How do u improve your experiment and prevent bias

No, no, no contamination in your experiment. Cut off all sources of possible bias. 

Always be as precise as you can.

That's just to name two ways.

Considering it's talking about bias, I believe it's referring to more of a "social" experiment-- not a laboratory one-- so contamination won't be a problem. To improve your experiment, you need to have a wide range of test subjects, and a large number so you can get more accurate results. And to prevent bias in an experiment, you might want to do a double blind study.