There's a distribution of 'dickness' in the population. That's true regardless of what other metadata you use to segment your sample. The causal direction between extremism and religion isn't easy to prove though - are religious people more likely to be extreme in their view, or are people who are more inclined to extreme viewpoints attracted to religions simply as means to validate those extreme views?
The number of muslims heading off to ISIS to the utter shock of their much more moderate families suggests it's a lot more complex than 'you inherit your parent's views'.