Top Page | Upper Page | Contents | About This Site | JAPANESE

Sampling of Experiment Data

Beginners often put basic levels of factors, experiment each factor one by one and report effects of each factor.

Design of Experiments teaches us that such experiment is not enough because the experiment using a basic levels does not experiment the case that more than two factors are changed togother.

Three Rules of Fisher

Fisher is the pioneer of DE and taught us the points, especially for agriculture.

• Repetition : To estimate the size of error
• Randomness : To minimize systematic error
• Local Management : To manage systematic error using block factors

If we take many levels for one factor, we can estimate the size of error. And repetition is not so important.

Comparing to agriculture, factories and laboratoris make homogeneous places. So local management is not so important in this case.

Three rules of Fisher are considered for the experiment of agriculture. So these are not always important in other fields.

Kinds of Sampling of Experiment Data

• Location Experiment : Basic sampling
• Orthogonal Array : Reduce the number of experiments

NEXT Location Experiment