An orthodoxy of social science is that studies of ethics should be descriptive and not prescriptive; they should describe what is without evaluating what ought or oughtn’t be. But approaching this ideal – which not all aspire to – is no easy task, especially when important cultural values are involved.
A good illustration is a recently published academic article, ‘The Negative Association Between Religiousness and Children’s Altruism across the World’ (you can read it here). In addition to being intended as descriptive it is also critical: it compares the judgements and practices of children of faith and non-faith using the standards of ‘pro-sociality’ and ‘altruism’. As its title suggests, religiousness was found to be inversely correlated with altruistic behaviour; in addition, some religious children were found to display greater punitive tendencies than the non-religious. But these contentious findings rely on a number of assumptions. Exploring them provides a good opportunity to examine problems of the relation of values to science.
First some details of the study: participants were selected from seven different locations in six different countries (Turkey, Jordan, USA, Canada, South Africa, and China). 43% identified as Muslims, 23.9% as Christians, and 27.6% as non-religious (the remaining 5.2% from other denominations were too few for significance). The parents of children participating in the study were asked to rank their children’s empathy and sensitivity toward justice, they also responded to a questionnaire ranking their own religiousness. Two behavioural experiments were then conducted on the children. The first experiment: A resource allocation experiment called ‘the dictator game’ where some children are given a greater share of scarce resources (stickers) and told “these are yours to keep” and their behaviour is subsequently recorded (do they share or not?). The second experiment: children observe scenes displaying ‘interpersonal harm’ (pushing, bumping by an individual) and are asked to rank the ‘meanness’ of the behaviour; they rate the severity of the punishments that should be meted out to the bullies.
The results of these experiments are as follows: the parents of religious children rated their children’s empathy and sensitivity to justice higher than did the parents of the non-religious children. The dictator game results showed that non-religious children were more likely to share stickers with others than both Christians and Muslims; they were therefore deemed to be more altruistic. The results of the ‘interpersonal harm’ experiment indicated that both Christians and Muslims were more likely to find aggressive behaviour “meaner” than the non-religious, and that Muslims alone were more likely to advise harsher punishments for the bullies than the Christians and the non-religious*. These findings it is stated, challenge “the view that religiosity facilitates prosocial behaviour”; the authors conclude the study by claiming to have supported “the idea that secularization of moral discourse will not reduce human kindness – in fact, it will do just the opposite” (p3).
This study is intended as descriptive social science, but, in order to support the idea that a secular discourse might increase human kindness, have the authors introduced bias into the study and into their analysis? Actually, those concluding remarks just noted are a case in point: it is not required that kindness itself be evaluated as ‘good’–even though the word kindness suggests a good in most people’s lexicon. The question in this case is how is ‘kindness’ to be defined? It is questions like this which are central to the problem of value bias. How are the other concepts which are used (such as altruism and pro-social) defined? And how are the behaviours which exhibit or do not exhibit these characteristics to be judged? There are also some potential issues of bias as a result of the composition of the study which I will get to later.
Altruism was defined as “cost for the donor and benefit for the recipient” (but what cost, what benefit?). A definition of prosocial was not offered, but the statement “if religion promotes pro-sociality, children raised in religious families should show stronger altruism” suggests the authors consider the terms interchangeable. In the analysis of the second experiment the standard of ‘punitive tendency’ is used; this refers to the quantified responses of children for the level of punishment the bullies deserve. But it is the significance of this result which is important – is it good to have ‘punitive’ tendencies, or is it bad? Without making a value judgement this question cannot be answered explicitly, an the authors are careful not to do so. Yet, switch the word good for ‘kind’ and the answer becomes obvious: punitiveness is diametrically opposed to kindness, at least in a conventional interpretation of the words. Thus even without value-judgment, the use of this term, along with ‘intolerance’ and ‘judgemental’ which are also used to describe the behaviours of the “religious” is likely to be received by many readers by the lights of their own values. In other words, the deliberate use of these words may serve the purpose of value-judgements by proxy.
Regarding the findings of the experiments, the interpretation that the first experiment indicated that non-religious children behaved more altruistically than the religious children seems reasonable on the face of it, simply because they distributed more stickers (a personal loss)– though the substantiveness of this finding is questionable given the paltriness of this ‘loss’. But how can the ‘less altruistic’ behaviour of the religious be explained? The authors suggest a phenomenon known as ‘moral licensing’ (i.e. perhaps religious children simply assume themselves to be kinder because they are raised in an environment which stresses their moral character, and perversely (unconsciously) they don’t feel called-upon to act generously). This may be true; yet I can offer a plausible alternative: the religious children followed more literally the wishes of the experiment’s sticker distributor whose literal direction was “these are yours to keep”. This is a phrase with very different connotations to the alternative, ‘these are yours to do with as you like’ which might endorse a freer relationship to the gift. This alternative explanation is also consistent with other research cited in this study which suggests that religious people follow rules more rigidly; if it is true, selfishness may not be the guiding factor, but rather obedience. This is an important point given that the power of the article’s chosen title hinges on the claimed disparity between the altruism of the religious and non-believers.
And what about the finding that religious children (Christians and Muslims) were more likely to find ‘interpersonal harm’ ‘meaner’ than the non-religious, or that Muslims were more likely to invoke harsher punishments for the bullies? Are these negative traits, are they anti-social? As noted already, both traits were characterised as ‘judgemental’, a term which can used neutrally but which also carries evaluative connotations; to support this they cite literature that suggests that religion is “related to increased intolerance for and punitive attitudes toward interpersonal offences”. But what about the motivations of the non-religions (and in the second case, the Christians)? Were their actions more prosocial, kinder, or were they laissez faire, or even apathetic? These questions are not raised. Likewise, there is no acknowledgement in the paper that the religious’ higher ratings on the “meanness” scale may be a result of compassion and empathy toward the victims (terms commonly connoting virtue); nor whether the higher rating of punitive judgement of the Muslims is prosocial in the sense of upholding the mores, standards of justice and laws of the community – functions served by policemen, judges, teachers to some extent. The explanation offered, which is supported solely by citing literature about ‘fundamentalism’ is that the religious may be more inclined to black/white judgements and have an inability to conceive of a gradient for transgressions. By contrast, the authors characterise the non-religious behaviours with positive words: they displayed more tolerance by not judging the bullies; and when Christians and non-religious express lesser punitive tendencies, this is also judged as tolerance (and presumably empathy) toward the aggressors.
The choice of these terms suggest that the authors may have introduced a bias in how they choose to study/analyse, one that favours characteristics seen to be ideals of developed liberal democracies –a mode of ethical behaviour/reasoning which is individual oriented – and ignores those behaviours which may positively relate to extra-individual ‘community standards’. Returning to the study’s concluding sentence “that secularization of moral discourse will not reduce human kindness – in fact, it will do just the opposite”, while as already noted, a judgement about the value of kindness as such is not here made (this would violate the orthodoxy of description only); instead, what is implied is that religiousness is not necessary and perhaps even detrimental to the kind of society the authors have in mind, though there is no exploration of what the functioning of such a society would entail.
One further complication with a trans-national study of this kind is the difficulties of skewing and error in the sample itself. Although attempts were made to exclude the effect of country of origin, age and socio-economic-status, there is the question whether these measures were adequate to isolate religion from cultural factors (political/institutional context etc.).
It is highly likely that the majority of Muslims participating in the study (who made up 43% of the sample) were sourced in countries where Islam is predominant. That is, given that participants were sourced in Turkey and Jordan (where 99.8% and > 94% identify as Muslims) it is improbable that these two contributed anything but a majority Muslim sample. And yet these are countries which also have vastly different social/political/economic contexts to other countries in the study.
Likewise, despite sizeable Islamic and growing Christian contingents (along with Daoists, Buddhists etc.) in the city of Guangzhou, relative to the population these religious groups are relatively small, and it is likely a majority of participants identified as non-religious (it is another question entirely whether the categories ‘religious/non-religious’ of Chinese society are commensurable with the other countries studied); this is important because some studies suggest that Chinese children are more altruistic than western children (‘Children’s Moral Reasoning: Influence of Culture and Collaborative Discussion’, is one). Admittedly, these are only speculations, but if correct the study would have limited data as to how Christians or the ‘non-religious’ in Jordan or Turkey, or Christians and Muslims in China would respond to the same experiments; it would therefore be hard to justify that this study is about religion rather than culture.
The findings of this study are presented as knowledge (if provisional) of the ethical consequences of religion (or ‘non-religion’). Did they succeed? Is this knowledge? They established interesting correlations of differences of behaviours of religious/non-religion children in different countries (though I have argued the religions factor may be inadequately isolated); this may make a basis for other studies in the future. But the published text suggests a lack of disinterest in some of the analytic decisions and limited effort to try and understand what ‘really’ motivated the children’s behaviours. And while the authors were analytical in their use of the concept ‘altruism’ and remained descriptive in the presentation of the facts, by selectively citing literature that associated ‘negative’ characteristics (‘less tolerant, less sophisticated in ethical judgement and less altruistic) with the religious, they may have voiced their value-judgements by proxy.
This study is a good example of how factors which are not evaluative but which are value-relevant (related to how and what was studied) can be used to lead to (or at least to lead toward) ethical evaluations while ostensibly remaining descriptive. It would be easy for a reader of this paper to interpret the conclusions as an indictment on religion when in fact deeper issues of the relation between religion and society and institutions were not explored.
*Although all were statistically significant, none of these findings indicated extreme disparities.