Study on Religion and Killing

Abstract:

The claim is often made by the religious that religious persons are more moral than non-religious persons. The counter position, by secular humanists, is that one is driven to immorality by primitive ideologies. To answer this question empirically one must first consider a measure of “morality,” thereby establish a rate, and then see if certain populations stand out as less or more “moral.” If ideology does not influence morality, then there should be no difference in the empirical behavior across groups. One can consider killing as a measure of immorality and ask, does religion promote mass killing or does the teaching of religious morality alter a “natural” tendency to kill amongst the human animal? This paper attempts to empirically answer these questions. Certain religious groups are statistically significantly more likely to kill than others.

Introduction:

Religions and those that attack religions suggest that social concepts of morality are affected by teachings. Those teachings, in turn, lead to moral or immoral predisposition to certain acts of the followers (otherwise there would be no point in teaching morality). Certainly the wars fought in the name of religions have been the hallmark cry against the morality of religious teaching and a call to claim religions as medieval dark remnants of a horrific past. Atheists have called the crusades immoral, christians have called jihad immoral, and muslims have called the crusades immoral. Killing holds a unique position as a mark of group immorality amongst various groups.

Though immoral behavior is varied in its constructs, killing for personal gain is certainly not commonly considered a moral behavior. Where theft can be moralized by need, adultery by survival of the species, dishonesty by compassion, killing stands out as a uniquely immoral act. It is difficult to identify moral killing. Being killed for one's belief has also been recognized as a great moral act by all groups. The subtlety of just intention in killing is difficult to ascertain. The person being killed often does not consider the killer as more moral even when the killer makes a moral claim. Thus we can look at killing as an act, moral or immoral, that is a signal of the effectiveness of social moral teaching. As most religions and humanists would consider excessive killing immoral, this can provide a measure or immoral behavior.

Killing is clearly a universal act. A rate of killing that is common to all groups would suggest that man has a certain preponderance to kill and moral teaching is insufficient to alter that. A rate of killing that is higher in any group (religious or otherwise) suggests that the teachings of that group promote killing. A low rate of killing suggests that the group's teaching prevents killing. Further, a historical recurrence of killing suggests an intrinsic value that transcends the temporal culture interpreting the morality of killing – an ideological / doctrinal origin. A single event can be attributed to an individual, a small extremist group, or historical circumstances forcing a poor choice. However, a recurrence of history suggests an underlying teaching supporting (or unable to inhibit) killing.

Persistent small scale killing has been ever present throughout history varying by a myriad of factors as temperature, resources, oppressive local regimes, etc. Thus this ongoing carnage holds little information when comparing the effects of long term ideologies. To find a socially significant effect one must look at killing at large scales. One must recognize, though, that looking at large scales biases selection to institutional actors that have the feasibility to cause mass killing - potentially leading one to conflate institution with religion as the correlate to mass killing. Of course, this in turn merely shifts the question to the effect of moral teaching on institutional actors. For example, colonialism is clearly well associated with mass killing. Yet colonialism needs a moral justification. This is attained by religious or secular moralities.

Using killing as a measure of morality, recognizing a signal in mass killing beyond small scale killing, this study compares the rate of killing across religious and atheist groups and then across particular religious groups to assess the effect of moral teaching. The kill rate is denominated by the number of adherents and the duration of the ideology as described below. The study does not assess causality.

Methods:

Every war and man made famine (considered as a political tool causing death) in which more than 100,000 estimated dead was included. Such events were segregated into those considered by various authors as religious or those targeting specific religious / ethnic groups. This included wars within religion as well as across religion. Data were taken from various Wiki sites (1,2) where maximum and minimum estimates were available. Where such a range was not available the given number was used as both. Mass killing that involved two groups were counted in both groups (perhaps falsely assuming equal responsibility) unless the power of one group far exceeded the power of the other group (i.e, Anglican church supported killings in India). Although Wiki sites are not academically reviewed, the data they provide allow a starting point to evaluate trends across large data sets.

Religious categories were assumed as commonly understood – Christian, Muslim, Hindu, Jewish, Buddhist. Atheist inspired mass killings were those targeting traditional religious groups and thus represent an expression of ideology. Secular mass killing were included as those wars where multiple religions fought on multiple sides without a specific religious goal. The concept of just war was not considered.

Questions to be addressed:

Are religions more likely to kill than non-religious or atheist groups?
Are all religions equally likely to kill?

As raw numbers do not translate into rates, two rates were defined:

the number killed divided by the current adherents of a group as defined by the Pew Research group (3). The current adherents is a simplification as the more correct denominator would be of the adherents at the time of the killings. Using the current adherents allows a more objective denominator that is comparable across groups, avoiding bias in local regional variation. This simplification would reduce the killing rate of the groups growing faster (christianity, islam, and atheism) while relatively increasing the estimated kill rates of the religions that have lost substantial numbers over time (judaism, hinduism).
the number killed divided by the centuries a group has existed (from founder or believed). This denominator is simpler for those groups founded in “modern” times (buddhism, christianity, islam, and atheism). The origins of hinduism and judaism are at best estimates but certainly they preceded the more modern groups by millennia. The origin of atheism is also difficult. Jainism is an atheist group that has not been involved in mass killing extant since at latest 500 BCE. Political atheism was manifest in modern times by the development of secular politics, particularly the American and French revolution, and later by communism.

Minimum and maximum rates were calculated using the ranges of dead provided by the sources. The difference in kill rates was calculated between religious groups and atheist groups. The relative risk of killing was calculated by dividing the kill rates by that of the lowest kill rates (hindus).

The data was also segregated into those deaths prior to widespread use of guns (below 1500 ce), those where guns were available but modern warfare as limited (before 1900 ce), and those occurring when modern warfare was available (after 1900 ce). Though this represents a certain zeal to kill versus an ease to kill, a clear quantitative scale is not available (number of people needed to kill x number of peoples...). Geographic data was also evaluated but kill rates could not well be correlated as these often included forces of christian, islamic, or secular colonial spread.

Chi squared contingency statistics were used to assess statistical significance.

Results: (The full data are tabulated in the appendix.)

Religious v Atheist v Whole population:

All killings with greater than 100,000 persons killed resulted in somewhere between 382.2 and 891.3 million deaths. With a current population of 6.2 billion and the modern historical record lasting 9013 years this results in a kill rate of 61.6-143.8 deaths per 1000 living humans or 4.2 to 9.9 million humans killed per century. This is our inheritance as a species. This can be taken as the natural rate or at least as the expected rate.

All killings with greater than 100,000 persons killed by atheist groups targeting religious groups resulted in somewhere between 36.7-95.1 million deaths. With a current atheist population of 1.1 billion and the atheist political historical record lasting 2513 years (since the Mahaviras) this results in a kill rate of 24.5-63.4 deaths per 1000 living adherents or 1.5-3.8 million humans killed per century.

All killings with greater than 100,000 persons killed by religious groups resulted somewhere between 96.2-263.3 million deaths. With a current religious population of 4.23 billion and the religious political historical record lasting 9013 years this results in a kill rate of 22.7-62.2 deaths per 1000 living adherents or 1.1-2.9 million humans killed per century.

Major religions v each other.

We can use the same methods above to see if certain religion's are empirically more or less likely to kill. That killing may be just or unjust. That remains a different question.

Christians have been involved in 36 recognized religiously motivated wars or engineered famines with more than 100,000 killed. This does not include the colonization of Europe, the persecution of the pagans and other Greek and Roman religions. The estimates vary between 67.1-217.9 million killed. The largest variance is in the killing of indigenous of the Americas (2-100 million). This results in a kill rate of 30.8-100 persons killed for every 1000 christians alive today or 3.3-10.8 million killed in christian conflict for every century christians have been in the world. The christians have been involved in killing in large scale in every epoch of war in large numbers: 8.2-11 million by the sword, 33.1-159 million by simple guns, and 25.8-47.9 million by modern warfare. In the Americas there have been 3.4-102.46 million killed. In Southeast Asia, 21.5-34.5 million killed. In Europe, 22.1-51.6 million killed. In the middle east, 1.5-4.6 million killed, 0.4 million killed in the far east.

Muslims have been involved in 19 recognized religiously motivated wars with more than 100,000 killed. This does not include the colonization of the Middle east, North Africa, and the Far East. The estimates vary between 38.4-57.1 million killed. This results in a kill rate of 24-35.7 persons killed for every 1000 muslims alive today or 2.7-4 million killed in muslim conflict for every century muslims have been in the world. The muslims have been involved in killing in large scale in the pre-gun and modern era: 23-30 million people by the sword, 5 million in pre-modern warfare, and 10.4-22.1 million people by modern warfare. In Southeast Asia there have been 21.3-33 million killed. In Europe, 10.2-12.2 million killed. In Africa, 4.9-7.9 million killed. In the Middle East, 1.9-3.9 million killed., 0.1 million killed in the far east.

The Crusades (1-3 million), Reconquista (7 million), Nigerian Civil war (1-3 million), Second Sudanese war (1-2 million), and Balkan cleansing (2.7 million) are counted in both christian and muslim numbers.

Hindus have been involved in a single religiously motivated war with more than 100,000 killed – partition of India in 1947. This is counted for hindus and muslims. The estimates are between 0.1-1.0 million killed. This results in a kill rate of 0.1-1 person killed for every 1000 hindus alive today and 0.001-0.01 million deaths for every century hinduism is recognized to have been practiced on the earth (9013 years). Note that dropping the number of years to 5000 years only doubles the latter rate to 0.002-0.02. Of course this would be in Southeast Asia and in the era of modern warfare only. The conquests of Asoka vary as to the number killed but this would not have been considered a religiously motivated killing though the cessation of killing was due to buddhist teaching. Also note that the population of hindus has been halved into the modern era since colonization by muslims and christians. If that had not happened the theoretical rate would be 0.05-0.5 persons killed for every 1000 hindus.

Buddhists have been involved in a single religiously motivated war with more than 100,000 killed – the sino-tibetan war 1930-1932. The estimates are of 1.0 million killed. This results in a kill rate of 2.04 person killed for every 1000 buddhists alive today and 0.04 million deaths for every century buddhism has been around. Of course this would be in Asia and with modern warfare available.

Jews have been involved in a two sets of religiously motivated wars with more than 100,000 killed – the Jewish-Roman wars between 66 and 135 and the Arab-Israeli conflict. The estimates are of 1.5-2.1 million killed. This results in a kill rate of 150-210 person killed for every 1000 jew alive today and 0.03-0.04 million deaths for every century judaism has been around (5013 years). Of course this would be in Europe with simple warfare and the Middle East with modern warfare available. Here also one notes that the population of jews has been essentially halved by the holocaust. If that drop had not happened the theoretical rate would be 75-105 persons killed for every 1000 jews.

This data is summarized:

The Chinese killing of religious groups was largely atheist attacks on the religious and are included elsewhere above. (Sino-tibetan war - 1 million, Maoist inspired great famine of China – 15 to 55 million.)

The martyrs killed for their religious belief were (double counting of wars is corrected for in generating these numbers):

It must be recognized that the data reflect only killing over 100,000 persons. And thus underestimates the effect of low scale long term killing. However, if we were to include systematic killing of 10,000 to 100,000, 10-100 such events would result in a change of only 1 million. Further, if we were to include systematic killing of 1,000 to 10,000, then 100-1000 such events would result in a change of only 1 million. Thus the net effect of such low level, even long term killing would be negligible to this data (though not to those religious groups made extinct by such practises).

The degrees of order of differences in rate of killing are important and apparent. But one can take the lowest rate of killing in any group as denominator and then calculate the killing risk associated with other groups. This estimates a relative risk of killing behavior, an odds ratio perhaps, that can be ascribed to any group with the assumption of the lowest risk as the basis for comparison.

Note that the minima are based on the lowest estimates of those killed and the maxima are based on the largest estimates of those killed (both as numerator and denominator). Also recall that the rates for judaism and hinduism could both appropriately be reduced based on population reduction due to the effects of religious killing.

Statistical analysis:

To do a contingency Chi squared analysis, we calculate an expected rate of kill based on the kill rate of all wars in the data base. This can give us an expected kill rate for the total population over the years of human history and thus see if subgroups are different than the overall kill rate. The extant years of an ideology can be seen as the opportunity to kill. If the kill rate is equal per person, then the proportion that that ideology represents of the current population would represent the expected part of the killing in those years. The product of the overall kill rate (millions per years), the years extant, and the proportion of the population belonging to an ideology creates an expected number of kills. The difference of the actual and the expected then represents overkill or under kill by an ideological group.

Using the data for minimum kill rates one finds that christianity, islam, and atheism have killed in excess of their expected kills while hinduism and buddhism have killed less than would be expected if the killing was equally distributed over time and persons. The finding is the same if the data for maximum kill rates are used.

Based on the a Chi squared contingency, the odds of this being a random difference is less than 10^-120

Conclusions:

Religious v Atheist v Whole population:

In general, the overall kill rate is similar amongst the atheists and religious groups. These, in turn, have a generally lower kill rate than the total for humanity. Both ideologies seem to moderate the effect of human killing, equally so.

Major religions v each other.

But amongst religions, clear differences can be seen as to the kill rate. These demarcations subsist when comparing to the atheist kill rate as well.

Hindus have the lowest rate of killing per capita and over time. Buddhist and hindu rate of killing can be seen as similar though the buddhist rate is double of that of the hindus. Because the data is so limited for these groups, broad comparisons across these groups seem to push the data too far.

Christian, muslim, and jewish rates of killing are similar and approximate the atheist kill rate per capita. Christians and muslims exceed the atheist kill rate over time. The jewish kill rate over time approximates that of hindus and buddhists.

Thus we can conclude that hinduism/buddhism teach non-killing, that christianity/islam/atheism teach non-killing less well, but humanity without an ideology is far more effective at killing. Jews hold an intermediate position but perhaps best understood in their lessening of killing over time. Of course the teaching may apply to the act of killing or are merely represent a failure to control the natural / expected preponderance.

One also must stop to recognize that the most killed are the indigenous, hindus, and buddhists. The very same groups that are less likely to kill. Christian killing has largely been from christians and atheists.

Based on the weighted (by years extant and percent of current population) expected kills, jews are about at expected. Christians, muslims, and atheists have substantially and significantly exceed their expected number killed. The hindus and buddhists have killed significantly far less than expected. Note that the weighted average accounting for years extant and current population can be used to assess smaller groups as a general model.

Discussion:

If God only loves one group then whether that group is good or not is not relevant. Those on God's side go to heaven, those not so go to hell. Morality is irrelevant. But if morality is an empirical event that can be studied as acts in the world, then the effect of a moral tradition can be assessed.

It is apparent that the rate of killing does vary mildly when comparing the ideological motivation in killing. Humanity has been killing without ideology for a long time at a high rate. Either as religious or as atheists, we moderate that urge.

The atheist killing rate may well reflect the ease of killing in modern times rather than a historically consistent predisposition in the atheist stance. Hindu, buddhist, and jewish killing seem to have been single events and do not suggest historical preponderance to killing. Christian and muslim kill rates are well sustained throughout history. This contrast suggests that doctrinal differences are significant and playing a role.

If we take killing as an immoral act then indoctrination has been used to reduce kill rates – improve moral behavior. But where all groups do better than the “natural” kill rate, the fact that the abrahamic religions exceed their expected kills and the dharmic religions undershoot their expected kills, one must conclude that the dharmic religions are more effective and doctrinally predisposed to improving moral behavior beyond that of the abrahamic or atheistic traditions.

Due to limits of historical data, one must admit that other groups may well have killed in higher and more consistent rates, but equally one must admit that no other groups killed in higher or more consistent rates. Thus the limitation of historical data is just that and precludes judgment on groups outside of the historical record.

The inclusion of groups into large monolithic data points is a challenge to all such studies. Neither are all atheists alike nor all christians, nor all muslims, nor all hindus, etc. Because a group has a preponderance of killing does not suggest that an individual does. But the exclusion of killer identity based on the killing itself (the true atheists do not kill, true christians do not kill, true muslims do not kill, etc) also can not be accepted as these killers are self described as religiously motivated and thus proclaim in their identity and action a disagreement with the apologist. From an objective standpoint, we must recognize that religions are composed of those that would kill and those that would not. A statistical analysis such as this merely points to a preponderance of the group to kill – not the individual.

The distinct rate of killing between religious groups suggests that the moral teaching of these groups does indeed effect behavior and likelihood of killing. The actual cause of that killing remains elusive without a deeper analysis into the actual doctrines of those groups. However, if the differences were minor one might be compelled to argue for a minor role of religious teaching in killing. Were the events unique rare historical events, one might argue that killing is not intrinsic to the teaching, rather a rare circumstantial effect. But this is not the case. Christians and muslims have consistently killed throughout their history where as other religions have not. Christians and muslims have consistently killed in significantly higher numbers than other religions. Atheists have only begun killing in modern history (perhaps in part reflecting political empowerment).

One could ask if the fault lies in the teacher or the teaching. Certainly these are interlinked. A poor teacher with a good teaching may well cause harm. A good teacher with a poor teaching may well compensate for such. But if we are to assume that humans are by and far equal in teaching and learning ability, then the focus of sustained killing must lie in the teaching rather than the teachers that have interpreted the scriptures over millennia.

The martyrdom of the peaceful is perhaps not surprising. But the data strongly contrasts the rate of killing compared to those being killed. Clearly those groups that do not kill also avail themselves to accept dying rather than becoming killers. Though this study does not examine such motive, one must respect the morality that one takes to one's own death. The high number of indigenous, hindu, buddhist, and jewish deaths reflect their population's moral tendency and the onslaught of christianity, islam, and atheists against them. The relatively lower rates for christian and muslim martyrdom is surprising but perhaps points to a tendency to more easily persecute those outside the tribe than those within. For christians this comes largely at the hand of other christians and then atheists. For muslims this also come to some extent by other muslims and largely from christians.

Unfortunately, modern warfare and policy has made it easier to kill many more people with less effort. The effect of this on the religious field is somewhat difficult to discern with this study. Christian killing has been ongoing on a mass scale prior to such “advances.” Muslim killing has accelerated in the modern era, prior to that it was largely focused on killing of hindus. Atheism, largely in the guise of communism, has also suggested a high tendency to use modern warfare to advance its moral teaching. Prior to the modern era, secular killing was prominent and significant. But as an ideology, the onset of modern atheists political thought has not been effective in reducing killing rates.

The conflation of colonialism with ideology (christian, muslim, communist, or capitalist) is clearly difficult to segregate out as an independent cause of killing. To some substantial correlation, the moral justification of colonial expansionism is based in the religious and other moral constructs underlying geopolitical aspirations. To the extent that the sense of moral superiority, whether religious or secular, informs the right to colonialism, one would certainly expect that to contribute to an increased rate of killing. This was not evaluated in this study.

The effect of institutionalized religions and proselytizing religions has also been suggested. The data do not suggest that as islam and buddhism are prominent counter examples to these suggestions.

One could argue that the natural rate of killing is zero. Any killing is wrong. Indeed such a utopian perspective is respectable, but, alas, merely utopian. One could argue that killing is justified when under attack. Based on this data, the right to kill, then, would belong to the indigenous religions, jews, buddhists, and hindus. Yet this is not the stance of those being killed.

A more detailed evaluation of kill rates accounting for population estimates at the time of killing may add nuance to these findings. Likewise, inclusion of rates of ongoing murder, rape, sequestration of wealth and opportunity, mis-education of adherents, limits on the rights of women, promotion of hate or hateful killing as moral duties, and other oppressive regimes in various societies as informed (or not regulated by) religious teaching would add to the available data as to the effect of religious teaching on moral / immoral behavior. Indeed, a scriptural analysis of the hatefulness of the god worshiped (perhaps the number of verses in which god is said to hate divided by the total verse count in the main scriptures) could well point to a causal link to the adherent's rate of killing. Such studies would add dimension to the concern for the benefit of various groups to human society. These are beyond the scope of this study.

Clearly one would want to attribute a specific causal relationship. This study was not designed to do so and caution must be applied when such flights of fancy are entertained. But the study does find that, if religion can be said to influence moral behavior, if killing is immoral, then certain religions (ie Hinduism, Buddhism, and the indigenous) cause a greater morality to be evident in their adherents. Causality may well be related to the doctrinal differences in abrahamic teaching versus that of dharmic groups. That doctrinal difference needs to be explored. A deeper understanding of why various groups are more or less likely to kill may well allow a greater movement towards a more peaceful world, reducing killing rates, despite a more complex world with ever greater capacity for easy killing.

Sources:

(1) http://en.wikipedia.org/wiki/List_of_wars_by_death_toll accessed May, 2013 and 1-14-2014

(2) http://en.wikipedia.org/wiki/Religious_war accessed May, 2013 and 1-14-2014.

(3) http://www.pewforum.org/files/2014/01/global-religion-full.pdf accessed May, 2013 and 1-14-2014

Appendix:

Raw kills data (x – denotes redundant counting) (click on tables to enlarge).