3. Faulty Rulings and Misplaced Column Headings

TABLE L

TABLE SHOWING ACCIDENTS CAUSED BY FALLS OF WORKMEN - BY CAUSE AND DISABILITY

Causes. . 1,387 100.0 48 2 30 425 384 110 346

Falls down 52 3.7

On the basis of the manner of treatment and the controlling factor in statistical arrangements, tables are of three types. First, those which express historical data; Second, those which describe a situation or condition in cross-section; and Third, those which express variable data of a non-historical character. Each of these types deserves brief consideration.

The controlling factor in tabulations which express historical data is, of course, chronology. Normally, the arrangement is simple and easily comprehended. All of the facts, no matter how diverse in frequency or divergent in type, are controlled by this consideration, thus giving a continuous view from the standpoint of time. This arrangement does not, however, suit all data equally well. Only when a table serves primarily as an instrument of record and when considerations of time are significant should chronology absolutely dominate. In cases where the time element is incidental it should be reduced to a subsidiary position. The degree of prominence to be given to it depends in each case upon the purpose of the table.

The second type of tabulation from the standpoint of contents is that in which a situation or condition is described in cross-section. The controlling facts are the relationships which maintain between the respective things described. The following table relating to scales of wages for plumbers in Massachusetts municipalities will serve as an illustration:

TABLE M

TABLE SHOWING UNION SCALES OF WAGES FOR PLUMBERS ON OCTOBER 1, 1913, BY MUNICIPALITIES. (LABOR BULLETIN No. 97, MASS. BUREAU OF STATISTICS, p. 39, Boston, Mass.)

The data refer to a single period of time and reflect the methods of wage payment, among municipalities, and the different rates of wages at the period to which they apply. That is, the table shows not only geographical distribution but also the relationships maintaining between hourly, daily, and weekly wage-rates. For cross-section tabulations of this type commanding importance should be given to those considerations which are most suggestive. Related things should be placed in juxtaposition in order to facilitate comparisons. Before the form is decided upon the relationships which it is desired to emphasize should clearly be determined and the table be prepared to register them. Tabulation is rarely the first step in analysis; frequently it is the last step, the early ones having been taken in deciding upon the form to be used. A large part of the exposition necessary to make plain what it is intended to show can be obviated if a table on its face unmistakably reveals its purpose. There is nearly always a best form, and it is the peculiar function of the person using statistics to discover it. After all, tabulation is only a method of summary expression where lines and columns are used to reveal relationships and sequences.

The third type of table, from the point of view of its contents, is one which expresses a variable fact at a single period of time. In describing a characteristic of a natural phenomenon one is impressed immediately by the regularity which the measurements, in which the characteristic is given, assume. Regularity of distribution around a central tendency approaches the absolute when dealing with numerous samples and with pure chance selection. If one were to compare the lengths of a great number of leaves, chosen at random from a particular tree, he would be impressed by the degree of uniformity and by the regularity of the graduations on either side of those lengths which

might be called normal or typical. The same uniformity of distribution characterizes the stature or weight of men, size of apples, weight of eggs, or of any other natural thing where chance has freely operated in the choice of the samples.

Similar regularity of distribution occurs when one thing is measured many times. The measurements tend to differ because of the limitations of the physical instruments and of judgment in their use, but these tend to be corrected as the number of measurements is increased. That which is typical or characteristic tends to be established, and the exceptions above and below it to become fewer and fewer as the distance from the norm increases.

In the measurements of certain economic phenomena the same tendency toward regularity of distribution as between that which is normal and that which is extreme is noticeable. Wage-rates vary within narrow margins for the same type of labor for a given district, and between districts the differences are not startling. For a given occupation a norm or typical wage tends to be established. Wages above and below this standard may be thought of as exceptional both as to the amounts paid and the number of individuals receiving them. The foot frontage value on a certain residence city street tends to vary only within a narrow margin, the amount of deviation from the extremes being relatively small and the frequencies relatively few. Down-town business blocks tend to be about six to eight stories in height. There are a few blocks higher than twenty stories and a few old-time buildings - misfits which are but two or three stories high. Most American freight cars have a capacity of from thirty to fifty tons; very few now in use for freight services have a capacity of less than fifteen tons, while few are built with a capacity beyond one hundred tons. The ruling interest rates on real estate mortgages, in Wisconsin in 1904,

L

were 5 and 6 per cent. 3 per cent; and a few others at more than 10 per cent. The most characteristic rate was 5 per cent. A degree of normality in these examples is noticeable, but it does not maintain generally in the same rigorous fashion in economic as it does in natural phenomena.

Some loans were made at less than

TABLE N

FREQUENCY TABLE SHOWING CLASSIFIED WEEKLY WAGES FOR EMPLOYEES IN ALL MANUFACTURING INDUSTRIES IN MASSACHUSETTS, 1912.

(27th Annual Report, Statistics of Manufactures of Massachusetts, 1912, p. xxii, Boston, Mass.)

1 Note the changing widths of the groups and the treatment of the residuum.