120 200 in Chart II. It will be seen that the variability of the percentages of those of voting age who actually voted increases in practically the same ratio as that of the other two series and follows closely their fluctuations. See Chart III. CHART II. Variability of percentage voting (based on median) The extent to which this increasing variability is evidence of real independence in voting must now be considered. The question of causation is here concerned with what I shall term statistical causes and extra-statistical causes. If the increasing variability of the series is due to mere chance arrangement of the figures then we have a case of statistical causation. If the increasing variability is due to independent voting then we have a case of extra-statistical causation. If the increasing variability is due to mere chance arrangement of the figures then the two most powerful statistical causes are: (1) The admission to the Union from time to time of new states I. Republican II. Democratic III. Percentage Voting. CHART III. Variability of popular vote and variability of percentage voting with a small voting population. This would serve to intensify the divergence between the number of votes of the small states and the number of votes of the large populous states and would increase TABLE I VARIABILITY OF REPUBLICAN AND DEMOCRATIC VOTES AT PRESIDENTIAL ELECTIONS Column II, Standard Deviation of Republican Votes, was obtained by computing the standard deviations of the votes by states for each election year. For example, the figure 73, opposite the year 1856, was obtained as follows: the Republican votes by states for 1856 were arranged in series of descending magnitudes; the figure 53 is the standard deviation or mean square variation of this series. The procedure was similar for the other figures in column II. Column III was obtained in manner similar to column II. Column IV, Standard Deviations of Percentages of Males of Voting Age Voting, was obtained by computing the standard deviations of the percentages in the columns of Table II. Only the standard deviations of the percentages for the years 1872 and following were computed, as it was believed that the accuracy of the percentages for years prior to 1872 was uncertain. Column V, Medians of Percentages Voting, gives the simple medians of the series of percentages of the columns of Table II. Columns VII and VIII were obtained as follows: a careful study of the percentages of Table II showed that for the period 1868 to 1872 a certain number of states showed decreasing percentage voting while others showed increasing percentage voting and some remained practically constant; the figures in columns VII and VIII represent the number of states showing this increasing or decreasing percentage for the period indicated. SOURCE The source used by the writer for the number of votes cast by political party, by state, by presidential election was the compilation published in the Fact Book (published by the Current Literature Magazine Publishing Co.) under the direction of Dr. Francis Rolt-Wheeler of the Current Literature Publishing Co. of New York. The data were compiled for the first time from official figures provided especially for the purpose by the secretaries of state for every state in the Union, by the respective chairman of the Republican, Democratic, Socialist, Labor, Prohibition, and other parties for every state, by the chairmen of the national committees of the political parties, by the personal records of presidential candidates, and forms the most authoritative data upon the subject. the variability of that year's series. (2) The unequal increase of population of voting age as between different states. Thus, a very large increase in the voting population of the most populous states would intensify the divergence of potential voters as between the states and might cause the increasing variability of that year's series. That the variability is not a result of the first consideration is shown by Table III. The year of greatest increase in the Democratic variability was 1876, when the figures increased from 81 to 107 and yet no new states voted Democratic in 1876. In 1880, when Colorado voted for the first time, Nevada was already a member of the Union and cast a smaller Republican and Democratic vote than did Colorado. In 1892 five states voted for the first time, but were obviously not the cause of the Republican variability because Nevada, already in the Union, cast a Republican vote smaller than any of the five. The variability of the Democratic vote for 1892 may have been increased by the voting of the new states, Wyoming or South Dakota, because these states brought a new vote smaller than any other vote, thereby increasing the divergence. But it is to be noted that 1892 was a year of decreasing variability for the Republican vote and the year of slightest increasing variability of the Democratic vote, so that it cannot be claimed that the admission of the new states was a cause of the variability. Table III also shows that the admission of Utah in 1896 and of Oklahoma in 1908, was not the cause of vari ability for those years, for at both of those years there were other states in the Union with smaller votes, i.e., Wyoming and Arkansas. Table III therefore demonstrates that the variability of the popular vote is not due to the admission to the Union of new states with small vote. That the variability is not a result of the second consideration is shown by Table IV. The second statistical cause assumed that a very large increase of the voting population had occurred, increasing the divergence and hence the variability. Taking the year 1876 for the largest increase in Democratic variability (from 81 to 107), we find by Table IV that the increase in voting population for the decade 1870 to 1880 for the four populous states was considerably less than the increase in the decade 1880 to 1890 at which period the Democratic variability did not increase so rapidly. Taking the year 1896 for the largest increase in Republican variability (from 143 to 202), we find by Table IV that the increase in the voting population for the decade 1890 to 1900 was not as great an increase as that for the decade 1900 to 1910, at which period the Republican variability did not increase so rapidly. Table IV therefore demonstrates that the increasing variability is not due to the unequal increase of the population of voting age as between states of the Union. Since the increasing variability is not due to these two statistical causes it is probably due to extra-statistical causes. Is the increase due to a growth of political independence on the part of a margin of voters? I will examine two considerations: I If a majority of the states show at the years of greatest increasing variability a higher percentage voting than at the preceding or succeeding election year, then we have evidence that the increasing variability is due to intelligent voting. 2. If, at the years of greatest increasing variability, there has been the shifting of a margin to one political party or to the other, then we have evidence that the increasing variability is due to intelligent voting. For the more careful study of the first consideration Tables A and B have been constructed based in turn upon the more extensive data presented in Tables I, II, and V. It will be seen that for 1876 |