Abstract
Apparent conflicts between religion and science are often observed in the United States. One consequence of such conflicts might be that religious individuals will be less likely to recommend their children pursue a career in science, technology, engineering, and mathematics (STEM). We examine this possibility using a nationally representative survey focused on a variety of issues related to religion and science. We find that, compared to religiously unaffiliated individuals, evangelical Protestants, mainline Protestants, Catholics, and Jews are less likely to say that they would recommend a child enter the pure STEM careers of physicist, engineer, or biologist. These differences are weaker or nonexistent for the more applied STEM careers of physician and high school chemistry teacher. The religious tradition effects observed for the pure STEM careers are primarily mediated by lower levels of interest in science and higher levels of creationist views among those groups relative to the religiously unaffiliated.
Keywords
A variety of academic, governmental, educational, and business actors have called for the United States to increase the size, quality, and diversity of its science, technology, engineering, and mathematics (STEM) workforce (e.g., U.S. Department of Labor, 2007). Teachers and counselors attempting to get students into STEM classrooms and from there into STEM careers face a number of potential challenges, especially when attempting to increase the representation of women and racial minorities. Research has shown that women, for example, tend to express less interest and confidence in math even when their achievement is equivalent to a male peer (Riegle-Crumb, King, Grodsky, & Muller, 2012).
Although it does not receive as much attention as gender and race, religion could also pose a challenge in getting students to choose science careers. Religion has long been seen as a potential obstacle to the education of students in science classrooms, especially in regard to teaching evolution (Dotger, Dotger, & Tillotson, 2010). These religion-related obstacles could translate into the underrepresentation of religious individuals in science careers. Ecklund’s (2010) study of scientists in elite universities found that 54% of scientists did not identify with a religious tradition compared to 16% of the general population. Based on these percentages, it seems that religious students are not pursuing science careers at the same rate as less religious students.
In the research presented here, we ask two specific questions. First, is religious affiliation, religious identity, and/or religious behavior associated with an individual’s likelihood of saying she would recommend a child enter a STEM career? Second, if such measures of religion are associated with the willingness to recommend that a child enter an STEM career, then what appears to mediate this relationship? Regarding the first question, we hypothesize that individuals belonging to conservative or evangelical Protestant traditions, those who attend religious services more frequently, and those who identify as very religious will be less likely to recommend that a child enter an STEM career compared to those who are religiously unaffiliated, those who do not attend religious services frequently, and those who do not identify as religious. Regarding the second question, we hypothesize that these effects are mediated by these individuals’ lower level of interest in science, a greater likelihood of perceiving scientists as hostile to religion, and a greater likelihood of adhering to creationist views on the origin of life and the universe.
Parents, Religion, and Career Recommendations
A large body of research has examined parental influence on children’s career aspirations and development. This research has identified a number of mechanisms, ranging from the passive (genetics) to the active (direct recommendation or disapproval of career) to those somewhere in the middle (socioeconomic resources, general educational encouragement) (Hughes & Thomas, 2003; Splete & Freeman-George, 1985). Parents can represent a type of support and a type of barrier for a child’s career choice both over the long-term (distal) and in short-term (proximal) (Lent, Brown, & Hackett, 1994, 2000). As a distal factor, parents can expose (or not expose) a child to particular career role models and experiences that shape the child’s interests and goals (Ferry, Fouad, & Smith, 2000). Parents can also represent a more acute influence on a child’s career choice. For instance, a parent could refuse to financially support a child in college if he or she would choose a particular major (Lent et al., 2000).
Some research has suggested that parents might be particularly influential in shaping a child entering a science career (Archer et al., 2012; Dabney, Chakraverty, & Tai, 2013; Ferry et al., 2000). Dick and Rallis (1991) found that high school students who said they were going to enter careers in science or engineering were more likely to cite the influence of parents than students planning on entering nonscience careers. Similarly, research has found that higher levels of parental involvement in a child’s high school education had the impact of increasing the number of math and science courses the child planned to take in high school (Ma, 2001; Maple & Stage, 1991).
How might the religion of parents influence the career development of children? There are several mediating mechanisms that could link the former to the latter. Some research has shown that, at least for some religious groups, parents could indirectly shape the career choices for their children by restricting educational attainment (Darnell & Sherkat, 1997; see also Beyerlein, 2004). Such broad educational effects, however, do not suggest anything about how a parent’s religion might specifically be associated with whether a child enters an STEM career. There are potential mediators, though, that could lead to such a specific outcome.
Interests and values
Some research has found that religion could influence attitudes about careers by shaping an individual’s values and orientation toward work. For instance, Keister (2003) argues that instead of “having an orientation toward the afterlife … Jewish families encourage this-worldly pursuits … such as high income careers and investing” (p. 179). To the extent that a religious tradition values or devalues certain types of work, or the outcomes of that work (e.g., money and prestige), and to the extent that this overlaps with scientific careers, the parent might be more or less open to her child entering an STEM career.
Scheitle (2011) examined how undergraduates’ religiosity varied across areas of study. He found that students majoring in the areas of business, education, or professional studies (e.g., medicine, health technology, nursing, and pharmacy) scored higher on a scale of religious commitment compared to the overall student population. At least in the case of education and health, this corresponds to research showing that religious students are attracted to “helping professions” out of a sense of religious duty or calling (Hernandez, Foley, & Beitin, 2011). A parent’s religious tradition or religiosity, then, could lead him or her to encourage implicitly or explicitly a child pursue careers seen as more religiously relevant or valuable (e.g., helping others), as these are the careers that the parent himself or herself sees as more relevant and valuable.
Scientifically problematic beliefs
A long history of social thought has assumed that religious belief and scientific pursuits are fundamentally at odds. Indeed, scientific advancement has been a cornerstone of theorizing about how individuals and societies become less religious (Swatos & Christiano, 1999). The argument is that specific scientific findings and the general mode of scientific thought reduces the plausibility of religion’s claims (Berger, 1967). We set that debate largely aside. Whether or not religious belief and science are inherently at odds is outside the purview of social science; what we do know is that many religious individuals perceive them to be in conflict. If a parent’s religion leads them to hold beliefs that they see as directly threatened by scientific findings, the parent could steer their children away from STEM. Although a variety of religious beliefs might run into conflict with scientific claims, one of the most common areas of tension surrounds the origins of life and the universe (Scott, 2009).
View of scientists
It is also possible that a parent may not worry that science will challenge specific religious beliefs held by the child but instead worry that the scientific profession as a whole is hostile to religion and, under some circumstances, even immoral (Evans, 2013). In recent years, scientists such as Richard Dawkins (2006) and Sam Harris (2005, 2006) have taken very vocal stances against religion, often framing their arguments in scientific language. If a parent feels like such voices are representative of how their faith will be viewed by a child’s science teachers or colleagues, then they may try to guide them away from exposure to scientists.
Figure 1, then, summarizes the potential conceptual links between an individual’s religious adherence and the likelihood that he or she would say that he or she would recommend a child enter a science career. These mediators have implications for both distal and proximal influences in a child’s career choice. For instance, if a parent thinks that science threatens particular religious beliefs (e.g., concerning creation), then he or she may not take the child to science museums, which could affect the interests and goals of the child. Or, the parent could present a more immediate or proximal barrier by refusing to pay for a child to pursue a science major in college if he or she thinks that scientists are hostile to religion (Lent et al., 2000).
Theoretical model relating an individual’s religion to her likelihood of recommending a child enter a science, technology, engineering, and mathematics (STEM) career.
Data and Method
The data used for this analysis come from the Religious Understandings of Science (RUS) study. The survey portion of the RUS was conducted by the survey firm GfK using its KnowledgePanel. The KnowledgePanel is an online research panel that is representative of the U.S. population. Panel members are recruited using a statistically valid sampling method with a published sampling frame of residential addresses that covers approximately 97% of the U.S. households, reflecting the U.S. Census. When non-Internet households are recruited they are provided a netbook computer and free Internet service so they may also participate as online panel members. The KnowledgePanel consists of about 50,000 adult members (ages 18 and older) and includes persons living in cell phone–only households. Panel members receive incentive “points” that can be redeemed for cash. The KnowledgePanel has been utilized in numerous research articles published in top social science journals (e.g., Pedulla & Thebaud, 2015). Chang and Krosnick (2009) found that probability-based online panels provide the representativeness of random-digit dialing while reducing measurement error, survey satisficing, and social desirability response bias.
For our survey, GfK randomly selected 15,868 panel members from the full KnowledgePanel and an oversample of 878 individuals who had indicated upon entering the panel that they worked in a science-related occupation. This equates to a total of 16,746 individuals invited to complete the survey. Of those invited, 9,900 of the randomly selected panel members completed the survey along with 341 individuals identified in the oversample of those in science occupations. This equates to 10,241 total responses. The survey was fielded in December 2013 and January 2014. Because the individuals selected from the oversample of science-employed panelists are likely different than the overall KnowledgePanel and therefore different from the general U.S. population, when presenting descriptive statistics (Table 1), we utilize a poststratification weight that adjusts for this oversample and nonresponse patterns based on population benchmarks from the October 2012 Current Population Survey. In other words, when applied the data weights ensure that the point estimates from the data match the target population (i.e., the U.S. adult population) and are not biased by the oversampling and/or the lower response rates of some groups. In the analyses below, we exclude individuals who provided a missing value on any of the measures used here. This produces a final analytic sample of 9,368 individuals.
Descriptive Statistics (N = 9368).
Note. All religious tradition measures coded 0 = not specified tradition; 1 = specified tradition; religious service attendance coded 1 = never to 9 = several times a week; religious person is coded 1 = not at all religious to 4 = very religious; science-related occupation coded 0 = no, 1 = yes; political conservatism coded 1 = extremely liberal to 7 = extremely conservative; education is coded 1 = less than high school to 4 = bachelor’s degree or higher; income is coded 1 = less than $5000 to 19 = $175,000 or more; number of kids is coded 0 = 0 children to 4 = 4 or more children; age is coded 1 = 18–24 to 7 = 75 or older; female is coded 0 = male, 1 = female; race measures are coded 0 = not specified race; 1 = specified race; interest in science is coded 1 = not at all interested to 4 = very interested; scientists are hostile to religion is coded 1 = strongly disagree to 5 = strongly agree; creationist views are coded 1 = definitely false to 4 = definitely true. SE = standard error.
Measures
Outcomes
We focus on a series of questions from the survey in which respondents were provided a randomly ordered list of 13 occupations and asked, “How much would you recommend to a child of yours to enter the following occupations as an adult?” Possible responses were 1 = would not recommend at all; 2 = would recommend somewhat; and 3 = would recommend strongly. The 13 occupations were engineer; physicist; biologist; plumber; anthropologist; TV weather forecaster; high school chemistry teacher; psychologist; physician; pastor, minister, or clergyperson; electrician; sociologist; and nurse.
We focus on two groups of careers from this list. First, we examine what we label the “pure STEM” careers of engineer, physicist, and biologist. Second, we consider what we label the “applied STEM” careers of physician and high school chemistry teacher. We include this second category for two reasons. First, it will provide the ability to examine some potential differences across types of STEM careers. Second, as we noted above, the so-called helping professions are sometimes pointed to as being more attractive to religious individuals or families. Being a science teacher or a medical practitioner could be seen as a more acceptable type of STEM career for some religious individuals. Including these in our analysis will provide some ability to assess this. We recognize that nursing could also be considered an applied STEM career, but we do not include it here simply because we were concerned that responses could be conflated too much with perceived gender of the career. Similarly, one could potentially view an electrician as a type of applied STEM career, but we felt that this might conflate social class or professional status with our interests. Ideally, we would also have an occupation that was clearly not STEM related but was also not problematic from a gender or social class perspective. Unfortunately, we do not have such an ideal comparison occupation in our data.
Looking at the pure STEM careers, 41% of respondents “would recommend strongly” that a child become a biologist or a physicist, while 56% would recommend strongly that a child become an engineer. It is not clear why engineering would be more appealing to respondents. It is possible that individuals have a better image of what an engineer does, or it is possible that they perceive an engineer as less hostile to religion, more applied, more prestigious, higher paying, and/or as having better job prospects. Examining the applied STEM careers, 56% of respondents said they would recommend a child become a physician, while 31% of respondents said they would recommend strongly that a child become a high school chemistry teacher. Again, these differences could be due to perceptions of earning potential and/or prestige.
Religion measures
We include several measures of religious identity and behavior in our analysis. The first is a series of indicators representing the religious tradition of the respondent. These indicators are constructed from a series of questions that began with a broad religious affiliation question (Religiously, do you consider yourself to be Protestant, Catholic, Jewish, Mormon, Muslim, not religious, or something else?) and then follow-up questions to specify particular denominations or traditions. The constructed indicator categories are: evangelical Protestant; black Protestant; mainline Protestant; Catholic; Jewish; Mormon; Hindu, Muslim, or other non-Western; other religion; and no religion. In creating the three Protestant categories, we relied on common classification standards (Steensland et al., 2000).
We also include a religiosity measure that asked, “To what extent do you consider yourself a religious person?” Possible responses were 1 = not religious at all; 2 = slightly religious; 3 = moderately religious; and 4 = very religious. Finally, we include a measure for frequency of religious service attendance. This question asked, “How often do you attend religious services?” Nine responses were offered ranging from 1 = never to 9 = several times a week.
Mediators
We include several measures assessing respondents’ interest in science and their views on science and scientists. The first measure comes from a series of questions asking, “Please tell me how interested you are in the following things.” We focus on the item asking about interest in “New scientific discoveries.” Possible responses were 1 = not at all interested; 2 = moderately interested; and 3 = very interested. We expect that a higher level of personal interest in science will increase a parent’s desire that a child will become a scientist.
Also included in the analysis is a measure assessing the respondents’ level of agreement with the following statement: “Most scientists are hostile to religion.” Possible responses were 1 = strongly disagree; 2 = disagree; 3 = neither agree nor disagree; 4 = agree; and 5 = strongly agree. We expect that this measure will have a negative impact on a parent’s likelihood of recommending their child become a scientist.
Finally, we include a measure of the respondent’s view of creation and evolution. This measure comes from a series of questions that provided respondents six different narratives of the origin of the universe and the development of life on earth. These narratives ranged from a young-earth creationism narrative to a purely natural evolution narratives. For each narrative, respondents were asked whether they thought it was definitely false, probably false, probably true, or definitely true. Respondents could also say that they were not at all sure. For the analysis below, we recoded this last response to be the midpoint between the true or false statements. We focus on respondents’ level of adherence to the most restrictive or problematic narrative from a scientific perspective: “God created the universe, the Earth, and all of life within the past 10,000 years.”
Other controls
Our analysis takes into account a variety of other control measures. These include the respondents’ highest educational degree and household income, since we know that there are significant differences across religious groups on these variables (Smith & Faris, 2005). The former has four categories: 1 = less than high school; 2 = high school; 3 = some college; and 4 = bachelor’s degree or higher. The income measure has 19 categories ranging from 1 = less than $5000 to 19 = $175,000 or more. Because research has shown that views of science are increasingly politicized (Gauchat, 2012) and because religious views and affiliation correlate with political views, we include a measure of political ideology. This measure ranges from 1 to 7, with 1 representing extremely liberal and 7 representing extremely conservative. We refer to this as a measure of political conservatism. We also include measures for the respondent’s age, gender, and race, as all three factors have been shown to be associated with religiosity (Argue, Johnson, & White, 1999; Miller & Hoffmann, 1995; Yancey, 2005). Age is measured with seven categories ranging from 1 = 18–24 to 7 = 75 or older. Race is measured according to four indicators: White, non-Hispanic; Black, non-Hispanic; other, non-Hispanic; and Hispanic.
Finally, we include a measure representing the number of children the respondent has ever had. This is measured on a 5-point scale ranging from 0 to 4 or more. We include individuals who have not had children because the issue of whether they would recommend a child enter a particular career is still relevant as many of these individuals may eventually have children. As with most survey research, we are ultimately assessing attitudes about recommending a child enter a career. We are not able to measure actual recommendations or behaviors regarding these attitudes. Even for those individuals who have had children, the question is often hypothetical; their children may be adults and beyond the point of being open or able to take a career recommendation (note: we do not have data on the age of survey respondent’s children, so we cannot limit to individuals with children of particular ages). Descriptive statistics for all of our measures are shown in Table 1.
Results
We utilize bootstrapped generalized structural equation models to assess the mediation model presented in Figure 1. Bootstrapping the estimates allows more flexibility for distributions to violate assumptions of normality, which indirect effects often do due to their multiplicative nature (Preacher & Hayes, 2004). In determining the significance of effects we examine the 95% confidence intervals produced by the bootstrapping procedure. We began by estimating models in which the religious measures and controls were modeled as having direct effects on the outcomes and the mediators were not included in the models. These initial models allow us to examine whether there is any evidence of a direct effect to be mediated. We then estimated models in which the religion measures and controls were modeled as having direct effects and indirect effects through the mediators. These models allow us to assess whether a direct effect remains after the mediators are included and whether there are significant indirect effects through the mediators. Analyses were conducted in Stata/SE 13.1. We used the “collin” command in Stata to assess potential multicollinearity among our independent variables. The largest variance inflation factor was 3.31, which is well below the levels at which multicollinearity becomes a concern (Regression with Stata, 2016).
Table 2 shows the pre- and postmediation direct effects of our predictors. The first column for each outcome shows the premediation effects. We focus our discussion on the religion measures since they represent the theoretical focus of this research. Looking at the three pure STEM outcomes we find that, compared to the religiously unaffiliated, evangelical and mainline Protestants, Catholics, Jews, and those of another religion are significantly less likely to recommend a child become a physicist, biologist, or an engineer. The Jewish effect is somewhat surprising, given other research showing that Jewish families tend to encourage “this-worldly pursuits” (Keister, 2003, 2008). It is important to emphasize, though, that these effects are after controlling for differences in political ideology, education, income, and the other factors included in the model. If we exclude the controls we find that the significant Jewish effect only comes out when controlling for these other factors. That is, on the surface Jews do not seem like they are less likely to recommend a child enter pure STEM careers, but once we take into account their education and income, which both have positive effects on recommending, then Jews have a lower likelihood of saying they would recommend a child enter these careers. This is not the case for the other religious tradition effects, which are significant before entering any other measures into the model. Frequency of religious service attendance also has a negative association with recommending a child become a physicist or biologist but does not have a significant association with recommending a child become an engineer. Interestingly, self-identification as a religious person has a positive association with recommending all three of the pure STEM careers.
Direct Effects From Bootstrapped Generalized Structural Equation Models of How Much Respondent Would Recommend Their Child Enter Careers (N = 9368).
Note. All religious tradition measures coded 0 = not specified tradition, 1 = specified tradition; religious service attendance coded 1 = never to 9 = several times a week; religious person is coded 1 = not at all religious to 4 = very religious; science-related occupation coded 0 = no, 1 = yes; political conservatism coded 1 = extremely liberal to 7 = extremely conservative; education is coded 1 = less than high school to 4 = bachelor’s degree or higher; income is coded 1 = less than $5,000 to 19 = $175,000 or more; number of kids is coded 0 = 0 children to 4 = 4 or more children; age is coded 1 = 18–24 to 7 = 75 or older; female is coded 0 = male; 1 = female; race measures are coded 0 = not specified race; 1 = specified race; interest in science is coded 1 = not at all interested to 4 = very interested; scientists hostile to religion is coded 1 = strongly disagree to 5 = strongly agree; creationist views is coded 1 = definitely false to 4 = definitely true. STEM = science, technology, engineering, and mathematics.
*Significant based on bias-corrected 95% confidence intervals.
Turning to the two applied STEM careers, we find much weaker evidence for any initial direct associations with religion, and we even find some positive associations. Evangelical Protestants, Catholics, and Jews, do not significantly differ from the religiously unaffiliated on these outcomes. Black Protestants are significantly more likely than the unaffiliated to say they would recommend a child become a physician. Mainline Protestants, however, are less likely to say they would recommend a child become a physician. It is somewhat surprising that mainline Protestants are less likely to say they would recommend a child become a physician, while evangelical Protestants show no difference from the unaffiliated. It is possible that evangelical Protestants’ theological desire to see a child help others and alleviate suffering (e.g., healing as a physician) leads them to see becoming a physician as attractive, outweighing their overall conservatism relative to mainline Protestants. There are almost no religious tradition differences in willingness to recommend a child become a high school chemistry teacher. The only such difference is between Jews and the religiously unaffiliated, with the former showing less willingness to recommend such a career path. Religious service attendance has no significant association with the physician outcome and has a positive association with the high school chemistry teacher outcome. The identification as a religious person measure is positively associated with both of the applied STEM outcomes, which is the same pattern we saw with the pure STEM outcomes. It is possible that individuals identifying as a religious person are simply more open to or willing to guide career choices in general compared to those who do not identify as religious.
To summarize, we find much stronger evidence for negative religious direct effects for the pure STEM fields compared to what we find for the applied STEM fields. To what extent are these effects mediated in the manner outlined in Figure 1? The second column for each outcome in Table 2 shows the postmediation direct effects of our measures. Looking at the pure STEM fields, we find that many of the religion associations become insignificant or at least weaker once we include the mediators. For instance, evangelical Protestants and Catholics no longer significantly differ from the religiously unaffiliated in their willingness to recommend a child become a physicist, biologist, or engineer. Mainline Protestants no longer differ from the unaffiliated in their willingness to recommend a child become a biologist or engineer. The previously significant negative association between religious service attendance and willingness to recommend a child become a physicist or biologist also becomes insignificant in the mediation model.
In examining the effects of our mediators, we find that the respondent’s interest in science is positively associated with his or her reported likelihood of recommending a child pursue all of the pure and applied STEM careers. Interestingly, creationist views have significant negative associations with the likelihood of recommending the three pure STEM careers, but no significant association with the applied STEM fields. This suggests that individuals who hold such beliefs do not view those beliefs as problematic for more applied STEM careers as they do for the pure STEM careers. Viewing scientists as hostile to religion has a negative association with the physicist, biologist, and the high school chemistry teacher outcomes. This is less of a pure STEM versus applied STEM pattern that we see with the creationist views mediator, but it might have something to do with the perceived work context or culture of these careers. That is, it could be that respondents perceive a higher likelihood that their child could be interacting more closely with scientists who are hostile to religion as a physicist, biologist, or high school chemistry teacher, while they see this as less of an issue with a child who becomes an engineer or physician.
Table 3 shows the indirect effects of the religion measures through each of the three mediators and their total indirect effects. We only show the three pure STEM outcomes in this table since the evidence of mediation was much stronger than for the applied STEM outcomes, and we do not show the indirect effects of our control measures since they are not our primary theoretical interest here. We see that many of the differences between the religiously unaffiliated and the religious traditions are filtered through respondents’ interest in science and creationist views. Viewing scientists as hostile to religion has a weaker role as a mediator of the religious tradition differences, although it does seem to play a role in evangelical Protestants’ lower likelihood of recommending a child become a physicist or biologist. It also mediates some of the frequency of religious service attendance association with the outcomes. Interestingly, identifying as a religious person has a positive indirect association through interest in science, but this is weakened somewhat by its negative indirect association through creationist views and, for the physicist and biologist outcomes, viewing scientists as hostile to religion.
Indirect Effects From Bootstrapped Generalized Structural Equation Models of How Much Respondent Would Recommend Their Child Enter Careers (N = 9368).
Note. All religious tradition measures coded 0 = not specified tradition, 1 = specified tradition; religious service attendance coded 1 = never to 9 = several times a week; religious person is coded 1 = not at all religious to 4 = very religious; interest in science is coded 1 = not at all interested to 4 = very interested; scientists are hostile to religion is coded 1 = strongly disagree to 5 = strongly agree; creationist views is coded 1 = definitely false to 4 = definitely true; indirect effects of controls not shown.
*Significant based on bias-corrected 95% confidence intervals.
Discussion
Broadly, the analysis showed that compared to the religiously unaffiliated, Catholics, evangelical Protestants, mainline Protestants, Jews, and those of “other” religions are more hesitant to say they would recommend their children enter pure STEM careers, such as becoming a physicist, engineer, or biologist. Religious service attendance is also negatively associated with the reported likelihood of recommending a child become a physicist or biologist. Generally, this hesitancy is not seen with the more applied STEM careers, such as becoming a physician or a high school science teacher. These religion effects on recommending pure STEM fields are explained by lower levels of interest in science among those religious traditions and higher levels of creationist views in comparison to the religiously unaffiliated. Viewing scientists as hostile to religion also plays a role by mediating some of the religious tradition differences, particularly for evangelical Protestants, and by mediating the influence of religious service attendance.
What do these findings mean for policy makers, educators, and career counselors? Our analysis showed that a major factor explaining why individuals in some religious traditions would be less likely to recommend a child enter a pure STEM careers is simply the lower level of interest in science within these religious traditions. The question then becomes how to spur more interest in science within these religious traditions. Part of the answer might simply be more outreach on the part of scientists to religious congregations and communities. The framing of that outreach could be important, though. Some prior research has shown that students and parents within certain religious traditions tend to be attracted to professions that are often perceived to be more “helping” or “public service” in nature, such as teaching and medicine (Hernandez, Foley, & Beiten, 2011; Perry, 1997; Scheitle, 2011). Indeed, our analysis showed weaker or nonexistent religious tradition differences in willingness to recommend applied STEM careers, but stronger and clearer negative relationships for pure STEM careers. It is possible that individuals of these religious traditions do not perceive teaching science or practicing medicine as fundamentally about science as they are about helping others. Trying to emphasize the potential helping aspects of pure STEM careers could be one way to address the religious tradition differences that are unique to them. Labels could also be important in this regard. For instance, labeling a potential career as “vaccine developer” instead of “biologist” might be perceived as more appealing by individuals in these religious traditions.
Although negative views of scientists do not seem to play as large of a role in mediating the influence of religion on recommending a child enter STEM careers, our analysis did indicate that such views are not entirely irrelevant. Viewing scientists as hostile to religion did significantly mediate the negative religious service attendance and evangelical Protestant effect on the physicist and biologist outcomes. Given this, counselors might consider pointing to examples of accomplished religious scientists. In her study religion among academic scientists, Ecklund (2010) described religious scientists who openly discuss connections between their faith and science as “boundary pioneers.” In her interviews, she found that many religious and nonreligious individuals pointed to Francis Collins, the geneticist and director of the National Institutes of Health, who has frequently written and spoken about his Christianity (e.g., 2006), as a well-known boundary pioneer. Collins is not the only example of such a boundary pioneer, either historical or contemporary. Such boundary pioneers could help alleviate the perception that all scientists are hostile to religion and increase the willingness to recommend a child enter that career.
The research presented above does have limitations, of course. As we noted earlier, the nature of our data does not allow us to actually assess individuals’ behaviors. We cannot be sure that respondents reporting that they would or would not recommend a child enter an STEM career would follow-through with such a stance. Similarly, our data are cross-sectional, so we are inherently limited in assessing the causal nature of the relationships we highlight above. Ideally, we would have panel data that would allow us to see how changes in religion are associated with changes in the willingness to recommend a child enter an STEM career. Such an analysis could be a direction for future research. It would also be beneficial if qualitative research examined more in-depth parental attitudes toward children entering STEM careers and the ways that religion appears to shape such attitudes. Despite the limitations just noted, this research provides unique evidence using nationally representative survey data of how religion could shape parental recommendations for children considering STEM occupations.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research for this article is part of the Religious Understandings of Science Study, funded by the John Templeton Foundation (grant 38817, Elaine Howard Ecklund, principal investigator).