1828 RURAL ECONOMICS 



before being used, they can lead to very erroneous conclusions. The present 

 paper deals solely with methods of testing data and with the means adopted 

 for making them more reliable. 



Data collected from different sources are not always comparable and 

 in order to check such figures it is often possible to make use of calculated 

 values drawn from a single secondarj^ source For instance, in order to 

 check the data on the production of forage crops in 27 districts of the 

 Province of Rorfie (i), the unit consumption was worked out for each district 

 (Table I) from the weight of live stock maintained and the amount of fodder 

 required to feed a unit weight of live stock per annum, assuming that it 

 takes 12 parts of hay or hay equivalents to maintain i part of live weight 

 for I year. There is little agreement between the two sets of figures, show- 

 ing that the returns for the forage crops, or for the live stock or for both were 

 unreliable. Some sources of error there must always be even in the most 

 careful records of this kind, but they should not be of an order to suppress 

 all correlation between the two sets of returns. Assuming that the live stock 

 returns are more likely to be correct than the forage returns, it follows that 

 the latter should only be used with caution. 



Another method of checking the reliability of data is based on the fact 

 that there should be some sort of order connected with the chronological 

 sequence. For example in Tables II and III, data are set out representing 

 the production of olive trees and vines over the period of their grovii:h, the 

 data being collected by various investigators and in difterent parts of Italy. 

 When plotted, the data give rise to very irregular curves which are a direct 

 proof of errors in the returns, as except in cases where considerable modifi- 

 cations take place in the treatment of the plantations or where a disease 

 appears, the average yield from such permanent crops shotild give a fairly 

 smooth curv^e. 



Ordinary statistical returns collected by the Government are often 

 very imperfect. During the Enquiry into the conditions of the agricultural 

 population in the south of Italy, it became apparent that the data on the 

 numbers of metayers, farmers, etc., published in the population census were 

 almost useless. A confusion had been caused by the bad wording of the 

 schedules on which the returns were collected, the various classes of holders 

 not being suificiently well defined, so that under the group metayage were 

 entered not only the various kinds of tenant parternerships, but also 

 ordinary tenant farmers and other types of tenancies. 



Besides testing the reliability of data, a critical examination often 

 indicates a means of improving them and of smoothing out curves by the 

 use of arithmetic, algebraic or graphic methods (2). Bj^ thus eliminating 

 oscillations due to incidental causes the general trend of the phenomena 

 under discussion is more clearlj^ defined. The use of arithmetic methods 

 is particularl}^ adapted to chronological series where the irregularities are 



(i) Minister di Agricoltwa, Industria e Commercio, Ufficio di Statist tea agraria, Catasto 

 agrario del Regno d' Italia, Vol. VI, No. 3, Part 2, pp. 11-65. Rome 1916. 

 (2) See R. 1914, No. 845. 



