Back to: Assess for Risk of Bias
Let’s say that the study researchers used a low risk of bias method to randomize participants, but by chance they happened to randomize sicker people into one group more than into the other.
It’s possible that this can happen, and doesn’t mean that the researchers did something wrong. The process was random, but it created seemingly non-random groups. You would be surprised how often you can get four, five, or even more heads in a row a if you toss a coin long enough!
The problem is, when you find a pattern in a randomization sequence, you don’t know whether it happened by chance or because of a flaw in how the randomization was carried out. To determine which is more likely, consider the extent of the pattern and the source of the data.