After my recent blog post on the U.S. Presidential primary calendar, I was very interested to read an article in this week’s Economist on Republicans in South Carolina moving their primary date earlier, potentially leading New Hampshire to move its primary earlier (apparently by state law, its primary has to be earlier than other states’) and possibly causing Iowa to move its caucuses earlier to 2007 (which its Governor doesn’t agree with).
According to a graphic in the Economist’s story, a number of states or state parties are considering moving their primaries forward. If the states agreed amongst themselves, a more organised primary calendar could be agreed to that didn’t systematically advantage or disadvantage any state.The first step is to determine the shape of the primary calendar. Let’s assume that the first step in the primary calendar is that a few states hold their primaries and then the remaining states either hold their primaries on one election day or on two or three election days. An organising principle in this is that there shouldn’t be any lame-duck primaries, i.e. primaries that don’t matter as a candidate has already won a majority of conference delegates. While it’s impossible to guarantee this, the number of lame-duck primaries can be minimised.
In the linked-to Economist article, it suggests that a few states could be chosen randomly to start the primary season. This would be an improvement on the current system, but it could be easily improved on. One could ensure that states at the start of the primary calendar were geographically dispersed by grouping the states into, say, 5 or 10 geographical regions (each region containing an equal number of states) and randomly select a state from each region to be part of the start of the primary calendar. The ordering of these states in the primary calendar could be randomised.
Alternatively, one could ensure that both high- and low-population states appear at the start of the primary calendar. Order the 50 states by population and divide them into population cohorts (eg the bottom 5 states by population, then the next 5 states and so on), and randomly select a state from each cohort.
One could have both a geographical distribution of states and a population distribution of states in the initial part of the primary calendar. Whatever system one adopted, after the initial state-by-state primary battles, one could have either a mega-primary of the remaining states (which might make U.S. Presidential politics even more beholden to money-politics), or hold a combined primary of, say, 10 states, and a few weeks later, a primary election for the remaining states (or perhaps staggered over two dates). For instance, one could group the 50 states together into 10 regions each containing 5 states (we’ll ignore D.C. in this). One state could be randomly selected from each of the 10 regions and these 10 states could be ordered randomly to be on the initial phase of the primary calendar.
During this initial phase, each of the 10 states would hold its primary in turn week-by-week (or two states hold their primary each week). After the first phase of the calendar is completed, another state from each of the 10 regions is selected randomly, and a “super-Saturday” primary takes place across each of these 10 randomly selected states, ensuring a geographical distribution of primaries.
After this stage, 40% of the states would have chosen their delegates and the remaining few Democratic and Republican Presidential candidates would be known. If, by this stage, the winning candidates for the Democratic and Republican nominations were likely to be known, the system could be redesigned so that fewer states will have held their primaries.
Then, perhaps a month later, the remaining states could hold their primaries simultaneously. The winning candidates would hopefully be clear from the results. It may be the case, however, that there is no clear winner even if only three candidates are remaining, possibly leading to haggling at the convention – an unfortunate outcome!
There are no doubt many practical ways of improving the primary calendar. Introducing a bit of random sampling of states is one possibility.