Here’s a thought experiment: what if empirical software engineering papers were first reviewed with the results redacted? Once the reviewers had submitted the reviews of the redacted paper, they were then shown the full paper and did a full review. I wonder how much the final reviews would change in practice.
Redacting results in reviews
Published
Surfing Complexity 
But what would you do with the discussion & abstract? Or do you mean to redact only the concrete findings (results of stats tests, for instance)?
The discussion would have to be fully redacted, and the abstract would need to be at least partially redacted. The idea is to have the reviewers read about the study without having any information about the outcome.