1.If the scientist does a blind study, were he or she does not know which item they are testing
2) Tool used. making sure whatever measurement the investigator is using should be valid and reliable. avoiding the possible results to become biased and unstable due to errors which should be avoided
3) Sampling method. obtain the samples which are appropriate and that would cover the entire participants studied- not a half but the entirely of what you want to study. entirely means as much as the possible sample covers.
4) Have some one else double check your findings. hope that helped