d'Hondting Pennsylvania

by: Joseph

Sat Apr 23, 2011 at 5:29 AM EDT

After reading about the recent Finnish parliamentary elections, I wondered how their Nordic variety of open party list, highest averages seat allocation would work if applied to a bothersome dummymandered state like Pennsylvania. Thus inspired, I applied the d'Hondt method, which they use in Finland, to the 2008 and 2010 election results. The d'Hondt method metes out seats according to party vote totals across a large jurisdiction -- a state, in the US.  There would be no general election districts and no first past the post (FPTP) competitions.

Without districts, Pennsylvania's parties would nominate a list of 19 candidates. In an open party list system like Finland, the party's full list is shown on every ballot, and the electorate votes for a specific candidate on the list. That vote counts toward the party as a whole but also the candidate's ranking on the party list. For this exercise, both parties have conveniently decided to select their list of 19 candidates through regional primaries in jurisdictions that just happen to match up with with the 19 current House districts. Thanks, imaginary parties. An alternative might be statewide primaries with the first-19-past-the post making the list. Of course, the primary campaigns would have been different under such circumstances, but let's not worry about that just yet.

In 2008 Democrats won a new seat, with Kathy Dahlkemper taking PA-03. Does the Obama wave wash another seat up on the shore with proportional representation? For now let's imagine that voters were given a ballot with all of the Democratic and GOP candidates and chose the exact same candidate that they did on actual election day 2008.

Here's the Democratic party list:

Notably, FPTP losing candidate Bob Roggio (who challenged Jim Gerlach) received more votes than FPTP winning candidates Chris Carney, Jack Murtha, Kathy Dahlkemper and Paul Kanjorski. There were quite a few wasted Democrat votes in eastern Pennsylvania. In western Pennsylvania, Murtha won convincingly with 57.9% in his FPTP race but received an underwhelming number of actual votes in his underpopulated district.

Now the Republican party list:

In 2008 Republicans didn't oppose Mike Doyle, which is a bit inconvenient for this exercise, so there are only 18 candidates on their list. The seven Republicans who won their FPTP races in 2008 hold the top seven party list spots. But Melissa Hart, who lost her FPTP contest with 44.1%, is not too far behind Glenn Thompson, who won his with 56.7%. Coincidentally, FPTP opponents Jason Altmire and Melissa Hart both hold the eighth slot on their party lists.

In total, Democratic candidates won 3,209,168 votes, and Republicans received 2,520,805 votes. Third party candidates did not receive enough votes to matter. The seats would be distributed thus:

Democrats win 11-8 rather than 12-7 under FPTP. The Obama wave fails to bring in new seats, and d'Hondt limits Republican losses. Democrats Kathy Dahlkemper and Paul Kanjorski miss out on the seats they won in FPTP, and Democrat Bob Roggio and Republican Melissa Hart take their spots.

As a result, there are no representatives from PA-03 and PA-11, the northwest and the northeast, while there are two each from PA-04 and PA-06, in the Pittsburgh and  Philadelphia suburbs. The FPTP data favor districts with high turnout and close races over modest winning candidates in low turnout and low population districts.

Of course, there are several huge and massive and very large flaws with this exercise.

For one, Republicans certainly would have fielded a full party list of 19 candidates, since every vote helps the party. But the Republican list would have needed an additional 104,882 votes for Tom Manion to pass Jack Murtha. The missing Republican in Doyle's Pittsburgh district was succeeded by a candidate in 2010 who failed to clear 50,000. Even with the increased 2008 presidential turnout, it seems unlikely that a Republican would have won 100,000 votes in central Pittsburgh. (PA-14 representatives went unchallenged in 2000 and 2004, so there's no good reference.)

It is also unlikely that parties would use geographic primary districts in the first place, at least not in the same configuration as current congressional districts.

But the more glaring issue is that candidates would campaign differently under a proportional system, and many voters would have selected different candidates on their party list or a more palatable candidate from the other list.

Campaigns would probably be centered on media markets. The candidates in PA-01, PA-02, PA-06, PA-07, PA-08 and PA-13 would campaign in "greater Philadelphia" and compete for many of the same votes, while candidates in PA-10 and PA-11 would campaign in "northeast Pennsylvania." A cursory scan suggests that the only big media market base splits are for the PA-03 and PA-12 candidates. The discrete media market of the Lehigh Valley would be virtually guaranteed a representative.

Without head to head competitions, voters would be free to find their best ideological fit in what might otherwise be a "lesser of two evils" situation. I'm not sure how this would pan out for the Blue Dogs or "moderate" Republicans. Independent voters in PA-11 who supported Lou Barletta over unpopular incumbent Paul Kanjorski would have had other Democratic options like Chris Carney.  

Also, candidates with demographic advantages or clear ideological differentiation could perform better within party lists by grabbing same-party votes from other parts of the state. Schwartz, Dahlkemper and Hart might benefit from the dearth of women on the ballot. Doyle would likely earn progressive votes from Altmire's district, while Altmire could win more conservative Democrats from Doyle's district. Similarly, Republicans turned off by Pitts could have switched their votes to a more palatable option like Platts or Gerlach.

There would also be the question of ordering the party list. In the Democratic party, the urban politicians would probably be ranked first, regardless of whether the metric is seniority, primary vote totals or power broker decision-making. On the Republican side I'm less clear who would benefit. But high ballot position would be an advantage.

So, yes, there are huge problems with applying data collected from one type of election to a completely different system. If this method were actually implemented in the 2008 election, I imagine the primaries would have produced a Democratic party list with an eastern urban/suburban bias and a Republican list with at least a couple strong urban candidates -- the wealthy businessmen and lawyers who live in cities but aren't stupid enough to run in Democratic strongholds under FPTP.

An 11-8 party split sounds reasonable in 2008, but the big city/suburban to small city/rural split could approach something like 14-5.

In 2010 four Democratic incumbents lost and Republicans flipped a Democratic-held open seat. Proportional representation buffered the 2008 Obama wave -- what about the 2010 tea tsunami?

Here is the Democratic party list:

The Democratic list doesn't change too much from 2008 to 2010. But FPTP winner Mark Critz brought in fewer votes than the three Philly area candidates who lost their FPTP races, and Kathy Dahlkemper drops down close to "some dude" territory. It's clear that the Democratic list is powered by the east, where blue votes are squandered in relatively close FPTP losses and massive FPTP wins.

Now for the Republican list:

In 2010 Republicans again fielded only 18 candidates, with Bob Brady getting a free ride. Charlie Dent drops from 3rd on the party list in 2008 to 10th in 2010. I guess he benefited from Obama surge ticket splitters in the Lehigh Valley? Or just less tea fuel in 2010. Patrick Meehan does very well, besting Philly suburb veterans Jim Gerlach and Mike Fitzpatrick.  Mike Kelly, who won a seat in FPTP, occupies the 16th spot on the list, below four FPTP losers. Somehow he got fewer votes than Allyson Schwartz's opponent. PA-03 really got wiped out in this election.

In total, Democrats won 1,860,644 votes and Republicans won 1,990,529. The third party vote was again not big enough to matter. The seats are allocated thus:

Republicans win the first seat and alternate with Democrats afterwards, for a close split of 9-10 in favor of the GOP. Proportional representation helps keep down Democratic losses; although four FPTP incumbent Democrats still lose, the party overall does two seats better than under FPTP.  Carney, Kanjorski and Dahlkemper still don't make the cut, but the fourth candidate out is Critz rather than Patrick Murphy. In fact, Patrick Murphy wins the 12th seat just ahead of the Republican who beat him under FPTP. Lentz and Trivedi, also FPTP losers, win seats as well.

The unlucky Republicans are Lou Barletta and Mike Kelly, while Charlie Dent, who got the 6th seat in 2008, just squeaks by with the 19th seat.

Manan Trivedi wins the 18th seat despite having fewer votes than Charlie Dent or Lou Barletta, the two Republicans immediately below him; the padding provided by competitive lower candidates like Critz, Carney, Kanjorski, Callahan et al was enough to compensate for his  modest vote count. Although Brady lacked an opponent in this election, the Republicans were very far from winning their 11th seat at the expense of the Democrat's 9th seat.

Following the election of this group of candidates, there are no representatives from PA-03, PA-11, and PA-12 while there are two representatives each from PA-06, PA-07 and PA-08. The west and the northeast seats migrate to the Philly suburbs, basically. It's worth noting that the threshold for Seat 19 is only 199,053. Several strong Libertarian (or other third party) candidates could plausibly round up enough votes (~10,000 per primary district) to seat one of their own in congress.

Given the flawed data, this system appears to effectively moderate swings and realign geographic representation in Pennsylvania, which is probably what would happen in similar states like Illinois, Ohio and Michigan. Bye bye, Joe Walsh? But then Massachusetts, for example, would have Republican representation, and Democrats could probably win additional seats in South Carolina and Louisiana. Third parties would also have a decent shot at a seat in big states like California and Texas. One unattractive feature for mappers: decennial reapportionment would be as simple as cutting one slot off the list in Pennsylvania. But otherwise it seems like it could be fun.