Last year’s partisan redistricting, or gerrymandering, of Wisconsin’s legislative districts by Republicans dramatically altered the political geography of the state. The immediate effect of gerrymandering was to allow Wisconsin Republicans to retain control of the legislature despite losing the popular vote in the November 6, 2012 elections. This basic point has been widely reported by the press, and illustrated with analyses of election results.
These reports and analyses, while generally accurate in their description of how gerrymandering affected the November 6 election, fail to accurately explain the implications of gerrymandering for future elections; in particular how gerrymandering creates risks for Republicans that in the long-term may be as bad or worse than the short-term damage it has inflicted on Democrats. These risks can be seen by carefully comparing the results of the 2010 and 2012 Assembly elections.
The Effect of Gerrymandering on the November 6th Election.
On November 6, 2,668,874 votes were cast for Republicans and Democrats in Wisconsin Assembly elections. Republicans got 1,253,587 votes (47%), while 1,415,287 (53%) went to Democrats. Despite this, Republicans won 60 (61%) of 99 Assembly seats, while Democrats won 39 (39%).
These widely reported numbers do not convey how profoundly the November 6 results were distorted by gerrymandering, taking Wisconsin very far from the one person, one vote ideal. This effect is best seen by looking at the election results from the bottom up, rather than the top down.
How did redistricting affect the power of individual voters?
The average Republican Assembly member won their seat with 20,893 votes, the average Democrat with 36,289 votes. It took 1.74 Democratic votes to have the same electoral effect as 1 Republican vote. In other words, each Republican citizen has 1.74 more times representation in the Assembly than each Democratic citizen.
Even before redistricting, Republicans were overrepresented in the Assembly. In the 2010 elections the average Republican won with 17,742 votes, the average Democrat with 21,825 votes, which gave each Republican voter 1.23 times more representation than each Democratic voter.
Several media reports have suggested that the Democratic edge in the popular vote for Assembly is artificial, because it includes votes cast for Democrats in uncontested races, where no Republican votes were recorded because there was no Republican candidate to vote for. Because 19 Democrats and only 4 Republicans won in uncontested races, the Republican vote total would have increased more than the Democratic total had both parties had candidates in all races.
The notion that Democratic vote totals in uncontested races should not be included when comparing popular vote totals ignores the fact that creating uncontested Democratic districts was one of the primary goals of gerrymandering. Before redistricting, 13 Republicans and 8 Democrats won uncontested Assembly races. After redistricting those numbers were 4 and 19, respectively. These 19 uncontested districts are heavily Democratic by design. Any Republican who ran in one likely would receive few votes. Including results from uncontested races in popular vote totals does not over-represent the Democratic vote by nearly as much as some reporters have suggested.
However, if Republican candidates had run in the uncontested races, Democrats still would have won the popular vote. Even under the unlikely assumption that hypothetical Republicans in these races received the same number of votes as Republicans received in the average race where a Democrat won, Democrats would have won the popular vote, albeit by a smaller margin (50.5%). Republican voters still would have been overrepresented in the Assembly by 1.58:1 compared to Democrats.
Long Term Effects of Gerrymandering
The Wisconsin Legislature was little changed by the 2012 election, but the status quo was maintained only because Republican gerrymandering radically altered the compositions of Wisconsin’s legislative districts. Had the 2010 legislative districts been in effect in the 2012 election, Republicans probably would have lost control of the Assembly. This much is common knowledge. But there is more to gerrymandering than meets the eye.
Reporters covering the election frequently describe gerrymandering as having both increased the number of districts Republicans control and decreased the chances that they would lose those districts in future elections. For example, Chris Rickert writes that supporting divisive legislation is, “of little practical impact for the legislators who, thanks to safe, gerrymandered districts, are unlikely to lose an election over controversial legislation.”
This is a mistake. In fact, it is mathematically impossible. Gerrymandering can either make it more difficult for a party to lose elections, or give it a greater share of legislative seats. On a statewide level, it cannot do both at the same time. The 2011 redistricting did make some seats safer for Republicans, but the overall effect was to make Republican seats less safe (see figures). In fact, it is possible that Republicans have pushed gerrymandering too far for their own long-term good, making them vulnerable to greater electoral losses than they would have experienced with the old districts. At the very least, gerrymandering has created new vulnerabilities for Republicans while leaving Democrats relatively secure in their remaining seats.
To see how gerrymandering can hurt Republicans, imagine a state divided into 100 districts where the electorate initially is evenly split between party A and party B within each district. Each party initially controls 50 seats, but by a coin toss Party A gets to redraw the districts in the way that give it the most political advantage. What should party A do? There are two basic and opposing strategies, a defensive one and an aggressive one. At the defensive extreme, party A could redraw the districts to give itself 98% of the vote in 51 districts, and party B 100% of the vote in 49 districts. This would lock Party A into a majority from which it could be dislodged only if its share of the vote in two or more of its districts dropped to below 50%, a big fall from 98%. Such a majority would be extremely stable but not very effective, because any measure it backed could be blocked by any single member of party A voting against it, and it would be easy for party B to peel off enough votes from party A to get its own measures passed.
At the aggressive extreme, party A could give itself 50.5% of the vote in 99 districts. A 99:1 majority would be very effective in the short term. Party A could pass almost any measure it wanted, and party B could do nothing, as long as the majority held. However, a 99:1 majority would be extremely unstable. A 1% shift in the popular vote, if evenly distributed among all of the districts, would cause party A to lose all of its seats.
An important point here is that by extending its majority as far as mathematically possible, party A is putting all of its seats at risk, not just the 49 additional seats it got through redistricting. With a large legislative majority held by small margins in each district, any substantial shift in the popular vote toward party B would be a catastrophe for party A, while shifts in the popular vote in its favor would give Party A no additional power. Under these circumstances, Party A could retain its power only by reducing the turnout of party B voters, and/or by moderating its positions enough to reliably attract a sizable number of party B voters in every election.
The defensive and aggressive extremes are end members of a spectrum of gerrymandering strategies. One extreme creates stability at the expense of political power, the other extreme increases short-term political power at the expense of stability. Clearly, party A will maximize its political power if it gives itself a legislative majority large enough to allow it to effectively pursue its own policies, but not so large that it makes it excessively vulnerable to future changes in voting patterns. Gerrymandering is very much like playing the game Risk. In Risk, it often happens that a player who controls most of the map with a few armies in each country ultimately loses to a player who has spent the game amassing large armies in just a few countries.
The situation in a real place like Wisconsin obviously is more complicated than in the abstract example, but the fundamental dynamics are the same. When Republican lawyers sat down to redraw Wisconsin districts last year, their task was to create Republican majorities in as many districts as possible without diluting the Republican vote to the point where they would lose control of the legislature if the popular vote were as heavily Democratic as it was in 2008.
Republicans lawyers did a good job of protecting Republican legislators from the 2012 election, but will this advantage hold into the future? As shown in Figs 1 and 2, Republican Assembly members won their races by smaller margins in 2012 than in 2010, with a much larger fraction of them winning with 57% of the vote or less. At the same time, the number of vulnerable Democratic Assembly members dropped dramatically. Overall, Republicans went from being somewhat less vulnerable to catastrophic defeat than Democrats to being substantially more vulnerable. This is the expected and necessary consequence of retaining the same majority despite a 7% drop in share of the popular vote. The essential point is that disproportionate power comes at a price. Gerrymandering is a two-edged sword. A party cannot gerrymander itself into a larger share of the legislature than it has of the popular vote without increasing its risk of an electoral defeat that will leave it with a smaller share of the legislature than it has of the popular vote.
 Vote totals used here included some preliminary results, and may differ slightly from other vote totals being circulated. These differences are trivial and do not affect the analysis.
 Documents relating to gerrymandering released under court order show that Michael Best & Friedrich, the firm hired by Wisconsin Republicans to redraw districts, used the 2008 election as a benchmark of Democratic electoral success.
Figure 1: Victory margins (%) of Republican and Democratic Assembly victors in 2010 and 2010, and percent of all Republican and Democratic
Assembly members who won with 57% or less of the popular vote (“vulnerable” members).
In short, the 2011 gerrymandering ultimately could hurt Republicans. Democrats need to pick up 11 seats to take control of the Assembly. In the last election, 11 Republicans won by 3% or less. If the Democratic share of the popular vote shifts to 56% from 53%, there is a good chance that Democrats will win enough seats to take control. That is a large but not unrealistic shift, and it would still leave the Republicans with a larger share of Assembly seats than the popular vote.
However, if the popular vote shifted to 60% Democratic, Republicans likely would lose 25 seats, leaving them with 35% of Assembly seats while winning 40% of the popular vote. Under the 2010 districts, the same shift would cause the Republicans to lose only 15 seats, giving them 45% of the Assembly vote based on 40% of the popular vote. Thus the point at which the current gerrymandered districts will begin to work against Republicans will be reached when the Republican share of the popular vote drops to less than 40-44%.
The odds for Democrats getting 60% of the popular vote from the current Wisconsin electorate are low, although they will increase as demographic changes continue to erode the Republican base. However, there is one more factor that needs to be considered: incumbency. Incumbents have an inherent advantage in elections, especially in local elections like Assembly races, where there is little advertising and many voters go to the polls without even knowing the names or party affiliations of challenges to the incumbent. Gerrymandering means that many Republican Assembly members come from districts where Democrats run strong challenges to Republican incumbents. As these incumbents retire, Republicans will have a much harder time retaining their seats than they would have had with the 2010 districts.
Figure 2: Victory margins of Democratic and Republican Assembly members in the 2010 and 2012 elections. Note the large fraction of Republican seats that were won by relatively small margins in 2012.
The power of incumbency may currently be shielding the Wisconsin GOP from the negative consequences of its own gerrymandering. As shown in Fig 2., if seats won by incumbents with less than 60% of the vote are vulnerable to Democratic takeover, Republicans are in danger of losing to attrition not just their majority, but the disproportionate legislative power that they sought to expand through gerrymandering.
Even if Republicans can retain their Legislative majorities, it is not true that Republican legislators can advance divisive legislation without risk, as Rickert claims. Just the opposite. Republicans are more, not less, vulnerable to voter retaliation with the current districts than they were before. The current circumstances dictate a cautious, defensive strategy for the Republicans, along with the implementation of as many voter suppression measures as they can get away with in the courts.
Democrats, on the other hand, are operating from a secure base, and could benefit greatly from aggressively attacking Republicans. In particular, targeting vulnerable Assembly and Senate incumbents at the local level, where advertising is cheap and political advertising beyond yard signs is almost non-existent, could be a very effective way of accelerating the collapse of an over-extended GOP.