Experimental Design Considerations
- Quality data only result from thoughtfully designed experiments. So, take your time, think through each experiment in advance of beginning your work in the laboratory, and discuss your plans with others including your advisor.
- When designing experiments identify all of the potential variables in the system, control them, and vary only one variable at a time.
- Look for and eliminate all possible sources of error.
- Use the highest quality experimental methods, reagents, and instrumentation available. Purify your reagents if you know that they are impure. Collaborate if necessary in order to obtain access to instrumentation and methods that are of the highest quality. Make sure that your methods and instruments will produce data with the required degree of accuracy and precision.
- Establish good sampling methodology. This means determine what size sample must be analyzed in order for the results to be statistically meaningful for your research problem. Replicate analysis should always be performed on a series of independently prepared samples. Too often in the classroom, laboratory experiments emphasize the analysis of three replicate samples. Students may walk away with the mistaken impression that three is the magic number that should be used in all experiments. Analysis of three samples, however, is justified as useful pedagogically in that it provides students with the opportunity to execute the experiment several times and therefore develop their lab technique in what is admittedly a very finite block of time. The reality is that rarely is three samples an adequate number of samples that will produce statistically significant results in an experiment.
- Always carry out any necessary control experiments.
- Record all data in a permanent laboratory notebook. If the data are obtained using a computer-interfaced instrument, a minimum of two copies of the data should be preserved.