1) have others replicate your experiment and compare results.
2) "choose" at random. What ever it is that you have to select from, let it be chosen randomly.
3) use a computer to store, analyze and organize your data, a computer won't give you a biased response.
