[ Pobierz całość w formacie PDF ]
.228 There are many other cases in variousdepartments of economics in which theory takes us up to a certain point,but in which the theoretical discussion needs to be supplemented bystatistics if we are to reach a determinate conclusion.If from considering how general theorems are to be established andtested we pass on to enquire how particular concrete problems are to besolved, we find that the aid to be derived from statistics is relativelyeven greater.There are many important problems of fact especiallywhere a comparison is instituted between different times or places, thoughnot in this case alone which are statistical in their very nature: forexample, the enquiry whether during a certain period of years there hasor has not been an appreciation in the value of gold; the comparison The Scope and Method of Political Economy/159between the position of the labouring classes at different times; the analy-sis and explanation of a given depression of trade; the investigation ofthe relative pressure of taxation, under existing conditions.Upon differ-ent classes of the community.A sound knowledge of theory is requisitefor a satisfactory treatment of problems of this kind.Theory guides usin our selection of statistics, and teaches us how to turn them to the bestaccount.But the data for the solution of the problems must necessarilybe numerical.In the majority of cases, moreover, aggregate regularity has to beevolved out of individual irregularity, and hence a special reason whywe must deal with phenomena in the mass, and not individually.Thusagain taking as our examples the enquiries above referred to if wecompare prices now with prices at an earlier period, some will be foundto have fallen, some to have risen; similarly if we compare wages nowwith wages a number of years ago: even in a year of depression sometrades are found to be flourishing: the pressure of taxation varies in thecase of different individuals in the same class.Averages must, there-fore, be taken;229 and it is clear that the essential conditions for the rightsolution of the problems are reliable statistics, and ability to use thestatistics in a sound manner.The right use of statistics is, indeed, far from being a simple matter.Statistics, it is often said, can be made to prove anything.And if theyare used without special knowledge or grouped simply with the objectof establishing a foregone conclusion, the charge is well founded.isagainst ignorant or prejudiced statisticians, or against the casual em-ployment of a few figures picked up at random and regardless of whatmay be called their context, it is not difficult to defend the paradox thatthere is nothing more misleading than facts except figures.For rea-soning from statistics, in addition to the dangers which it has in com-mon with all empirical reasoning, is subject to difficulties and dangerspeculiar to itself.230 If, however, the limitations of statistics are clearlyrecognized, if they are accurately collected over an adequate range, ifthey are employed without prejudice and after full enquiry into theirtrue significance, and if they are fairly and properly-grouped, then theirvalue is unique, and the statistical method easily makes good its claimto rank as a most effective and reliable instrument of science. 160/John Neville KeynesNote to Chapter XOn Some of the Precautions Requisite in the Use ofStatistics in Economic Reasonings§1.Conditions of the reliability of statistical data. If arguments basedon statistics are to be of any value, particular attention must be paid tothe following points: (a) the sources from which the statistics are ob-tained, with special reference to their reliability; (b) their true meaningand significance; (c) their completeness or incompleteness as coveringthe whole range of the phenomena to which they relate; (d) the mannerof their grouping, with special reference to the taking of averages.Eachof these points may be briefly considered in turn.231The initial difficulty in the use of statistics is the possible inaccu-racy of the original data.Statistics may be obtained and published offi-cially, or they may be collected through private channels.Under theformer of these conditions, the accuracy of the figures is sometimespractically unquestionable; as, for example, in the ease of railway traf-fic receipts.But this is by no means the universal rule, even when offi-cial statistics are forthcoming.Thus up to 1854 the values of importsinto this country were calculated at the prices of the end of the seven-teenth century.From 1851 to 1870 they were officially computed ac-cording to the best information obtainable.At the present time bothimports and exports are in this country valued according to the declara-tions of the importers and exporters.The returns are of course checkedby the officers compiling the statistics; but still their accuracy dependsto a considerable extent upon the good faith and carefulness of the con-signees and exporting agents, who it is said are often insufficiently in-structed by their principals
[ Pobierz całość w formacie PDF ]