Abstract
“Faith is an evil precisely because it requires no justification and brooks no argument. Teaching children that unquestioned faith is a virtue primes them … to grow up into potentially lethal weapons for future jihads or crusades” (Dawkins, 2006, p. 305).
In Ginges, Hansen and Norenzayan (2009; GHN), we presented studies investigating the relationship between religion and popular support for suicide attacks. We identified two possible reasons why religious groups carrying out suicide attacks might find it easier to obtain popular support from communities they are seeking to represent. The first we termed the religious belief hypothesis. This hypothesis, neatly encapsulated by the above quote from Richard Dawkins, is that devotion to religious belief itself encourages suicide attacks, because, for example, religious belief might lead to hatred of non-believers (Dawkins, 2003). The second hypothesis we termed the coalitional-commitment hypothesis. This hypothesis focused on the role religion plays in binding people together into cooperative groups of non-kin (Graham and Haidt, 2010; see also Norenzayan and Shariff, 2008). In particular, we built on theoretical and empirical work suggesting that participation in collective religious ritual encourages costly commitment to coalitional identities (Atran, 2003; Irons, 2001; Sosis and Ruffle, 2003). We proposed that because suicide attacks may be thought of as an extreme form of parochial altruism (where the attacker gives his or her life to the group while killing others), frequent attendance in collective religious ritual might facilitate positive attitude towards parochial altruism in general and, in relevant contexts, suicide attacks in particular.
In four studies carried out with different religious groups across diverse political contexts, using survey and experimental methodology, we tested these two hypotheses by investigating the extent to which frequency of prayer (as an index of religious devotion) and frequency of attendance at mosque, synagogue temple or church predicted support for suicide attacks (in Studies 1, 2 and 3) and parochial altruism more generally (in Study 4). While prayer frequency proved a reliable index of devotion to religious belief, it did not independently predict either support for suicide attacks (in Studies 1 and 2) or endorsement of parochial altruism (in Study 4). In contrast, attendance in places of collective religious ritual strongly predicted support for suicide attacks or parochial altruism in all studies whether this independent variable was measured or manipulated.
Since frequency of attendance at sites of collective religious ritual always strongly predicted support for suicide attacks but frequency of prayer never did, we concluded that our results strongly supported the coalitional commitment hypothesis, but failed to support the religious belief hypothesis. We also argued that this more broadly implies that any relationship between religion and suicide attacks may be independent of devotion to specific religious creeds and instead is a function of the way religions help to bind people together into communities of parochial altruists.
Liddle, Machluf and Shackelford (LMS; this issue) comment on our findings, suggesting that there may be more to the religious belief hypothesis than we allow in GHN. LMS do not dispute our results, but they worry that the strength of our claims may deter other researchers from investigating the hypothesis further, and argue for different methods of testing the hypothesis. In particular, they suggest that an important test of the religious belief hypothesis would be an investigation of the relationship between endorsement of specific religious beliefs (e.g., the afterlife) and support for suicide attacks. Below we respond to their comments and offer suggestions of our own.
As LMS note, we did not investigate the relationship between specific religious beliefs such as belief in the afterlife, and support for suicide attacks. Our reasoning is as follows: devotion to a religion typically involves a general endorsement of the core beliefs and values of that religion, such as monotheism or the existence of an afterlife. In this way, measures of devotion to Islam should account for the influence of belief in the afterlife on support for suicide attacks. That said, we encourage research on this topic and for those who wish to conduct further empirical tests of the relationship between religious belief and support for suicide attacks, we would like to offer some advice.
First, people may manufacture post hoc rationales for their support for violent attacks on others just as they often manufacture rationales for moral decisions (Haidt and Bjorklund, 2008), or other behaviors (Nisbett and Wilson, 1977). Insurgent political violence has been conducted in the name of any number of beliefs, including religion, human rights, freedom, and preferred forms of economic organization (Ginges, 1997). That these various rationales are employed does not tell us whether the beliefs themselves—the tenets of religion, democracy, or freedom—actually cause people to go to war, or whether they are merely epiphenomenal post hoc rationales to justify intergroup violence. Therefore, further tests of the religious belief hypothesis ought to measure degree of belief independently of justifications in the context of intergroup violence.
To retest the belief hypothesis empirically one needs to do so in a manner that does not measure independent and dependent variables that are so close in meaning as to make relationships between these variables entirely unsurprising—or tautological. Thus, we do not advise that one should use belief in martyrdom or belief in the duty to kill infidels as predictors of support for suicide attacks, as LMS suggest. This approach may produce positive correlations, but they would be trivial. Of course those who believe in martyrdom are more likely to support martyrdom; likewise, those who believe it is a moral obligation to kill are more likely to support killing. We accept as an obvious fact that religious Muslims who support suicide attacks are also more likely than religious Muslims who do not support suicide attacks to believe that Islam supports such attacks. A less trivial question—and the question we addressed in our research—was whether it is devotion to a specific religious belief system (e.g., Islam, Christianity, or Judaism) that drives such support. It seems critical then that the religious beliefs to be measured should be beliefs defined by the religions concerned as core beliefs.
Related to this last point, a third critical issue in testing the religious belief hypothesis is that there is often very little variance in core religious beliefs in many of the populations one might be most interested in sampling. For example, Islam's most fundamental belief is that there is no God but God and Muhammad is his messenger. Not only is this the central belief (the Shahadah) of Islam, but to hold such a belief implies endorsement of the canon of Islamic beliefs more generally. While this would seem a good candidate for a predictor variable to test the religious belief hypothesis, the belief does not vary much in the populations concerned. For example, in multiple surveys of Palestinians we have asked participants to rank order how important different “virtues” are—including the “Belief that Allah is the one true God and Muhammad is his prophet”. This belief is ranked as the most important by more than 95% of the population in each survey. Similarly, there is likely little variance in afterlife beliefs in highly religious populations of interest, and therefore this variable is difficult to test empirically tested. To give another example, we interviewed students in four Indonesian madrasas (religious boarding schools), one of which, Al-Islam, was associated with Jemmah Islamiya. Students at Al Islam were highly radicalized, with a majority believing it was their duty to fight and kill non-Muslims. Yet religious beliefs did not predict such endorsement of violence as all students we sampled were extremely religious. For instance, almost all students in all schools believed it was very important for Indonesia to be ruled by Sharia (Ginges and Atran, 2009).
As we demonstrate in GHN, a tractable way to deal with these issues is to investigate devotion to the body of beliefs that make up a religion in creative ways—by asking how important religion is to the lives of participants, or to measure religious devotion indirectly through prayer frequency, which we have found varies even in highly religious populations. In one study we found that a reminder of prayer in fact dampened support for a specific suicide attack, and in three other studies we found a lack of correlation between measures of devotion to a participant's religion and support for suicide attacks or parochial altruism. Thus, while we remain open-minded about whether religious belief is related to other forms of intolerance and violence (see, for example, Hansen and Norenzayan, 2006), regarding support for suicide attacks and related forms of parochial altruism, we have found little support for the validity of the religious belief hypothesis.