Empowering Ethnic Diversity

In any case, the possibility of anomalous demographics in early contests should be examined. In the current system, the first stage small-state sample (namely Iowa and New Hampshire) is already grossly unrepresentative of the U.S. electorate… disproportionately white, rural, and middle-class (see Table 1). In contrast, consider as a group the 38 jurisdictions that would be eligible for the first round. The first round eligibility pool has a population of 88.5 million people, or 31.5% of the total population (see Table 3). Clearly, this is a far more representative pool from which to select the first states.

While it is certainly possible that in a given year, the first round might comprise a predominately African American population, and in another year a predominately Hispanic population, et cetera, it is unlikely that the initial round would be skewed in the same way cycle after cycle as it is now and as it historically has been. Over several cycles, the potential for anomalous ethnic distributions averages out.

Considering a single election year, not only is the eligibility pool for the second round even larger than for the first, it is twice as large a sample, thus the probability of anomalous ethnic distributions is much less than in the first round. Also, any anomalous ethnic distribution in the second round would most likely be something other than whatever anomalous ethnic distribution occurred in the first round, if any. The first round might include a disproportionately large number of Asians, for instance, while the second round might be skewed toward whites. The potential for ethnic skewing of the sample diminishes in each successive round and tends to counter any ethnic skewing in the previous round, and each successive round carries more weight than the previous one. The combination of these effects should dampen ethnic (and any other type of demographic, such as household income) skewing down to the noise level by the third or fourth round.

The 20 sample schedules in Appendix 2 were analyzed for anomalous ethnic distributions in the early rounds of the baseline Graduated Random Presidential Primary System (see Appendix 3. As expected, the anomalous distributions tend to balance each other from one round to the next, and to diminish with each successive round. By the third interval (at which time only 11% of the country has voted) the cumulative deviation from the total US population for any ethnic group is generally less than 5%. The standard deviation for the major ethnic groups falls off rapidly from the first interval to the third interval, and for all but one of them the decline continues into the fifth interval (see Figure 19). The surprise is that the standard deviation for the Hispanic population increases from the third round to the fourth round; however, this is easily explained. It is in the fourth that Texas becomes eligible, and the second largest state in the union has nearly triple the percentage of Hispanics as the national average (see Table 10). Because of this, the representation of Hispanics can vary more widely in the fourth round, depending on whether or not Texas is selected to vote in that round. The standard deviation for Hispanics declines from the fourth interval to the fifth interval, and it can be expected that it would further diminish in the sixth round, rise in the seventh round as California (which has the same percentage of Hispanics as Texas) becomes eligible, then steadily decline in the remaining intervals.

However, it is noticeable in the 20 baseline simulations in Appendix 3 that Hispanics and Asians are consistently under-represented in Rounds 1 through 3 (see Figure 20). This is due to the fact that the four largest states, none of which are eligible in the first three intervals, all have large Hispanic populations, and two of them have large Asian populations (see Table 10). This under-representation is somewhat alleviated in Round 4, when Florida, New York, and Texas become eligible, but this bias cannot be completely be eliminated until California becomes eligible in the seventh interval (see Figure 21). Accounting for one-eighth of the population of the US, California has triple the percentage of both Hispanics and Asians compared to the national average. There is no way to balance the ethnic books in the early voting rounds without advancing California's eligibility.

Figure 22 shows the effect of inserting the seventh interval before the fourth, the scheme referred to earlier as Mod 1. This zeroes out the ethnic biases for that round, but they persist for the next three rounds in which California is barred from voting. The Mod 2 scheme, which re-orders the rounds to more evenly spread out California’s eligibility, breaks up this "bias bubble" into smaller bubbles and distributes them evenly through the end of the primary season (see Figure 23), thus smoothing out the ethnic bias as much as possible. As noted earlier, the other argument in favor of the Mod 2/Mod 2A scheme is that it brings the treatment of California in line with that of the other large states: Florida, New York, and Texas.

Figure 24 shows the standard deviation in ethnic group representation in the first five rounds under the modified schemes. Comparing this with Figure 19, it can be seen that the standard deviation for Asians and Hispanics increases in the fourth interval due the "California effect" (and for Hispanics, the "California effect" is added on top of the "Texas effect"). This measures the variability in the representation of these groups, depending on whether Texas and/or California are selected to vote in the fourth round. While there is definitely more variability under the modified schemes, comparing Figure 20 and Figure 25 shows that the modified schemes brought the average representation of all ethnic groups to within one percent of the national average in the fourth round, whereas in the baseline scheme, whites were cumulatively over-represented by an average of 2.6% in the fourth interval, and Hispanics were cumulatively under-represented by 3.3%.

In the baseline Graduated Random System, and even more so in modified schemes, the initial over-representation or under-representation of any ethnic group is short-lived, and the initial advantage or disadvantage of each ethnic group changes from schedule to schedule. This is unlikely to have a dominating effect on a campaign. Rather, this constitutes a healthy variation in the starting conditions of a presidential campaign within a system that inexorably converges toward consensus and solution. However, for the sake of argument, let us consider the possibility that an initial anomalous ethnic distribution could materially affect the outcome of a campaign. For instance, when whites had the initial advantage, as they currently do in Iowa and New Hampshire, a white candidate might end up winning the presidential nomination. To point out that this is not an uncommon occurrence under the current system may seem rather fatuous. On the other hand, in the unlikely event that the Graduated Random System produced a schedule that offered some other ethnic group a significant and persistent initial advantage... might we see the first Hispanic, African American, or Asian nominated for national office by a major party?

It might be argued that such an outcome might not produce a party candidate most fit to compete in the general election. Naturally, this is always the objective of a political party, but is the absolute assurance of such an outcome in every presidential cycle imperative to the health of the republic? Can it not be argued that democracy gains from the occasional glorious failure that sets an important precedent? Alfred E. Smith, the first Catholic to be nominated for president by a major political party, lost spectacularly in the general election of 1928, but surely that precedent mitigated the issue of John F. Kennedy's religion in 1960. In the long run, what is more important: that Geraldine Ferraro lost by a landslide along with Walter Mondale in 1984, or that she was the first woman to receive the vice-presidential nomination of a major political party?

The best measure of the system is whether it opens the future to such possibilities, over whether it guarantees the most competitive candidate every time. This is not to suggest that the two concerns are always mutually exclusive, rather I suggest that the political parties--and American polity as a whole--are best served by a judicious regard for both. An economic analogy may serve to illustrate the distinction. The political parties compete in the marketplace of ideas. As self-interested entities, they naturally seek to maximize the outcome of each transaction. However, specifically who gains the maximum advantage from each outcome is not the concern of the market system. The broader issue is whether the marketplace is maximally open to participation and competition. The political parties should not operate in an environment in which they play it safe within narrow limits by erecting, as they have, barriers to political competition; rather they should operate in an environment in which they are equally empowered to take risks, and to more openly and efficiently appeal to constituencies that have historically been marginalized. The "invisible hand" is prevented from evolving such a market because the major parties constitute an oligopoly. Therefore the objective should be to engineer a more open marketplace on their behalf, and implement it via their simultaneous assent.