The received wisdom for some time has been that looking at CP makes it more likely that a pedophile will offend against a child. When men who have committed a contact offense are interviewed, they often describe CP viewing as part of the process that led up to the offense. But there is no evidence of a causal relationship -- whatever process was in motion that led to offending could have led to CP viewing as a side effect. There is no control group -- it is entirely possible that a great many pedophiles view CP but never abuse a child.
Milton Diamond and colleagues in a <a series of studies> found that widespread availability of pornography in a society is associated with sex offending either going down or staying the same, and in particular that the availability of CP results in reduced or level sex offending against children. I discussed this in <an earlier post> . Here there is no need for a control group. They are measuring the overall effect on the population itself -- how many sex crimes are occurring in the society in question.
Earlier I gave a review of Michael Seto's "<Internet Sex Offenders>", . He criticizes the Diamond et al studies by noting that there has been a major drop in crime of all kinds during the past few decades. Although the causes are not understood, he says it is simpler to view the drop in child sex abuse as a part of that trend, and not associated with the availability of CP. This is a plausible alternative explanation, but it in no way settles the issue.
He raises in contrast evidence that CP increases offending against children.
Correlational studies are the formal version of the observation that increased CP viewing preceded offenses against children, and they suffer from the same flaw -- they don't distinguish between a causal role for porn and the idea that a rising tendency to offend causes both porn viewing and the later offense.. Most studies consider the different case of ordinary, legal pornography and sex crimes against adults. Some are limited to the viewing of violent pornography (not pornography in general). Seto himself notes that very little CP that people download is violent.
Experimental studies expose people to pornography (not CP) in a lab setting and compare them to a control group. Some find that more porn exposure leads to various indirect measures of increased aggression. I see two major flaws with them. One is that when people are shown porn in a lab, there is the implicit message that the experimenter and the institution sponsoring the research both think that it is OK to view this material, and that may be the cause of any effects in contrast to the normal case where an individual decides himself whether to seek out such material. Another is that subjects who are aroused in a lab setting never proceed to orgasm and release in the experiment. The frustration of arousal without release might itself lead to aggressive attitudes.
A group performed <a detailed and careful study> of sex crimes in different regions of Norway correlated with when broadband internet became available. They attributed 3% of sex crimes to this availability. It's a good study, though it's only one and there are alternative explanations. Limitations aside, note that 3% is a rather small number. The effects observed in the correlational and experimental studies are also quite small.
Putting it all together, what's the effect of CP? We don't know for sure. On the one hand, some small portion of a 50% drop in sex crimes could be attributed to CP availability. On the other hand, other studies with their own flaws suggest an increase on the order of 3%.
To me the most relevant result is that CP viewing does not cause a dramatic rise in sex offenses against children. Freedom and liberty suggest that if something is not inherently harmful, criminal penalties should not even be considered unless it has a dramatic effect on crimes that do cause harm. CP production of course should remain a crime because of the harm to children in the production process. When it comes to virtual CP, there is no reason to prohibit either production or viewing.
We could also do better by adopting different rules for different people. If an individual finds viewing of CP to be putting him on the verge of offending, any sensible person would argue strenuously that he should stop. Seto's view is that CP viewing has a bad effect on at-risk individuals, not on average people. One solution he suggests is banning those who have been convicted of a contact offense from viewing virtual CP while allowing it for everyone else.