Walking around the neighborhood after publishing my last post I had a sudden pang of anxiety: Were the things I wrote about STV elections really the case? In particular, people have always been able to take advantage of bloc voting to elect a favored candidate (or candidates); what’s so different about STV compared to a conventional at-large election?
Fortunately I was able to convince myself that STV worked (or at least could work) the way I thought I did. I’m now writing my thoughts down just to make sure—and of course in the hope that others might find this informative.
A good place to start is to think about how an STV election could allocate council seats between Democratic and Republican candidates. Consider again our hypothetical 2014 council election, in which 102,000 voters cast ballots (approximating the turnout in the 2010 council election). Further suppose that 57,000 of those voters are predisposed to vote for Democrats and 45,000 are predisposed to vote for Republicans (approximating the parties’ respective vote totals in the 2010 council election).
If 2014 were a traditional at-large election then the strategy for Democrats would be very simple: field a slate of five candidates and ask Democratic voters (and Democratic-inclined independents) to vote a straight party ticket. If this went according to plan then each of the five Democratic candidates would receive at least 57,000 votes, no Republican candidate would receive more than 45,000 votes, and Democrats would secure a 5-0 county council majority.
Note that it would not matter if Republicans ran one candidate or five in this example; the result would be the same. The only thing that could go wrong for Democrats would be if one of the Democratic candidates were significantly less popular than the others, so that a Republican candidate could sneak through with a victory. This model corresponds to the historical reality of the five at-large council elections held in Howard County through 1982: only in one election (1970) did Democrats fail to achieve 4-1 or 5-0 majorities on the council.
Now instead suppose that the 2014 election were held according to the STV scheme, with a quote of 17,001 votes (one-sixth of the total 102,000 votes, plus one). Now it’s Republicans who have a simple yet effective strategy: First, run only two candidates for county council, say Greg Fox and Robert Flanagan (the top GOP council vote-getter in 2010 other than Fox). Then ask half of the 45,000 Republicans and Republican-leaning independents to vote for Greg Fox as their first choice with Flanagan as their second choice, and the other half to vote for Flanagan with Fox as their second choice.
If all went according to plan then Greg Fox would receive 22,500 first-preference votes (45,000 divided by 2) and Robert Flanagan would also receive 22,500 first preference votes. Since 22,500 is greater than the quota of 17,001 both GOP candidates would be elected to the council. Note that this is independent of how the Democratic candidates fare; polling above the quota based on first preference votes results in an automatic win.
However suppose Howard County Republicans instead fielded three candidates, say Greg Fox, Robert Flanagan, and Dennis Schrader, with Fox receiving 18,000 first preference votes, Flanagan 14,000, and Schrader 13,000. In this case only Fox would be automatically elected to the council. Even if all of Fox’s 999 excess votes (18,000 minus 17,001) were transferred to Flanagan (i.e., as Fox voters’ second choice) he would still have only 14,999 votes, well below the quota of 17,001. Assuming that no Democratic voters selected Flanagan or Schrader as alternative choices, both candidates would remain below the quota and would be eliminated, and Democrats would take a 4-1 council majority.
Bad as this might be for the GOP, there’s an even worse scenario, namely that Republican voters split their first choice evenly among Fox, Flanagan, and Schrader, leaving each with 15,000 first preference votes, well below the quota of 17,001. Assuming that no Democratic voters selected a GOP candidate as an alternate choice, all three would be eliminated and Democrats would take a 5-0 council majority.
Note that unlike the effects of gerrymandering under a district scheme, such a result would be a self-inflicted wound resulting from not understanding how STV actually works. The key point is that in an STV election a party should resist fielding more candidates than the number of quotas (i.e., multiples of 17,001 votes in this case) it can reasonably expect to receive as first-preference votes, and should work to as much as possible ensure that those votes are spread evenly across all its candidates.
Thus at 45,000 expected first-preference votes in our example the Republican party has only 2.6 quotas (45,000 divided by 17,001) and should consider fielding no more than two candidates. On the other hand in our example the Democratic party has 3.4 quotas (57,000 divided by 17,0001) and should consider fielding no more than three candidates.
Of course in a given election a party may project it will do better than expected and take a chance on fielding more candidates. It may also in some cases expect to attract transferred votes from one or more of the other party’s candidates, if some voters designate as their second choice (or third, or fourth, or …) a candidate of a different party.
Attracting transferred votes is key to another challenge, namely how to get elected as a candidate representing a relatively small voting bloc; I’ll discuss that topic in the conclusion of this two-part post.