13. Past and present of the most usual of statements which are met in business life, is the comparison of current results with past achievements, the sales of this year, month or week with those of last year, month or week. In all such comparisons the two elements should have equal weight, tho it is an instinctive tendency of the mind to regard what is more remote as something normal with which to compare what has more recently been accomplished.
14. Results, year by men of business will explain why this year's sales exceeded those of last year. It never occurs to them to explain why last year's sales failed to reach the level of this year's, tho in some cases this would be the most reasonable way of stating the case. It may be that this year represents normal conditions, while last year had ab normally small sales.
Because of this tendency to accept the earlier dates as normal, it is well to choose what is instinctively re garded as the basis of the comparison with great care, and to be sure that it actually is of normal character for the purpose in hand. The results of the Census of manufactures for 1919, when available, are ex pected to show an enormous increase over those of 1914. As the latter were collected in a period of business depression a correct view of manufacturing growth will require a resort to the earlier figures of 1909.
15. Results, mcm,th by month.—What is empha sized regarding yearly results applies with equal force to monthly results. We must assure ourselves in comparing results, month by month, that there is no seasonal variation which would affect them. Proper comparisons are possible only when the corresponding month of a previous normal year is considered.
A point that often escapes attention is, that when any given facts occur with a very even, distribution thruout the year, the difference between one month and another will be very slight. Oftentimes the dif ference is fully accounted for by the fact that the months are not of equal length. It is therefore a good rule, when the difference is slight, to reduce the figures to daily averages and make comparisons on this basis.
16. Results,week by week.—The longer the period of -time to which the figures relate, the more likely is the effect of accidental variations to be diminished.
It is never wholly eliminated. Conversely, the smaller the period of time, the greater the caution which must be used. Comparisons of one week with another, whether the previous week or the correspond ing week of last year, which ignore special holidays, weather conditions and the like, prompt to•wrong con clusions. To illustrate, a shoe dealer would not be
likely to compare sales of rubbers in the first week of November in one year with the same week in another without first asking himself how the weather at the two periods compared. The use of the obvious illus tration is intended merely to call attention to the fact that similar conditions may in less degree bring similar effects, and that these are not negligible quan tities. It is therefore advisable, when conclusions are drawn from statistics, to qualify the comparison by the addition of the phrase "all other things being equal." 17. Results, day by comparisons must be handled with the utmost care. That is obvious, be cause it is so easy for some purely accidental circum stances to throw these comparisons out of gear.
One summer, when the public was eagerly watch ing day by day the figures of American exports the New York Journal of Commerce called attention to the fact that altho Saturday was a favorite sailing day for ocean steamers, the reported exports for Satur day were invariably less than for other days of the week. Investigation showed that the daily returns reported to the press were not those that were made each day, but those tabulated each day. As Saturday was a short ,working day in the government offices, fewer figures were prepared on that day and hence the falling off in exports reported.
18. Establishing standards of obvious remedy for some of the difficulties noted in the preceding sections is to establish proper bases for comparison. These usually take the form of a fair average of past conditions, but under certain cir cumstances it is possible to establish a normal result and observe the variation of actual figures from it.
The purpose of such an average is to eliminate the fluctuations incident to particular periods. As a gen eral rule the average used as standard should ap proach as closely as possible the period which is to be judged by it. An average of the years 1910 to 1913 might be very effective for judging conditions of 1914, yet comparatively worthless as a measure of re sults in 1919. The advantage of the average and the advantage of nearness in date, can sometimes be com bined by taking a moving average. Under such a plan sales for each of the years 1918, 1919, 1920, might be compared not with the same average but with the average of the five years immediately pre ceding in each case.