Requirements For A Good Experiment
1. Absence of Systematic Error
A good experiment should be free from any systematic error. This is possible if experimental units receiving one treatment do not differ in a systematic manner from other experimental units given another treatment. This is achieved by randomization.
2. Precision
A good experiment should have maximum precision. The indicator of precision is the magnitude of standard error, which should be sufficiently small.
3. Range of Validity
To the extent possible, an experiment should be valid in a wide range of conditions so that the researcher can have a high degree of confidence in its results. An experimental technique that gives encouraging results in a certain setting may not give favourable results when conditions are different.
4. Simplicity
If an experiment is a complicated one, it will be difficult to carry it out properly without the help of trained and qualified personnel. It is, therefore, desirable for the experiment to have a simple design, particularly when it is to be performed by relatively unskilled people. Also. it is desirable to use relatively simple methods in the analysis of data.
5. The Calculation of Uncertainty
A good experiment should enable the researcher to calculate the uncertainty in the estimates of treatment differences. This would mean that he should be able to ascertain the statistical significance of the differences between the treatments.
5 Comments