CHAPTER 



16 



Experimental Design 



We have seen some of the statistical treatments whose primary purpose 

 is to show the probabiHty that interpretations are correct. This formal 

 mathematical treatment gives a level of "significance" to the results. 

 Experimental design is a term applied almost exclusively to the advanced 

 planning of experiments in order to take advantage of the statistical pro- 

 cedures available. The statistical design takes into consideration all the 

 assumptions that went into deriving the various statistics, thus making 

 the significance tests valid. A more practical effect, perhaps, is that experi- 

 mental design enables the research worker to gain more information with 

 less effort. The designed experiment is as much more effective than 

 the haphazard experiment as that same haphazard experiment is more 

 effective than observation without experiment. The ideal experimental 

 design, of course, is the one that yields the most information and the 

 most reliable information with the least expense and effort. 



The most extensive user of experimental design, and in fact the area 

 responsible for the development of this branch of statistics, is agricultural 

 research. A little reflection shows how necessary it is to the agricultural 

 scientist that experiments be effective. A person interested in the effects 

 of fertilizers on the growth of field crops can raise only one crop a year, 

 and extensive test plots require large amounts of land. Imagine the 

 serious result if badly planned experiments yield no information. Sim- 

 ilarly, breeding and nutrition experiments with large animals are ex- 

 pensive and time-consuming even if carefully designed. Psychology and 

 certain other areas related to human affairs also depend heavily upon ex- 

 perimental design, for somewhat the same reasons. 



212 



