Sec. 1.2 PURPOSES OF STATISTICAL REASONING 5 



based on that attitude toward the purposes of statistical analysis. 

 Unfortunately, the statistical ideas and procedures which Student 

 introduced in 1908 did not become familiar to persons outside his own 

 firm for nearly a decade, at which time R. A. Fisher and his col- 

 leagues in England began to extend and to popularize the theory of 

 small samples and its applications. The theory of statistics was 

 developed extensively by Jerzy Neyman and Karl Pearson's son, 

 Egon S. Pearson. They placed special emphasis on rigor in statisti- 

 cal reasoning and led the way by publishing many papers in this field. 

 Many others have followed their lead since their papers began to 

 appear. The results of this research are being applied in many fields, 

 such as biology, the physical sciences, industry, economics, sociology, 

 medicine, education, and psychology. 



1.2 SOME OF THE PURPOSES OF STATISTICAL 

 REASONING 



Early in his history man displayed a desire to take numerical 

 measurements of the various phenomena involving himself and his 

 environment. At first, those measures probably consisted of simple 

 counts, or of crude measures of weight, volume, length, and area. At 

 present many instruments are available for the precise measurement 

 of those features of man's self and environment which interest him. 

 He constantly is taking groups of numerical measurements because 

 such a procedure can furnish a relatively precise and standard means 

 of obtaining the information desired, of using it efficiently, and of 

 transmitting that information to others. The general purpose of 

 statistical analysis is to assist in the collection and the interpretation 

 of sets of numerical measurements which supposedly have been 

 taken for some useful purpose. 



Once it is decided that a particular phenomenon should be meas- 

 ured numerically, one of two general classes of data is then ob- 

 tained. It may be that it was both possible and practicable to secure 

 every measurement of that particular kind which exists or could be 

 obtained under the particular circumstances. Such a complete record 

 is one type of statistical population of numerical measurements. An 

 example is the record of the ages of all the legal residents of the state 

 of Kansas on April 1, 1950, as contained in the official United States 

 Census for that date. Another example is a list of the I.Q.'s of all the 

 students entering a particular university in a given year. 



