|
|
|
|
Margin of Error The term “margin of error” is repeated incessantly during election campaigns in the reporting of opinion polls. However, since a tiny fraction of our population has ever studied statistics, few news consumers (or reporters) actually know what it means. (It refers to a confidence interval mathematically determined from the sample size and the shape of the statistical distribution.) Suffice to say that opinion polls are inaccurate predictors of how the electorate will actually vote. After the November 2000 election, we found that the adage “The only poll that matters is the vote on election day” gives no comfort when we discover that elections are not immune from their own margins of error. Putting sources of error aside for the moment, the balloting margin of error hardly concerned us in the past because it was usually exceeded by the margin of victory. But the situation was bound to happen when the margin of victory is nearly zero, making the margin of error loom large, a situation we faced in 2000. What was the appropriate solution? Believe it or not, the Founders foresaw the very real possibility of such an event, and provided for it. But before we discuss it, let’s look at the facts: 1. It is certain that the 2000 presidential election was a draw. If we look at Florida alone, fewer than 1,000 votes are currently in dispute out of 6 million cast. This amounts to 17/1000 of one percent. Since no polling system is perfect (and admittedly the systems in use in much of Florida are far from perfect, though better than average nationwide), this is much smaller than the polling error could possibly be. Nationwide, the edge is larger, but still within the margin of error. 2. The margin of error can be reduced somewhat by recounts under certain circumstances. If voting is performed by machine, recounts can uncover possible machine and clerical tallying errors. To uncover possible errors, the recount must be performed by different machines and tallied by different clerks and compared with the previous count. But recounts introduce another margin of error. If the two totals are within the recount margin of error, then the original total must be accepted as the most accurate. If not, then neither total is reliable and the law must be prepared for that eventuality (as is Florida law). Successive recounts may validate earlier counts but such validation becomes less likely with each recount because of compounding errors. 3. Human (commonly called “manual” but is actually visual) counting is inherently less accurate than machine counting. In small precincts where paper ballots designed for human counting are used, the error difference is negligible and acceptable. In populous precincts, where machine balloting is required to hold error levels within acceptable margins, human recounts are far less accurate than machine recounts and should only be employed if the calculated machine errors are unacceptable (see (2) above). However, even if a machine count cannot be certified, it is nearly impossible for humans reading machine ballots to reduce the margin of error and it is more likely that human counting will produce a less reliable result than originally obtained. 4. Because of (3), human recounting of machine ballots is appropriate only for the purpose of validating the machine results, but only by using a statistical sample, not the entire set of ballots. Once the machine margin of error has been verified, additional counting by humans produces no valid information and the results may not be relied upon to refine the margin of victory. 5. Election shenanigans introduce other “errors” which are detectible but not calculable. In precincts where significant irregularities are evident, the law has tools for redress. However, neither candidate has formally made a charge of irregularity, so we may ignore this factor for now, while keeping in mind that human recounts may introduce irregularities that did not already exist. Therefore, the presidential election was a draw and no amount of recounting (except as described above) will produce a more reliable vote count, even as totals change in either direction, because no recount is guaranteed to reduce the margin of error. State laws that recognize this fact place time limits on vote certification and establish recount criteria and restrictions. Otherwise, if a candidate loses in a precinct within the margin of error, it is theoretically possible to reverse the result if a sufficient number of recounts are performed. Likewise, when the margin of victory in a precinct is greater then the margin of error, and the recount discerns additional votes, the absolute number of additional votes detected will always favor the victor. Thus, an attempt to force multiple recounts should be recognized for what it is (a tactic designed to reverse the outcome of the election) with the understanding that it cannot resolve the issue of victory. The question therefore is not “Who won?” but “Who wins in a draw?” since we can’t have a tie. A lot of people have been calling for a runoff. This suggestion directly contradicts the intentions of the Founders, who opposed multiple balloting (and other ex post facto remedies such as re-votes) and carefully crafted Constitutional procedures to avoid them. What if the second ballot is also a draw? Others have suggested that the Electoral College is an anachronism, and that direct election would have avoided this “crisis” or that the electors should vote in proportion to the popular vote. Facts: 1. The candidates campaigned for the electoral vote, not the popular vote, and would have conducted their campaigns much differently if all they needed was the popular vote (for instance, they would have campaigned only in urban and populous areas and ignored most of the rest of the country) so we have no idea what the outcome would have been in this instance. 2. The U.S. is a federal republic, not a pure democracy, by design. Because of its huge geographic size and diversity (even in 1787), the Founders believed that a republic was necessary to moderate the adverse influences of regionalism and factionalism and protect minority (regional, economic, political) rights. For this reason, only members of Congress were directly elected, and only by their constituents. (In European-style democracies, legislative members are seated in proportion to the percentage of their party’s national vote obtained. This system produced Hitler and Mussolini.) 3. The 1992 and 1996 presidential elections were not resolved by popular vote, since no candidate received a majority. The Electoral College determined the victor. No one can predict who would have won if alternative methods were used. So it doesn’t matter that the election is a draw, since the Constitution (Article II, Section 1, as amended by the 12th Amendment) provides sufficient republican means to select the president. Even if Florida can’t appoint its electors (for whatever reason), there is no crisis, because the Constitution provides for any eventuality. Our leaders should have reminded us that we have solutions already in place, and it’s not court-ordered manual recounts, which accomplish no more than increase the uncertainty of the original outcome and served only to deepen the unnecessary electoral crisis. |
|
|