The selection of the best index number is a fascinating subject. It is a curious fact, however, that index numbers of different types usually agree with each other remarkably well, whatever the formula for calculation, the method of weighting, the number of commodities, etc.
In Chapter I some diagrams illustrating this important fact were given. Any reader unconvinced as to the correctness of this conclusion has only to consult the literature on the subject indicated in Appendix VI (especially the writings of Mitchell and Edgeworth) in order to be reassured.
Nevertheless, there are always some differences between the movements of different index numbers, and occasionally these differences are large. Therefore, just as, in determining the physical yardstick, it is worth while to eliminate the effects of temperature and other disturbing factors in order to obtain a unit as nearly perfect as practicable, so it is worth while to construct an index number as nearly perfect as possible.
I shall therefore indicate the points which seem to me of most importance.
The chief factors of an index number are: (1) the agency authorized to calculate the index number; (2) the markets and sources of quotations; (3) the kinds of prices (i.e. wholesale or retail, etc.) to be quoted; (4) the list of goods to be included; (5) the frequency of calculation; (6) the formula for calculation.
Out of the wide range of choice presented under each of these six heads I would, tentatively at least, make my own choices as follows:
(1) The calculation of the index number might well be put in the hands of the Bureau of Standards which now has charge of standardizing every unit other than the money unit. An alternative is the Bureau of Labor Statistics which now publishes excellent index numbers of prices for other purposes.
(2) The markets should be the chief public markets
