This story has been updated to incorporate the results of the October 10 primary. 

THE HEADLINES AFTER the 2022 state election did not look good for the Massachusetts Republican Party. The GOP lost all nine Congressional races and every statewide office, while seeing their already anemic representation in the Legislature erode even further, with the party now holding only 25 of 160 seats in the House and just three seats in the 40-member Senate. While Republicans increasingly look like an endangered species in Massachusetts, a closer look at legislative elections over the last 15 years leads to a more nuanced interpretation of the state’s partisan leanings — and a special election for state Senate this fall may serve to illustrate that point. 

A large number of Massachusetts voters are still willing to support a certain brand of Republican in statewide elections (read: Charlie Baker) and there continue to be conservative cities, towns, and legislative districts in Massachusetts that consistently elect Republicans to office. 

The power of incumbency can also mask a district’s potential for partisan competition. Districts with a history of sticking with a Republican or Democratic incumbent year after year can end up being surprisingly competitive when the incumbent steps down. 

That is exactly the situation that could unfold in the coming weeks in a special election to fill a vacant state Senate seat in Central Massachusetts. 

In May, longtime state senator Anne Gobi, a Democrat from Spencer, announced she would be stepping down from the Legislature to become the first director of rural affairs for the Commonwealth. Her resignation has triggered a special election for the Central Massachusetts-based Worcester and Hampshire District, with the primary taking place on October 10 and the general election scheduled for November 7. 

Gobi’s five-term tenure was part of a long run of moderate Democrats holding the Senate seat that goes back nearly 50 years, starting with the election of Robert Wetmore in 1976, followed by Gobi’s predecessor, Stephen Brewer,  who was first elected in 1996. Despite that long hold on the seat, an analysis of voting patterns and other predictive election variables shows that the looming special election represents a rare case where Republicans can flip a seat that has long been in Democratic hands. 

The district is made up of the western side of the city of Worcester, 20 other small Worcester County municipalities, and the Hampshire County town of Ware. After easily winning a contested Republican primary, state Rep. Peter Durant of Spencer will now face-off in November against state Rep. Jon Zlotnik of Gardner, who ran unopposed in the Democratic primary. 

Democrat Anne Gobi’s departure from the Senate has created a rare opportunity for Massachusetts Republicans to flip a legislative seat. (Photo by Andy Metzger)

An important factor to consider in assessing how competitive a race is likely to be between a Democrat and Republican in a legislative race is how votes  have split between the Democrat and Republican in other well-understood matchups. 

A widely used metric for making that determination is the partisan voter index. The PVI, introduced in the 1990s by the Cook Political Report, looks at the share of the vote won in a district by the Democratic and Republican presidential nominees in the last two presidential elections in comparison to their shares of the overall national vote. 

The Worcester and Hampshire Senate district has a partisan voter index of EVEN, meaning that its voters have given approximately the same proportion of their votes to Republican and Democratic presidential candidates as the country as a whole when averaged over the last two presidential elections. (Districts with clear partisan leanings are identified by the degree to which party presidential nominees overperformed their national vote share. A district identified as D+4, for example, is one where the Democratic presidential nominees won, on average over the last two elections, 4 percentage points more of the district vote than they did in the US popular vote total. An R+4 district indicates the same level of swing toward Republican presidential nominees.)

The interactive map below shows the 22 communities in the Worcester and Hampshire Senate district. Hover over the map to see the name of each community; click on a community to see its partisan voter index, the size of its citizen voting-age population, and what share of the Senate district it represents.

While the Massachusetts Republican Party has suffered from leadership woes over the last several years, Republican state legislative candidates who are on the same page as district voters continue to be able to win open seats and, once elected, gain a large incumbency advantage. 

The 2022 election cycle in Massachusetts produced two examples of Republican competitiveness in open-seat races for the Legislature without an incumbent, one in an evenly-balanced district and another that showed GOP competitiveness even in a Democratically-leaning district. 

In the race for the 9th Norfolk House seat, Democrats hoped to pick up the seat that had been held by Shawn Dooley, a Wrentham Republican who left to mount an unsuccessful challenge to incumbent Needham Democrat Becca Rausch for a state Senate seat. Though the 9th Norfolk is a D+6 district, Republican Marcus Vaughn, who received strong support from outgoing incumbent Dooley, prevailed over Democrat Kevin Kalkut by 360 votes. 

In the 1st Middlesex District, a D+2 district that had been represented by Republican Sheila Harrington, who did not seek reelection, Margaret Scarsdale, a Democrat from Pepperell, won a nail-biter against Republican Andrew Shepherd of Townsend by just seven votes.


While every election is different, there are fundamental attributes of each race’s district, candidates, and election context that can be used to model the broad contours of elections using historical data. For Massachusetts state legislative elections, a useful model can be created using the district’s partisan voter index, the incumbency status of the race (GOP incumbent, no incumbent, or Democratic incumbent), and whether the election coincides with a presidential election (with Massachusetts legislators elected to two-year terms, legislative elections alternate between presidential and non-presidential election years, while special elections for legislative seats never coincide with a presidential vote).

These three variables can serve as inputs to a model used to evaluate possible election outcomes using a statistical approach called Bayesian regression.  (See here for more details on the model.) When the model is applied to 521 contested general elections results for Massachusetts state representative and state Senate seats from 2008 through early 2023, it explains approximately 70 percent of the variance in the Democratic or Republican margin of victory. That makes it a useful tool, both for estimating the range of likely outcomes for upcoming elections and also to understand the impact and importance of the individual partisan lean, incumbency, and presidential election year variables.  

The following table gives the model’s estimate of how the margin of victory in a race is affected by each of the three variables.

       Variable                                                                 Margin Change

Partisan voter index                                        1 (each PVI point is roughly equivalent to 1 margin point)

GOP incumbent                                              11 (Republican incumbency advantage)

Democratic incumbent                                   6 (Democratic incumbency advantage) 

Presidential election                                       4 (Advantage for Democratic legislative candidate)


Incumbency provides a significant advantage in Massachusetts legislative elections, especially for Republicans. Our model for estimating Democratic/GOP margins shows that, on average, Democratic incumbency is worth 6 percentage points compared to competing as a new candidate in an open race, while Republican incumbency is worth 11 percentage points, all other factors being held equal. 

In the 2022 state legislative elections, 12 of the 13 incumbent Republican state senators and representatives who faced a general election opponent won their races. The remaining incumbent legislator, state Rep. Leonard Mirra of Georgetown, was defeated by a single vote in the D+9 2nd Essex District.  

A look at contested legislative races over the last 15 years gives a sense of just how powerful the combination of incumbency and partisan voter index can be in predicting election outcomes. Since 2008, there have been 52 contested Massachusetts legislative races with a Republican incumbent running in a district with a PVI of EVEN or Republican-leaning. The GOP incumbent won every race but one. 

The single outlier was the 2018 defeat of recent Massachusetts Republican Party chairman Jim Lyons by Democrat Tram Nguyen in the Andover-based 18th Essex District, which has an even partisan index. The affluent suburban district was not a good match for a Trump-supporting, culture warrior Republican like Lyons. 

The incumbency advantage makes clear the opportunity and/or risk for the parties when an incumbent moves on. The absence of an incumbent in a race drastically changes the partisan landscape. 


The legislative election model shows that competing in an election that coincides with a presidential election gives an additional advantage of about 4 points to Massachusetts Democratic legislative candidates in competitive races. It has been a consistent finding in US political science research that voter turnout tends to be higher in presidential election years. The research is less definitive on whether higher turnout is consistently and reliably better for Democratic candidates across geographies and political offices, but this is likely to be one of the reasons for the 4 point Democratic advantage for legislative candidates in deep-blue Massachusetts, which was last won by a Republican presidential nominee nearly 40 years ago, when Ronald Reagan carried the state in 1984.

No single example of the impact of presidential vs. non-presidential election years makes the case as consistently as the broad survey of the 521 contested elections from 2008 to 2023. A notable example of this dynamic, however, is the November 2008 election, which saw gains for Democrats in the Massachusetts Legislature in the high turnout Obama presidential election. 

The 2008 election brought in a class of new legislators noted for its age, gender, and racial diversity. An unusually large number of the newly-elected Democratic legislators from 2008, however, lost their seats to Republicans two years later in the non-presidential 2010 election. The dynamics of the 2010 election went beyond simple turnout differences—2010 has been characterized as the Tea Party backlash to progressive wins in 2008. (Disclosure: my wife, Jennifer Benson, was one of the Democratic legislators from the class of 2008 who were able to withstand the 2010 GOP wave.)


The largest set of eligible voters in the upcoming special election for Senate comes from Democratic hopeful Jon Zlotnik’s home of Gardner, with 13 percent of the district’s electorate. Republican Peter Durant’s home town of Spencer accounts for 8 percent of the eligible voters. 

While the overall partisan lean of the district is EVEN, the lean of individual cities and towns that make up the Worcester and Hampshire Senate district ranges from R+10 East Brookfield and Phillipston to the the precincts in the City of Worcester, which stands at D+14. 

As for the other variables in the election model, as an open-seat race, there is no incumbency advantage for either party to claim in the three-part model. Finally, as a special election, the race does not bring the Democratic advantage that the model assigns to legislative contests taking place during presidential elections. 

The historical data from more than 500 legislative races from 2008 to 2023 show that a Republican stands some chance in an open race with no incumbent against a Democrat even in D+14 district in a non-presidential year and in a D+9 district in a presidential year. The odds increase, of course, as the partisan voter index moves towards EVEN and crosses to Republican-leaning districts. A race in an even district like the Worcester and Hampshire Senate district – with no incumbent, in a non-presidential cycle – is a true tossup from a historical point of view.   

The three variable model can be used to forecast the most likely outcome of the Worcester and Hampshire special election – a close one or two point win by either party. But the model also allows us to run simulations to see the likelihood of other outcomes, with wider victory margins, that might occur due to factors not measured by the model’s three variables—candidate quality and fundraising, for example—and inherent randomness and uncertainty in the world.

The sampling process can be thought of as doing a set of random walks through a hilly landscape where the height of the hills correspond to higher probabilities, the goal being to find the highest point. Each random walk results in a single simulated election outcome and the search will end up finding more likely outcomes more often. 

It may seem strange to take a roundabout random-walk approach rather than simply reporting the single most likely outcome. But the real world is complicated and unpredictable and the random walk approach has been shown to do a better job of understanding a range of likely outcomes and the degree of certainty or uncertainty to the outcome. The uncertainty has been calibrated into the model based on the previous election outcomes and their corresponding variable values.

By using the sample process we get the most common/most likely outcome of a close win by either party. But this approach also estimates the chances of something other than a tight race. This analysis finds that there is an 85 percent chance that the winning margin in the special election will be in the single digits and a less than 1 percent chance that the margin will be greater than 20 points.

There are many other important factors in a state legislative race beyond partisan lean, incumbency status, and whether it coincides with a presidential election, including fundraising, candidate quality, voter outreach efforts, and media coverage. Given that the fundamentals point to an evenly split race, these other variables will likely be the determining factors in the Worcester and Hampshire state Senate race.

As a Democratic incumbent running for reelection last November, Gobi equaled the model’s expectations with a comfortable 6 point win over Republican James Amorello of Holden, but Gobi’s exit from the Legislature has turned this special election into a wide-open race. Democrats run the risk of losing a legislative seat and accompanying Democratic advocacy in the heart of Central Massachusetts, while the Mass. GOP,  with barely a presence in the state Senate, has a real opportunity to flip a seat and increase its  influence in the middle of the state.

Brent Benson has been analyzing Massachusetts elections, politics, and policy through a quantitative lens on his Mass. Numbers blog and other news outlets since 2012. Follow him on X @bwbensonjr