Religion is often thought of as psychological defense against bad behavior, but researchers have recently found that the effect of religion on pro-social behaviors may actually be driven by the belief in hell and supernatural punishment rather than faith in heaven and spiritual benevolence.

In a large analysis of 26 years of data consisting of 143,197 people in 67 countries, psychologists found significantly lower crime rates in societies where many people believe in hell compared to those where more people believed in heaven.

"The key finding is that, controlling for each other, a nation's rate of belief in hell predicts lower crime rates, but the nation's rate of belief in heaven predicts higher crime rates, and these are strong effects," lead author Azim Shariff, professor of psychology and director of the Culture and Morality Lab at the University of Oregon said in a university news release.

"I think it's an important clue about the differential effects of supernatural punishment and supernatural benevolence. The finding is consistent with controlled research we've done in the lab, but here shows a powerful 'real world' effect on something that really affects people – crime," he said.

Religious belief generally has been perceived as "a monolithic construct," Shariff said. "Once you split religion into different constructs, you begin to see different relationships. In this study, we found two differences that go in opposite directions. If you look at overall religious belief, these separate directions are washed out and you don't see anything. There's no hint of a relationship."

Previous research published 2011 in the International Journal for the Psychology of Religion found that undergraduate students were more likely to cheat when they believed in a forgiving God compared to those who believed in a punishing God.

Interestingly, in 2003 Harvard researchers found that the gross domestic product (GDP) was higher in developed countries when people believed in hell more than they did in heaven.

Researchers said their findings support previous findings that a society's predominant belief in either heaven or hell strongly predicted a country's crime rate.

"Here, we investigate these effects at the societal level, showing that the proportion of people who believe in hell negatively predicts national crime rates whereas belief in heaven predicts higher crime rates," researchers wrote.

"These effects remain after accounting for a host of covariates, and ultimately prove stronger predictors of national crime rates than economic variables such as GDP and income inequality," they added.

Researchers had accounted for factors like nations' dominant religion (Roman Catholic, other Christian and Muslim), income inequality, life expectancy and incarceration rate.

"Supernatural punishment across nations seems to predict lower crime rates," Shariff said. "At this stage, we can only speculate about mechanisms, but it's possible that people who don't believe in the possibility of punishment in the afterlife feel like they can get away with unethical behavior. There is less of a divine deterrent."

Shariff noted that because the findings were based off of correlational data, they do not prove causation.

The data for belief in hell and heaven, belief in God and religious attendance were taken from World Values and European Values surveys conducted between 1981 and 2007, and crime data were based on United Nations records of homicide, robbery, rape, kidnapping, assault, theft, drug-related crimes, auto theft, burglary and human trafficking.

"This research provides new insights into the potential influences of cultural and religious beliefs on key outcomes at a societal level," Kimberly Andrews Espy, vice president for research and innovation said in a statement. "Although these findings may be controversial, dissecting the associations between specific belief systems and epidemiologic behaviors is an important first step for social scientists to disentangle the complex web of factors that motivate human behavior."

The findings are published in the Public Library of Science journal PLoS ONE.