In this paper four studies investigating Christian, Muslim, and atheist job applicants’ perceived competence and likeability were reported. Significant differences in likeability ratings were found in several studies, where atheists and Christians preferred the target belonging to their ingroup. Significant differences between groups in competence ratings or hiring decisions were rare. In the first pilot study, no statistically significant findings were observed. However, the sample consisted of only 60 participants, which likely was too few for possible differences to be detected. In the three subsequent studies with sufficient sample sizes, participants rated applicants who mentioned religion (or atheism) as being less likable than applicants for which religion was not mentioned. These results were found both in the USA and in Sweden. Although a significant difference was not reached in the Swedish sample, applicants who mentioned religion were rated as less competent in both studies with American participants. In study 2, conducted in a US sample, Christian participants rated the Christian target as being more likable than the atheist target and also rated the Christian target as more likeable than atheist participants did. These results were replicated in study 4, which also found that atheist participants rated Christian targets as being less likeable than both atheist targets and Muslim targets. Atheist targets were also rated as more likeable by atheist participants than by Christian participants. In study 3, with a Swedish sample, atheist participants rated atheist targets as more likable than both of the religious targets. They also hired applicants who did not mention religion (76.3%) more often than the Christian applicants (23.7%). However, Christian participants did not rate atheist targets as less likeable than Christian targets.
Mentions of religiosity or lack thereof led to lower likeability ratings in both Swedish and US samples, as well as lower competence ratings in the US samples. This might be explained by the information in the control conditions, where applicants reported being part of a philosophical discussion group or an interest organisation, being perceived as more favourable than being part of a religious/atheistic discussion group or organisation. It might also be explained by mentions of religious (dis)belief being more likely to be perceived as negative in a recruitment context by people in general.
When a difference in ratings between targets was found, it was consistently people preferring their religious ingroup over a religious outgroup. Significant results were however only found for likeability ratings and not in competence ratings, i.e., people did not view religious ingroup members as more competent, but merely as more likeable. The exception was in study 3, where Muslim targets were deemed less competent than Christian targets by Christian participants. In study 4, the differences in likeability ratings did not affect hiring decisions. However, in study 3 atheist participants, who rated atheist targets as more likeable but not more competent than Christian targets, hired Christians (23.7%) less often than applicants who did not mention religion (76.3%). Similar differences in hiring decision were not found for the other targets. Thus, in some cases likeability might affect hiring decisions, even when there is no significant difference in perceived competence.
Christian Swedes rated Muslim targets as less competent than Christian targets. This result is surprising, since it was the only significant difference in competence ratings in all included studies and since it was not coupled with significantly lower likeability ratings or lower willingness to hire Muslims. One possibility is that Christians in Sweden do not dislike Muslims more than they dislike atheists or other Christians but still view Muslims as less competent. Another possibility is that Christian Swedes are more comfortable with rating Muslims as less competent than to rate them as less likeable. Swedish Christians might consider it a moral obligation to love and value all people equally, despite religious differences. However, their religion does not encourage them to view everyone as equally competent. It is also possible that increased power would be needed in order to determine if they also view Muslim targets as less likeable.
The only significant difference in hiring decisions was that Swedish atheists hired Christian applicants less often than they hired control applicants. Surprisingly, no such effect was found in the USA. The Swedish atheists hired Christian targets in 23.7% of cases, while American atheists hired Christian targets in 28.0% of cases. It is possible that no significant association was found in the latter case due to the lower number of atheist participants in study 4. However, it is also possible that Swedish atheists are less willing to hire Christian applicants than American atheists are. In any case, the result from study 3 indicates that Christian applicants are less likely to be hired by atheist Swedes, which could imply that this group has a negative enough view of Christians to warrant avoiding hiring them.
When likeability ratings and hiring decisions differed between targets, religious ingroup members were preferred. This is in line with ample research showing that ingroup members are generally perceived to have more positive qualities. In study 3, Christians rated Christian targets as more competent than Muslim targets. Apart from this finding, an ingroup preference was not seen in any competence ratings, indicating that religious ingroup members are generally not viewed as more competent than outgroup members. It is possible that significant, albeit small, differences in competence ratings between targets could have been detected with larger samples.
In study 3, only atheist participants were found to rate their religious ingroup higher on likeability. This might be explained by the fact that they constituted the majority of the sample (110 atheists and 87 nones) and thus generated enough statistical power to produce a significant difference, while the 54 Christian participants were too few for an ingroup likeability preference to be detected. Speaking against this interpretation is the fact that the difference between Christians’ likeability ratings of atheist targets and Christian targets was very small. Another possibility is that the stigma against atheists, which has been demonstrated in the USA and supported by the results from study 2 and 4, is not that pervasive in Sweden. Being nonreligious is common and perhaps therefore not as penalised by Christians in Sweden.
Atheists and nones in study 4 rated Christian targets as less likeable than both the other targets, which means that they liked atheist applicants more than Christian applicants despite living in a country where numerous studies have found that the population prefer religious people over nonreligious people. Thus, this study indicates that ingroup favouritism influences likeability judgements made by atheists and nones in the USA more than the general stigma against atheists in the country. Alternatively, outgroup derogation directed at Christians might affect these ratings more than ingroup favouritism.
Previous studies have found little or no gender bias in recruitment contexts when structured interviews or structured employment references are used [32, 33, 37]. The lack of variation in competence ratings and hiring decisions across targets in the studies presented here might be due to the structured information that participants received. The order and amount of information about experience, education, personality, skills, and references was similar in all cases. If participants had evaluated a less structured motivation letter, the results might have been different. The results at least indicate that mentioning religious affiliation in a recruitment context might make an applicant seem both less likeable and less competent. Moreover, the differences in results between likeability and competence ratings indicate that religious ingroup favouritism or outgroup derogation seem to mainly affect likeability ratings and more seldom competence ratings.
The religious affiliation of targets was not directly stated, but rather implied by their involvement with an atheist, a Christian, or a Muslim group. This was necessary to avoid suspicion by participants, but means that some participants might have interpreted the information in an unintended way. In addition, participants might think that mentions of religiosity are improper in a recruitment context, which might be an explanation for the finding that applications that did not mention religion were rated higher. Another consequence of this design is that participants might perceive people who are actively involved in a discussion group or an organisation with a (non)religious focus as more extreme than people who simply identify with a religion or as atheists.
Even though Muslims are a minority group in both Sweden and the USA, with a culture that is relatively distant from both atheists and Christians, Muslims were rated as likeable and competent as atheists’ and Christians’ ingroups in most studies. The two exceptions were found in Study 3, where atheists rated Muslims as less likeable than atheist targets and Christians rated Muslims as less competent than Christian targets. This could mean that Muslims are well-liked in both countries, but a more probable interpretation would be social desirability bias. Participants might want to appear to be unprejudiced and thus rate Muslim targets higher than they otherwise would have done. The religious information when the applicant was a Muslim might have become more salient for participants than when the applicant was an atheist or a Christian, since being Muslim is more uncommon in the two countries where the studies were done.
There are a few discrepancies from the preregistration for study 2. The initial plan was to conduct another study in Sweden, identical to study 2. For practical reasons, we could not conduct two studies of that sample size in Sweden and opted for two new studies with a changed design (study 3 and 4) instead. Thus, the hypothesis and intended analyses that we preregistered could not be investigated. We also opted to use ANOVAs with difference scores for our analyses instead of repeated measures, as we had preregistered in study 2.
Since participants were aware that the applicants were fictitious, they might not respond as they would if they were asked about real people and real positions. A study where participants were asked about ostensibly real cases, similar to Moss-Racusin et al. , would avoid this limitation.
A limited number of jobs was used in the studies. Participants rated applicants’ likeability and competence in relation to the positions they applied for, meaning that religious affiliation might affect perceived competence or likeability differently had other types of jobs been used. The intention was to use positions which would require a high degree of moral responsibility—taking care of a defenceless person (personal care aide), teaching children (teacher), and making decisions affecting people’s financial situation (administrative director). Other jobs might be more heavily dependent on qualities such as intelligence, industriousness, or cooperativeness.
The USA is a large country with religiosity levels and other cultural aspects differing between states. Since participants anywhere in the USA could participate through Prolific, the results might not be generalisable to all parts of the country.
Future studies could investigate jobs that require specific qualities, such as morality, industriousness, intelligence, and cooperativeness. Such studies could examine if people perceive atheists, Christians, or Muslims to have some of these qualities to a greater extent, if people perceive their religious ingroup members to possess these qualities to a greater degree, and if such perceptions differ between cultures, e.g., depending on the majority religion or religious history. In the present studies, jobs varying in responsibility and status were used. Future studies could investigate whether jobs of low or high status lead to different competence and likeability ratings for atheists, Christians, and Muslims, depending on the religious affiliation of participants. Another approach would be to ask participants to imagine themselves as being a customer or equivalent tasked with rating an employee, where similar information is available and religious belief is varied between conditions. This would make it possible to examine participants’ perceptions when they have more of a stake in the situation.
In conclusion, the present studies suggest that in some cases people in Sweden, as well as people in the USA, rate religious ingroup members as more likeable, but not more competent, than religious outgroup members.