Really good scientific research is done using a triple-blind approach.
Those who design the study do not gather data, and those who gather data do not analyze it and draw conclusions.
This process, of course, is quite rare. Its purpose is not to prevent intentional fraud, which is probably pretty rare, but to eliminate scientists finding what they expect to find.
We have a great deal of evidence, stretching over more than a century, that this is a HUGE problem in science.
It’s bad enough when such confirmation bias only helps scientists “prove” their own hypotheses. It’s far worse when (to use AGW as an example) finding A results in fame, fortune and the attentions of admiring female students, while finding B results in ostracism and loss of funding for future research.
The way Einstein’s theory that mass/gravity bend light was a beautiful example of how science should be done.
He presented his theory and the mathematics he believed would back it up and then he stepped aside as hundreds of astronomers tested and re-tested the theory. After several years the evidence became overwhelming and forced a fact based consensus.
If you had a telescope and an eclipse you could test it yourself today and get the same result.