Repetition of experiments

Even if you have taken great care to ensure that your experiment is well designed and statistically analysed, you are limited in the conclusions that can be made. Firstly, what you can say is valid for a particular place and time, with a particular investigator, experimental subject and method of applying treatments. Secondly, if your results were significant at the 5% level of probability, there is still an approximately 1 in 20 chance that the results did arise by chance. To guard against these possibilities, it is important that experiments are repeated. Ideally, this would be done by an independent scientist with independent materials. However, it makes sense to repeat work yourself so that you can have full confidence in your conclusions. Many scientists recommend that all experiments are carried out three times in total. This may not be possible in undergraduate practical classes or project work!