STATISTICAL INQUIEY 1003 



The influence of season would be ascertained by arranging 

 the cases in the four quarters of the year in which they 

 occurred. The influence of age would be obtained by 

 arranging the cases according to their ages, and the other 

 dividing characters would be examined in like manner. 



In grouping results, it is very difficult to arrange the 

 dividing characters so that all the cases selected will fit 

 into them, and that one case will fall only to one group 

 and to no other. A disregard of this latter rule— viz., that 

 no case can appear in two groups at the same time — will 

 render all groupings worthless. 



In observations extending over a series of years, we have 

 an opportunity afforded of ascertaining whether there is an 

 improvement in the general health, and a decrease in the 

 mortality. By a careful consideration of facts we can be 

 led to determine the causes acting beneficially or otherwise 

 in this direction. 



A sufficient number of facts is absolutely essential if any 

 correct deductions are to be drawn. Conclusions based on 

 a small number of observations are insufficient to establish 

 a fact, for the reason that it can be mathematically 

 demonstrated that where the number of units is small, the 

 chances that the same result will be obtained again with a 

 similar number of cases is highly improbable ; whereas, 

 when deductions are drawn from a large number of units, 

 the chances of error are less, and the probability is that 

 with a similar number like results will be obtained on 

 another occasion. The reason why this is so can be readily 

 appreciated; errors in a few numbers alter considerably the 

 value of the results, whereas errors acting through a large 

 series of numbers tend to neutralize each other, owing to 

 some being greater and others less than the real value. 



The larger the number of observations the greater the 

 accuracy, but the error only diminishes as the square root 

 of the number of observations, so that twice the accuracy 

 is not obtained by doubling the number of observations; 

 it requires four times the number of facts to effect this. 



When we have a large number of units collected which 



Digitized by Microsoft® 



