The three principles of experimental design as stated by R.A Fisher are – replication, randomisation, and local control. Following these three principles of experimental design allows us to have confidence in the validity of our analysis and enhances the precision of our experiments.
Replication:
Replication refers to the repetition of a similar treatment on different experimental units. The advantage of replication is that it allows us to average out the influence of random effects other than the treatment. The resulting average gives us a precise estimate of the effect of the treatment on the experimental unit.
For example, suppose that we want to test the effectiveness of a particular fertilizer on the growth rate of crops. Then it is better to apply the fertilizer to many different patches of our field. This allows us to remove the effect of the soil quality on the growth rate of crops. The average obtained is a good measure of the effect of that fertilizer on crop growth.
Randomisation:
Randomisation refers to the use of a random process to assign treatments to the experimental units. By assigning treatments randomly, each treatment has an equal chance of being assigned to the experimental units. The process of randomisation also eliminates the bias in any form.
Suppose that in the above example, the researcher was biased towards a particular fertilizer. The researcher may then purposely assign that fertilizer to a patch of good quality soil in order to manipulate the results. Randomisation ensures that the assignment of treatments is taken away from the hands of the researcher and thus any such bias is eliminated.
Local Control:
The process of reducing experimental error by dividing the experimental area into blocks is known as local control. For example, when testing the effectiveness of fertilizers it is better to divide our field into blocks. Now blocks closer to each other will have similar soil quality. So by assigning treatments in such a way that each treatment is assigned only once to a particular block type we can ensure that the conclusions of our study are unaffected by the soil quality.
The above three principles of experimental design were first stated by the renowned statistician R.A Fisher in his book – “The Design of Experiments”.
