Conference acceptance process is "random" -- is it any less so for journals? [duplicate]

This question already has an answer here:

• Studies over how noisy is it to accept/reject submissions 1 answer

A recent experiment at the NIPS conference in computer science basically shows that more than half of the papers appearing at the conference would be rejected if the review process were rerun. Is the acceptance process generally any less random for journals?

The answer could involve statistics/experiments, but since these are unlikely to exist for journals, it could simply be drawn from personal experience or based on personal thoughts.