual individuals are still rated less favorably. This suggests a pattern that is most apparent among transgender individuals, but can already be seen taking effect among homosexual and bisexual target groups. Homosexual and bisexual Christians and Jews are rated less favorably than are nonreligious homosexual and bisexual individuals. This suggests two things. First, it is more acceptable for nonreligious people, who are still considered “deviant” minorities in American culture, to have additional deviant identities, like being homosexual, bisexual, or transgender. The inverse is also true; in religious traditions where heterosexuality is normative, having a non-normative sexual or gender identity in effect “taints” (Goffman 1986) the religious identity. As a result, the intersected identities rated the least favorably are those that result in the greatest violations of normativity: transgender Muslims, Christians, and Jewish individuals.

Figure 2 replicates the analysis, but only for religious participants (nonreligious participants’ thermometer scores are excluded from Figure 2). This changes the results substantially, though one of the findings from Figure 1 remains. The first notable change is that there are much bigger gaps between sexual/gender categories than in Figure 1. There is also no overlap between the sexual/gender categories, as there was in Figure 1. However, the same shifted pattern of attitudes with nonreligious transgender people being rated more favorably than transgender Jewish, Christian, and Muslim individuals remains.

Figure 2 

Intersectional identity mean thermometer scores, ranked (only religious participants; n = 476).

Figure 3 presents the same analysis, but it is limited just to the nonreligious. Figure 3 is substantially different from Figure 2. The first notable difference in scores is that, for the nonreligious, sexual and gender categories are much less important than are religious categories. For instance, heterosexual Christians and Muslims are viewed substantially less favorably than are homosexual atheists and nonreligious individuals. Certainly sexual and gender identities still matter, but religious affiliation is a more influential characteristic for the nonreligious than it is for the religious. However, this may be because nonreligious individuals simply have less prejudice against sexual and gender minorities, which can also be seen in comparing Figures 2 and 3. At a very basic level, the range of scores is interesting. In Figure 2, the lowest rated intersected identity, Transgender Muslim Individuals, has a mean score of 57.91, almost 20 points lower than the mean score for the same identity in Figure 3. In other words, the religious people in our sample held substantially more negative views toward some categories of people than did the nonreligious individuals.

Figure 3 

Intersectional identity mean thermometer scores, ranked (only nonreligious participants; n = 137).

One of the problems with trying to examine intersectionality quantitatively is that you end up with lots of variables or categories in variables. We thus sought to further examine whether or not it was possible to reduce the number of variables. While this may appear counter to intersectional premises on the surface, it allows us to – as Collins (1990; 2005) suggests – explore both the unique variations created by intersectional categories (as shown above), and the most salient dimensions facilitating unequal outcomes for people occupying varied social locations (see also Grollman (2012) on Multiple Disadvantage patterns revealed via prominent intersectional effects in relation to health). As such, we utilize factor analysis to visually demonstrate the most prominent intersections of power facilitating variation in respondents’ interpretations of religious and nonreligious people of different sexual and gender identifications (see also Browne & Misra 2003).

The use of factor analysis makes it easier to identify prominent intersections or categorical distinctions respondents make when considering variations in religious, gender, and sexual identifications. To this end, we took the twenty intersectional gender, sexuality and religious identifications, and factor analyzed them using principle components factor analysis with Varimax rotation and Kaiser normalization. Table 4 presents the results of these analyses for all participants in our sample.

Table 4

Exploratory Factor Analysis of Thermometer Scores Toward Intersecting Sexuality, Gender, and Religion Identities: All Participants.

1 2 3

Heterosexual Christian Individuals –0.005 0.026 0.856
Heterosexual Jewish Individuals 0.094 0.276 0.820
Heterosexual Nonreligious Individuals 0.192 0.759 0.330
Heterosexual Atheist Individuals 0.247 0.787 0.288
Heterosexual Muslim Individuals 0.332 0.252 0.680
Homosexual Atheist Individuals 0.589 0.741 0.070
Homosexual Nonreligious Individuals 0.598 0.736 0.069
Homosexual Christian Individuals 0.608 0.574 0.261
Homosexual Jewish Individuals 0.627 0.619 0.192
Homosexual Muslim Individuals 0.671 0.531 0.274
Bisexual Nonreligious Individuals 0.736 0.576 0.042
Bisexual Atheist Individuals 0.744 0.567 0.040
Bisexual Christian Individuals 0.756 0.441 0.200
Bisexual Jewish Individuals 0.769 0.479 0.156
Bisexual Muslim Individuals 0.796 0.386 0.220
Transgender Atheist Individuals 0.888 0.308 0.034
Transgender Nonreligious Individuals 0.890 0.315 0.017
Transgender Muslim Individuals 0.922 0.203 0.195
Transgender Christian Individuals 0.927 0.186 0.179
Transgender Jewish Individuals 0.927 0.224 0.120
Initial Eigenvalues 13.507 2.001 1.086
Variance Explained 67.536 10.005 5.429

Note: Boxes highlighted in gray are the variables that load on the corresponding factors.

Table 4 indicates that three underlying commonalities resulted from the factor analysis, though the first accounts for most of the variation, 67.54%. It is common practice in factor analysis to assume that only those variables with a factor loading above .29 (i.e., .30 or above) are part of a factor. Using that approach, sixteen of the twenty variables load on the first factor. However, we have included all of the factor loadings in Table 4 because they tell a very interesting story. The first variable in Table 4, Heterosexual Christian Individuals, has a factor loading (or eigenvalue) that is basically zero. In other words, it does not load on Factor 1 at all. Given what does load on Factor 1, an interpretation of the factor loading for Heterosexual Christian Individuals could be that Heterosexual Christian Individuals are basically seen in society as the gold standard for what is normal. The further you move away from “heterosexual” and “Christian,” the further you move away from the gold standard of “normal” and an eigenvalue of 0.00 on Factor 1. This begins, immediately, with the next variable, Heterosexual Jewish Individuals, who have an eigenvalue of .09. This is not typically a strong enough loading on a factor to consider that variable as part of the Factor, but because it is not zero, it does indicate this category is “different from normal.” Heterosexual Nonreligious Individuals have a factor loading of .192, which is even closer to inclusion in Factor 1, and further from “normal.” This, of course continues, all the way up to Transgender Jewish and Christian Individuals, with identical factor loadings of .927. According to our sample and given the intersected identities we presented, the most deviant identity statuses are actually Christians and Jews whose gender identities do not align with their socially assigned biological sexes. Factor 1 is basically a reflection of marginalization via distance from what Americans consider “normal” and could basically be called: “marginalized American intersected identities.”

Factors 2 and 3 are very similar, but they provide slightly different insights. If we use the .30 standard cut off to determine which variables load on Factor 2, it turns out to be everyone but heterosexual and transgender Muslims, Christians, and Jewish individuals. This factor seems to be a reflection of a preference for religions of the book, or rather opposition to the Nonreligious and Atheists, as the variables with the highest eigenvalues are Heterosexual Atheist and Nonreligious individuals. Since transgender religious individuals anchor Factor 1, their distance from “normal” is largely reflected in that Factor, not in Factor 2. Factor 2 could be labeled something like “residual religiously and sexually marginalized intersected identities.” Factor 3 is almost perfectly inverted from Factor 1; it is a reflection of what is “normal.” Nonreligious heterosexual individuals barely make the cut off of being considered “normal” in American society with a factor loading of .330; atheists, on the other hand, remain too deviant to join the ranks of the “normal,” with a factor loading of .288. Factor 3 could be called something like “normative American identities.”


Utilizing an intersectional framework, we set out to answer several questions previously unaddressed in literature concerning religion and nonreligion in American society. Rather than simply comparing aggregated religious or nonreligious variables (i.e., the only real option for analyses in traditional cisgender and heterosexual based quantitative data sets, Nowakowski et al. 2016), for example, we sought to examine variations within these terms in relation to gender and sexualities. As Figure 1 shows, heterosexuality and cisgender status ultimately trump religious distinctions. As Queer theorists have long argued (Butler 1999), these aspects of the self significantly influence the ways people make sense of and evaluate any other social category or identity. Regardless of whether participants were religious or nonreligious, heterosexual identities were always rated highest by respondents and transgender identities were always rated the lowest. Further, our findings suggest that previous studies showing variation in attitudes toward religion – a pattern replicated in our respondents’ evaluations in Figure 1 – may actually be only capturing religious hierarchy if or when heterosexuality and cisgender subjects are assumed from the outset. Our findings lend empirical weight to intersectional assertions of variation embedded within and connected to multiple social locations, and reveal the importance of ascertaining what types (i.e., in terms of sexual and gender identity) of religious and nonreligious people respondents are evaluating.

Our findings also demonstrate some ways sexualities and gender complicate previous assumptions and theoretical conceptualizations of religion and nonreligion (see also Cragun & Sumerau 2015). Whereas researchers and the public have long assumed and argued that spiritual components – such as belief in a higher power, conceptualization of supernatural forces, and faith in the unseen – drive people’s understanding of religious traditions, memberships, and structures, our analysis reveals that in general and for religious respondents specifically sexualities and gender actually matter far more than anything having specifically to do with the supernatural (see also Sumerau et al. 2016). Respondents were more likely to accept nonreligious sexual and gender minorities than people who were both religious and within these groups. In fact, these effects were more pronounced among religious respondents, and less important for nonreligious people who appeared to actually focus more on the religious aspects of the subjects.

Our findings suggest that for some religious Americans, their attitudes toward others have much less to do with religious membership and belonging than with who one sleeps and what gender identity one claims (Barton 2012; see also Sumerau & Cragun 2015; Wilcox 2009). In contrast, for the nonreligious, it may be that some of these individuals adopt religious movement strategies (Cimino & Smith 2014) because religion is the primary frame of reference they focus on for making sense of themselves and distinguishing themselves from others as it is the salient component of their identity that singles them out from other Americans (Smith 2010; Edgell et al. 2006). In other words, it may be the case that religion is more important to nonreligious identity formation and boundary maintenance in some cases whereas, among some of those who have a more “normative” religious identity, sexualities and gender are more salient for identity formation and boundary maintenance.

Moving beyond the specific elements of variation demonstrated in Figure 1, our findings also suggest an interesting intersectional dynamic operating between identities, social status, and attitudes concerning certain groups of people. If, for example, one includes a really marginalized identity – such as transgender – with an identity generally accepted if not necessarily the most privileged in society – such as Jewish – that rejects the really marginalized identity as part of its “social script” or “expected norm” (Goffman 1986), then that combination becomes the most marginalized or disliked subject. If we look at figure one again, for example, transgender Christians, Jews, and Muslims (i.e., subjects combining a well accepted religious identity with a heavily marginalized gender identity) are far more negatively evaluated than atheists and nonreligious individuals even if these individuals also carry the same marginalized gender status. How might we explain this type of variation?

One answer may lie in what could be termed spill over or a pollution effect wherein the “stigma” (Goffman 1986) attributed to the heavily marginalized position is interpreted as a taint or stain upon an otherwise “clean” or “pure” (Goffman 1986) social identity. For example, a respondent may believe Christians are pure or clean in moral terms while holding transphobic biases at the same time. When said respondent encounters a transgender Christian, however, they may reject the combination instead of finding room for transgender people in their existing worldview (see Sumerau et al. 2016 for examples of this in process in the everyday experiences of transgender Mormons). At the same time, however, this respondent may already see atheists as damaged or unclean (Edgell et al. 2006) so a transgender atheist (i.e., a combination of two categories already devalued by the respondent) fits their worldview and causes little trouble in terms of evaluation. In the latter case, there is no conflict between the clean identity and the unclean identity – the respondent simply interprets the combination as further evidence of their belief in atheist or transgender failings. In the former case, however, there is a conflict between the clean and unclean assumption, and the respondent thus seeks to erase the conflict by even further devaluing the possibility represented by the combination – the respondent emphatically rejects the possibility with the most possible negative evaluation (see also Sumerau et al. 2016). These possibilities suggest the importance of ascertaining how any social category intersects with established religious teachings, beliefs, and interpretations of the world.

Another answer may lie in the source material that provides the foundation for religious beliefs, identities and communities (see also Barton 2012; Sumerau & Cragun 2015). Considering that over 60% of our respondents relied heavily on either Jewish, Christian, or Islamic scriptures (i.e., believed these writings to be the literal or inspired word of the divine), it is noteworthy that transgender people – and to a lesser extent gay/lesbian and bisexual people – do not exist or only exist in a negative fashion in these compositions (see also Sumerau & Cragun 2015 for interpersonal examples of such tension in relation to transgender and intersex people; Barton 2012 for interpersonal examples of such tension in relation to lesbian, gay, and bisexual people). While many communities and writings within broader Jewish, Christian, and Islamic cultures, traditions, and ideologies note the existence of such people in various ways and even positively in some cases, scholarship consistently shows that commitment to such scriptures tends to predict less favorable views of such groups (see Barton 2012 for reviews). Part of this issue lies in the promotion of distinction (rather than integration) of sexes into oppositional categories (i.e., male and female only) in many such traditions (see also Sumerau & Cragun 2015). This may lead traditions to expend considerable energy erasing the existence of transgender – and other non-scripturally verified or approved – people (Barton 2012; Sumerau et al. 2016; Wilcox 2009).

Our respondents may be doing the same type of erasure in their evaluation of others to reinforce their imagined female-male worldview or “cisgender reality” (Sumerau et al. 2016). Whereas scholars have noted that sexual and gender minorities often leave religious traditions due to negative reactions (Wilcox 2009), the “background expectations” (Ridgeway 2011) or “imagined realities” (Sumerau et al. 2016) that facilitate these negative reactions might be found in the ongoing erasure of these people by members of such traditions and maintenance of cisgender, monosexual and heterosexual worldviews. As such, it may not simply be the behavior or existence of actual sexual and gender minorities that establishes such conflict, but rather, it may be that some religious people imagine sexual and gender minorities in ways that are incompatible with their constructed realities and seek to dismiss the possibility.

In fact, such a scenario would reflect ongoing patterns of erasure noted throughout American structural and interpersonal relations in the past century. Whether we look to history (Stryker 2008), scientific instruments (Nowakowski et al. 2016), workplace policies and procedures (Schilt 2010), media accounts (Schilt and Westbrook 2009), assumptions about families (Pfeffer 2010), or healthcare diagnostic and treatment protocols (Davis 2015), researchers have noted many ways contemporary American structural and interpersonal norms erase any possibility that does not neatly fit into heterosexual, monosexual and cisgender assumptions (see also Sumerau et al. 2016). Further, researchers have noted the ways such patterns facilitate the ongoing marginalization of lesbian, gay, bisexual (Cragun & Sumerau 2015), transgender (Miller & Grollman 2015), and intersex (Davis 2015) people. Our respondents may be responding to widespread societal erasure and marginalization of non-heterosexual, non-monosexual, and non-cisgender possibilities in their evaluations of religious and nonreligious others.

At the same time, it appears that being nonreligious lessens – to an extent – expectations of sexual conformity. While other studies have demonstrated that nonreligious communities echo religious groups in emphasizing conformity, such studies show they typically require members to conform to agreed upon notions of religion and science (Cimino & Smith 2014; Dunn & Creek 2015; Smith 2013b) rather than established sexual claims. If we turn to Figure 3, for example, we see less variation in relation to sexual characteristics, and even see two groups – bisexual nonreligious people rated more highly than some heterosexual and homosexual others and heterosexual Muslims rated lower than other heterosexuals – that disrupt the type of uniform pattern found in the other analyses.

While nonreligious respondents may be more open to some types of sexual diversity than their religious counterparts (Cragun 2013; Hunsberger 2006; Lefkowitz et al. 2004), we see a similar pattern of non-acceptance of transgender experience. While attitudes are not as negative as are those of our religious participants, they mirror their religious companions by demonstrating the most difficulty or negativity in relation to this group of people. While this may seem surprising as a result of their apparent greater recognition of sexual diversity, it likely owes to the foundational assumptions they share with religious people or the wider patterns of transgender marginalization noted above (see Miller & Grollman 2015). Whereas religious groups typically rely on sacred texts like the Bible, Torah, and Koran for guidance, nonreligious groups typically rely upon scientific arguments and assertions from the past to the present (see Smith 2010 for the importance of scientific belief among nonreligious groups). Whereas these claims differ in many ways (see Cragun 2015 for an elaboration of such distinctions between religion and science), one place they agree – whether one looks to evolutionary or Biblical elaborations of world history, whether one looks to social surveys or Koranic Suras, whether one looks to medical protocols or laws in the Torah – is that they typically erase transgender experience in order to create an imagined worldview wherein only static and distinct female and male beings exist and matter (see Butler 1999). While there have long been transgender (and other sex and gender variant) people and groups challenging these fictional worldviews in both religion and science (see Butler 1999; Stryker 2008), at present these oppositional systems share a cisgender foundation (Nowakowski et al. 2016; Sumerau et al. 2016) that facilitates the marginalization and negative evaluation of transgender people whether the subject or respondent is religious or nonreligious. Our findings lend further weight to the studies cited above by suggesting that a cisgender worldview that erases the existence of transgender people is pervasive beyond religion and continues to permeate nonreligious and scientific worldviews.

Our findings also complicate previous research exploring prejudice against the nonreligious. Our analyses of variation coupled with the results from the factor analysis reveal that when it comes to prejudice, gender and sexualities overshadow religion. While researchers utilizing traditional quantitative samples and methods that only allow the comparison of singular categories to other categories have suggested the nonreligious may be the most ostracized in contemporary American society and experience similar harassment and discrimination in relation to other marginalized groups (Edgell et al. 2006; Hammer et al. 2012; Wallace et al. 2014), our analysis utilizing a convenience sample to compare multiple social categories at the same time suggests that this assertion depends on which nonreligious people we are talking about. In no place do nonreligious people represent the most negatively evaluated subjects for our respondents, but rather gender and sexual identifications lead to the most marginalization and religious people may be more marginalized than nonreligious others when they are sexual and/or gender minorities. What previous findings may have actually noted is that nonreligious cisgender heterosexuals are the most marginalized within cisgender heterosexual populations, but confirmation of this – or potential acceptance of previous assertions – will require data sets containing sexual and gender representative samples and the ability to compare and contrast multiple rather than singular categories.

Our findings from the factor analysis clearly demonstrate the way these complexities may play out when quantitative data sets become more capable of offering representative analyses of society rather than of only cisgender and heterosexual assumptions (see also Sumerau et al. 2016). Those who deviate at all from normative gendered, religious, and sexual expectations may be automatically marginalized. Stated another way, those who do not conform to being a heterosexual cisgender religious person will face marginalization. This marginalization then varies in relation to which or how many of these expectations one violates specifically. Atheists and other nonreligious people are seen as marginal; gay, lesbian, and bisexual people are seen as marginal; and transgender people are seen as marginal. However, sexual and gender deviation are even more frowned upon than religious deviation. If one merely deviates in terms of religion (i.e., nonreligious cisgender heterosexuals) they will likely do better than someone who only deviates in terms of sexualities (i.e., gay/lesbian/bisexual cisgender and religious) or in terms of gender (i.e., transgender heterosexual and religious). As suggested by the factor analyses, there is likely a completely distinct world inhabited by, on the one hand, heterosexual cisgender Christians, and those who deviate in any way from these three axes on the other hand, but this other world of marginalization is varied and complex depending upon which and how many violations a given person can be considered to have committed from the heterosexual, cisgender, and religious norm (see also Sumerau et al. 2016).

This realization becomes even more intriguing when we recognize that in some cases religious deviation (i.e., becoming nonreligious) can be a positive thing in American society. In the case of sexual and gender minorities, being nonreligious actually may reduce the ramifications of sexual and gender nonconformity because such an endeavor erases conflicts between religious assumptions and marginalized selfhood. Stated another way, people may easily make sense of an agnostic, bisexual, genderqueer person by relying on the combination of normative assumptions suggesting (1) agnostics are incompatible with American norms (Edgell et al. 2006), (2) bisexuals are incompatible with religion (Barton 2012; though welcomed in some religious cases, see Harper 2010) and American mononormativity (see Moss 2012), and (3) gender variant people are incompatible with religion (see Sumerau et al. 2016) and American cisnormativity (Westbrook & Schilt 2014). The congruency of these assumptions in such a case may allow people to ignore or easily dismiss any claims by such people, but may also protect such people from explicit religious conflicts with others since they are – on average – not expected to be religious in the first place (see also Sumerau 2014 for LGBT Christians’ experiences constantly having to explain their existence to other Americans).

On the other hand, people may have a hard time making sense of, for example, a transgender Christian heterosexual because normative assumptions paint transgender people as incompatible with religion and heterosexuality in the popular imagination (see also Sumerau 2014). Likewise, a gay transgender Muslim may cause conflicts for people who assume Muslims are heterosexual or people who hold cisnormative assumptions that being transgender automatically requires transition that may change whether someone’s partner is of the same or a different sex as they are (see Sumerau & Cragun 2015). Unlike the former scenario, such people may face constant conflict (and thus elicit even more drastic reactions and/or evaluations) in relation to religion as they try to create a space for themselves within a structure and belief system that does not uniformly agree that they exist in the world (Wolkomir 2006) and often relies upon cisnormative, mononormative, and heterosexist foundations (Wilcox 2009). Further, they may, like the LGBT Christians Sumerau (2014) studied, have to constantly manage other people’s assumptions that they do not exist in this world in their daily lives even as more and more religious traditions at least mention the existence of sexual and gender minorities.

Considering that researchers have already noted the lack of diversity in many emerging nonreligious organizations, movements, and even scholarship (Smith 2013b) and that some religious traditions have long worked in opposition to sexual and gender nonconformity in American society (Wilcox 2009), the intersection of (non)religion, sexualities and gender may be a fascinating microcosm of existing tensions between what people – religious or not – believe and the worlds they imagine (Sumerau et al. 2016). In much the same way religious and nonreligious groups each became more comfortable with and more welcoming to lesbian and gay people in recent decades by revising their scriptural and scientific writings to make room for these realities (Butler 1999; Robinson & Spivey 2007; Warner 1999), one must wonder what reactions members of these groups will develop as transgender, bisexual, and other sexual and gender fluid (see Moss 2012) experience increasingly permeates the mainstream and challenges much of their previous findings and teachings. For nonreligious scholars (see Dunn & Creek 2015 for a similar observation in relation to race and nonreligion), this may be an interesting opportunity to see just how similar or different religious and nonreligious groups are when faced with challenges to their foundational teachings, beliefs, and assertions about what the world looks like and how it operates.

Despite what we believe are important findings capable of guiding attention and empirical analyses toward the diversity of religious and nonreligious interpretations and experience, our study has several limitations. First, our data derived from people recruited by college students at one university in the Southeastern United States, which means generalizability from our sample is limited and would be better accomplished via reproduction of the same analyses with various populations in different places. Additionally, given that research emerged during data collection revealing significant variations in black versus white nonreligious experience (Dunn & Creek 2015), it may be worthwhile to incorporate racial variations into analyses of this type in the future (i.e., rank white atheist heterosexuals, black atheist heterosexuals, Hispanic atheist heterosexuals, etc.). While our analysis begins unpacking variation hidden by limitations in more traditional data sets (see Westbrook & Saperstein (2015) for discussion of such limitations especially in relation to sex and gender), we emphasize the need for more systematic research into the variations contained within identifications like religious and nonreligious. Another limitation with our study is that we included “transgender” as a gender modifier attached to other identities, but did not include “cisgender” as a gender modifier. This was an intentional decision as we thought the implicit cisgendering of other identities (e.g., Heterosexual Christian without adding “cisgender”) might result in our participants evaluating those categories in similar ways to how they are evaluated in most other surveys that include no gender identities, in future research “cisgender” should be compared to “transgender” explicitly as an intersected identity category.

Our findings here also lend weight to emerging criticism and consideration of traditional quantitative approaches predicated upon cisgender, monosexual and heterosexual assumptions, measurements, and samples (see, e.g., Nowakowski et al. 2016; Miller & Grollman 2015; Westbrook & Saperstein 2015). Due to such limitations in available large scale data sets, we utilized a convenience sample to direct attention to nuances and variations often hidden in national or (cisgender, heterosexual) “representative” survey designs. Our analysis points to a couple of important considerations for quantitative analyses of religion and nonreligion. First, as a convenience sample, our findings shed theoretical light on potential nuances and variations, but should be approached with caution and further study of such possibilities rather than as a source for generalizations about society as whole. Second, as a survey instrument that empirically measures sexual, gender, and religious diversity, our findings reveal the importance of also approaching previous generalized or representative findings with caution as such findings do not typically include information on aspects of society that do not conform to cisgender, monosexual, and/or heterosexual assumptions (see also Nowakowski et al. 2016).


In sum, our analysis directs attention to some ways examinations of religion and nonreligion may benefit from intersectional frameworks. Considering that unequal systems are built upon and maintained via their connections to other systems of oppression and privilege (Collins 2005), fully understanding religion or nonreligion requires deconstructing the categories themselves as well as the ways other social assumptions and locations influence what these categories mean to people. To this end, our analysis revealed that respondents’ reactions to religious and nonreligious people may be heavily predicated upon the gender and sexual identification of the subject in question and may vary considerably. Transgender identification, in particular, appears to be especially marginalized in some religious and nonreligious imaginations of the world. The privileging of cisgender status – in combination with monosexuality, heterosexuality, and Christianity – may become an issue both religious and nonreligious communities will have to wrestle with in the coming years. Although it may be tempting – and comforting for religious and nonreligious people – to assume concrete distinctions between these two worldviews, our findings suggest that while they may differ in some important ways they may also share some foundational assumptions in relation to gender and sexuality at present. Disaggregating and making sense of religion and nonreligion as separate, different, shared, and/or similar systems of understanding the world may require systematically analyzing intersectional influences on the ways people interpret religious and nonreligious selves and others in relation to other systems of social inequality.