In the comments thread of a previous post on the electoral demographics of Presidential elections since 1980, a commenter wondered:
To what extent did the structure of the Republican primaries contribute to Romney's final standing in the election? (Romney was my favorite candidate and yours too it seems).
The structure probably kept Romney from winning the nomination. When he dropped out of the race after coming up short on Super Tuesday (February 5), delegate counts showed McCain with a seemingly insurmountable lead. He had 695 delegates, Romney had 293, Huckabee 183, and Paul 16. Winner-take-all states were killers for Romney. California, the most lucrative contest to take place while Romney was still running, was the most painful--McCain took 42% of the vote but 93% (158) of the 170 delegates. Romney garnered 35% of the vote but earned a paltry 7% (12) of the delegates.
The commenter also asked what Republicans might do to ensure their party's nominee actually represents their preferences. If delegates had been assigned proportionally and only Republicans were allowed to vote, on the day Romney conceded at CPAC he would've been in the lead with 502 delegates, followed by McCain with 475, Huckabee with 294, and Paul with 74*. In this scenario, when Huckabee eventually dropped out, most of his votes would've went to Romney. To think that if Republicans were actually able to choose their nominee based on Republican voter preference, McCain's career would've forever peaked months ago. How sad.
But making that setup a reality is a long row to hoe. Even closed primaries do not keep people who will end up voting for the opposing party in the general election from influencing the contest, since party affiliation has no influence on what can be done on election day. Every Republican exit poll, including closed contests, reveal voters who report to be independent and also voters who report their party affiliation to be Democratic (my count above only includes the votes of those who reported their party affiliation to be Republican).
If their nomination is already decided, this is especially easy (it's what Rush Limbaugh was encouraging his listeners to do in what he dubbed "Operation Chaos"). Setting the registration deadline several weeks or months ahead of the primary helps prevent this, though.
The only way to eliminate independent and crossover influences would be for both parties to agree to simultaneously hold all state primaries and caucuses on the same day, in one super Super Tuesday. Presumably the putatively unaffiliated would vote for the side they were already leaning towards, essentially relegating them to moderate Republican/Democratic status for an election cycle.
- Only those who reported their party affiliation as "Republican" in exit polls were counted, and delegates were assigned as if those voters were the only ones casting votes in the contest at hand.
For the states without exit polling data, I simply went with the caucus results, assigning delegates proportionally. Inexplicably, no exit polling data for Delaware is available, although it appears that media outlets like CNN and MSNBC had expected there to be. Here, too, I assigned delegates proportionally based on total votes. In both cases, this is almost certainly to Romney's disadvantage, as he consistently performed better among Republicans than among independents (McCain and Paul did the opposite).
The delegate totals add up to more than media estimates at the time did, because of rules about certain delegates being assigned at the convention or otherwise uncommitted due to party rules. I presumed all delegates would be allocated based solely on a state's primary or caucus contest.