Edexcel A-level Politics, Paper 3: Comparative Politics (USA) | also relevant to AQA Paper 2: Government and Politics of the USA
“If we do it, we will never lose the midterms, and we will never lose the general election, because we will have produced so many different things for our people, for the people, for the country, that it would be impossible to lose.”
President Donald Trump made this claim to Republican lawmakers in May 2026, months before voters went to the polls in the November 2026 midterm elections. It is a bold prediction to test, because the political science on midterm elections is unusually consistent: the president’s party has lost House seats in nearly every midterm election of the last 90 years, regardless of who is in the White House.
This page uses that historical pattern, the competing explanations for it, and the evidence available (as of July 2026) on the 2026 elections themselves, to test whether Trump’s prediction holds up, and to reach a reasoned judgement on how far midterm elections really are beyond a president’s control. We will re-visit Trump’s claim at the foot of this page and this page itself will be updated after the 2026 mid-terms.
1. What are midterm elections, and why do they matter?
Midterm elections take place two years into a president’s four-year term. All 435 seats in the House of Representatives and a third of the 100 Senate seats are contested, but the presidency itself is not on the ballot. This makes midterm elections, in effect, a national verdict on a sitting president delivered without the president standing for re-election.
Midterm elections matter constitutionally and politically. A president who loses control of one or both chambers of Congress faces divided government (a situation in which one party holds the presidency while the other controls one or both chambers of Congress), which can block their legislative agenda, block appointments to the federal judiciary and executive branch, and expose the administration to congressional investigation. The clearest example is 2018: Democrats retook the House, used their new subpoena and committee powers to investigate President Trump’s conduct, and voted to impeach him in December 2019, something that would have been constitutionally impossible without their midterm victory.
| Key terms Midterm elections: congressional elections held two years into a president’s four-year term, when all House seats and about a third of Senate seats are contested but the presidency is not. Coattail effect: the tendency of candidates from the president’s party to win extra seats because of the president’s own popularity at the top of the ticket in a presidential election year; midterms remove this effect entirely. Generic ballot: an opinion poll question asking voters which party, rather than which named candidate, they intend to support in their local congressional race, e.g. “Democrats 46%, Republicans 43%”; used as a rough national proxy for the House popular vote. Divided government: a situation in which one party controls the presidency while the other controls one or both chambers of Congress, a common outcome of midterm elections. |
| Check your understanding 1. Explain why the absence of the coattail effect makes midterm elections structurally different from presidential-year congressional elections. 2. Using the 2018 example, explain one way in which losing a midterm election can constrain a president even though the presidency itself was not contested. |
2. The historical pattern: the president’s party almost always loses
Since Franklin D. Roosevelt’s presidency, the president’s party has lost House seats in nearly every midterm election bar two: 1998 and 2002. Averaged across the 22 midterm elections between 1934 and 2018, the president’s party has lost 28 House seats and four Senate seats. The table below sets out the results since 1994, the point at which the modern, highly nationalised pattern of midterm politics (beginning with the Republican Revolution) took shape.
| 1994 | Clinton | Dem | −54 | −8 | The Republican Revolution, led by Newt Gingrich’s Contract with America; Republicans took both chambers for the first time since 1953. |
| 1998 | Clinton | Dem | +5 | 0 | First midterm gain for a president’s party since 1934. Backlash against Republican pursuit of the Lewinsky impeachment, combined with a strong economy and Clinton approval above 60%. |
| 2002 | G. W. Bush | Rep | +8 | +2 | Post-9/11 national security environment pushed Bush’s approval above 60%; only the second president since 1934 (after FDR) to gain seats in both chambers at a midterm. |
| 2006 | G. W. Bush | Rep | −31 | −6 | Iraq War unpopularity and Bush’s sixth year in office (the Six-year itch); Democrats retook both chambers, 232–203 in the House. |
| 2010 | Obama | Dem | −64 | −6 | Tea Party wave; backlash against the Affordable Care Act and the 2009 stimulus. Largest House loss for either party since 1948. |
| 2014 | Obama | Dem | −13 | −9 | Obama’s sixth year (the Six-year itch again); low approval amid the Ebola and ISIS news cycles; Republicans took the Senate. |
| 2018 | Trump | Rep | −40 | +2 | Referendum on Trump’s first two years; Democrats retook the House and later impeached Trump (December 2019). |
| 2022 | Biden | Dem | −9 | +1 | Smallest midterm loss for a president’s party in 20 years, widely attributed to backlash against the Supreme Court’s Dobbs v Jackson ruling overturning the constitutional right to abortion. |
Source: House and Senate seat changes, 1994–2022, compiled from Ballotpedia and Britannica election results.
The two genuine exceptions, 1998 and 2002, share a common feature: both occurred when presidential approval was unusually high (above 60%), driven by extraordinary national circumstances (impeachment and the aftermath of 11 September 2001 respectively) rather than by normal political conditions. This points to presidential approval, not the mere fact of being a midterm, as the underlying variable that really explains the pattern.
| Check your understanding 1. Identify and explain two features shared by the 1998 and 2002 midterms that made them exceptions to the usual pattern. 2. Using the table, compare the scale of losses in 2010 and 2022. What does the difference suggest about the relationship between a president’s approval rating and the size of their party’s midterm losses? |
3. Explaining the pattern: competing theories
Political scientists offer several, overlapping explanations for why the president’s party so reliably loses ground at midterms.
| Key terms Surge and decline theory: developed by the political scientist Angus Campbell, the idea that turnout and enthusiasm “surge” for the president’s party in a presidential election year because of the coattail effect, then “decline” at the following midterm as many less-engaged, short-term voters stay home, leaving a more oppositional electorate. Six-year itch: the tendency for a president’s party to suffer even heavier midterm losses in the sixth year of a presidency (the second midterm of a two-term presidency) than in the first midterm, as accumulated grievances and “time for a change” sentiment build. Incumbency advantage: the electoral benefit sitting members of Congress enjoy from name recognition, campaign funding and constituency casework, which normally makes midterms less volatile at the level of an individual district than the national swing alone would suggest. |
Evaluating the coattail effect
Because the Coattail effect is the mechanism that produces the initial “surge” in surge and decline theory, it is worth evaluating separately how strong that surge still is. Split-ticket voting, where a district backs a presidential candidate from one party and a House candidate from another, is the clearest evidence of coattails at work, and it has collapsed over the last fifty years.
| Year | Split-ticket districts | Share | Note |
| 1972 | 190 of 435 | 44% | High-water mark of candidate-centred, locally-focused congressional voting. |
| 1988 | 145 of 435 | 33% | Split-ticket voting still common before the nationalisation of American politics. |
| 2012 | 26 of 435 | 6% | Sharp fall as partisan polarisation hardens under Obama. |
| 2016 | 35 of 435 | 8% | Small uptick, but still historically low. |
| 2020 | 16 of 435 | 4% | Lowest share on record; near-total alignment between the presidential and House vote by district. |
| 2024 | 16 of 17 flipped districts matched the presidential winner | n/a | Trump’s coattails were, in Brookings’ phrase, “short”: Republican House gains came mainly from a collapse in Democratic turnout rather than new voters drawn in by Trump himself. |
Source: Pew Research Center (2016); Axios (2022); The Downballot and Brookings analysis of 2024 results.
Two patterns stand out, and they point in slightly different directions:
- The near-disappearance of split-ticket districts, from 44% in 1972 to under 4% in 2020, shows that voters increasingly treat their House vote as an extension of their presidential vote. This suggests that national partisan identity, rather than an individual president’s personal appeal, is now the main force aligning presidential and congressional results, a more structural explanation than the Coattail effect implies.
- Even so, the strength of that alignment does not mean presidents actively pull new voters down-ballot. In 2024, Republican House candidates gained relatively little from an increase in their own vote, and far more from a 6.6 million fall in the Democratic vote compared with 2020; ticket-splitting persisted at the margins, with 19.7% of Black voters who backed Trump nonetheless voting for the Democratic House candidate, against only 3.6% of Harris voters crossing over the other way.
- This matters directly for 2026: if the Republican “surge” in 2024 was already shallow, resting on Democratic turnout collapse rather than a genuine coattail pull towards Trump and congressional Republicans together, then the equivalent “decline” facing Republicans in 2026 may be smaller in raw vote-share terms than the historical pattern in Section 2 would suggest. This is cold comfort for Republicans in practice, however, because their 3-seat House majority leaves almost no cushion against any decline at all, however modest.
On balance, the coattail effect is a weaker explanation for 2026 than surge and decline theory or presidential approval. Split-ticket voting’s collapse shows that district-level results now track national partisanship closely, but the 2024 evidence suggests this is closer to a permanent, polarised alignment between two committed electorates than to presidents actively expanding their party’s support down the ballot, reinforcing why a small number of genuinely competitive, newly gerrymandered districts, rather than a broad coattail wave in either direction, are likely to decide control of the House in November 2026.
Referendum theory holds that voters use the midterm election to pass judgement on the president’s performance, even though the president is not on the ballot. A 2014 Gallup poll found that 52% of registered voters said their vote was intended to send Barack Obama a message, and of that group, around 32 percentage points wanted to send a message of opposition against only 20 of support, evidence that midterm voting behaves like a national verdict on the presidency rather than a series of local judgements on individual House members.
The strongest single predictor is presidential approval itself. Gallup’s analysis of midterms since 1946 found that presidents with approval below 50% saw their party lose an average of 37 House seats, compared with an average loss of only 14 seats for presidents above 50% approval, a stark illustration of why 1998 and 2002 (both above 60% approval) broke the usual pattern.
The Six-year itch is a more specific version of the same logic. Since the Second World War, the president’s party has lost an average of around 15 House seats at a first midterm but around 48 at a second-term (sixth-year) midterm; the pattern is clearer in the Senate than in the House, and some political scientists question whether it is a distinct effect at all rather than simply presidents’ approval ratings naturally eroding over two terms. This matters for 2026: because Trump’s two terms were non-consecutive (2017–2021, then 2025–), November 2026 is technically his first midterm of this presidency, analogous to 2018 rather than to a classic sixth-year itch election, even though six years have now passed since he first took office.
| Check your understanding 1. Explain, in your own words, the difference between surge and decline theory and referendum theory as explanations for midterm losses. 2. Using the split-ticket voting data, evaluate whether the Coattail effect or partisan polarisation better explains why presidential and House results have converged since 1972. 3. Evaluate the claim that presidential approval rating is the single most important factor in explaining the scale of a president’s party’s midterm losses. Use figures from this section to support your answer. |
4. Turnout: who actually shows up?
Part of the explanation for midterm volatility lies in who turns out to vote. Midterm turnout is consistently far lower than presidential-year turnout, and the electorate that does turn out tends to be older, more partisan and more likely to be motivated by opposition to the sitting president.
| Year | Election type | Turnout | Notes |
| 2018 | Midterm | 49% | Highest midterm turnout since 1914, driven by high engagement on both sides in the first Trump-era midterm. |
| 2020 | Presidential | 66% | Highest turnout rate for any US national election since 1900. |
| 2022 | Midterm | 46% | Exceeded every midterm since 1970, though still 20 percentage points below the 2020 presidential turnout. |
| 2024 | Presidential | 64% | Second-highest presidential turnout on record, tied with 1960. |
Source: Pew Research Center, “Voter turnout in US elections, 2018–2022” and “Voter turnout in the 2020 and 2024 elections” (2025).
Even accounting for the unusually high turnout of the 2018 and 2022 midterms, the pattern holds: average presidential-year turnout from 1972 to 2024 was around 58%, against average midterm turnout of around 41%, a 17-point gap. Because the voters who drop off between a presidential year and the following midterm are disproportionately less partisan, low-propensity voters, the electorate that remains is skewed towards the more committed and often more oppositional voters of the party out of power, reinforcing the losses predicted by surge and decline theory.
5. Gerrymandering, redistricting and the 2026 map
One complicating factor for 2026 is that the electoral map itself has changed since the last election. Congressional districts are normally redrawn only once a decade, after the census, but a wave of mid-decade redistricting (redrawing outside the normal ten-year cycle) began in summer 2025 when Trump asked Texas Republicans to redraw their map to create additional Republican-leaning seats.
This is constitutionally possible because of the Supreme Court’s 2019 ruling in Rucho v. Common Cause, which held that claims of excessive partisan gerrymandering (the deliberate drawing of district boundaries to advantage one party) present a “political question” beyond the reach of the federal courts, leaving state legislatures largely free to gerrymander for partisan advantage so long as they do not breach separate protections against racial discrimination.
| State | Driving party | Projected impact | Status |
| Texas | Republican | Up to 5 seats | New map passed by the legislature in 2025, targeting Democratic-held districts; subject to ongoing federal litigation. |
| Missouri | Republican | 1 seat | New map passed in autumn 2025 as part of the same push. |
| North Carolina | Republican | 1–2 seats | New map passed October 2025. |
| Ohio | Republican | 1–2 seats | New map approved by the state redistricting commission in October 2025. |
| California | Democrat | Up to 5 seats | Counter-map approved by California voters in a November 2025 ballot measure, in direct response to the Texas move. |
| Utah | Democrat | 1 seat | Court-ordered redraw produced a more competitive map. |
| Virginia | Democrat | Up to 3 seats (pending) | New Democratic-favourable map awaiting final approval as of mid-2026. |
Source: Ballotpedia, “Redistricting ahead of the 2026 elections”; Democracy Docket redistricting tracker (2026).
Nationally, this leaves Republican-led states pursuing roughly nine additional Republican-leaning seats (concentrated in Texas, Missouri, North Carolina and Ohio) against a Democratic counter-push of a similar magnitude concentrated in California, Utah and potentially Virginia. The practical effect is that, for the first time in the modern redistricting era, both parties have partially redrawn the battlefield mid-cycle in direct response to each other, meaning the 2026 result will reflect not only the underlying national mood but also which side’s redistricting gamble pays off in court and at the ballot box.
Gerrymandering is treated here only as it affects the 2026 midterm result. A fuller evaluation of gerrymandering as a question of representation, including how well Congress reflects the electorate and the case for reform, follows in the later study notes on the functions of Congress.
| Check your understanding 1. Explain why Rucho v. Common Cause made the 2025–2026 mid-decade redistricting wave constitutionally possible. 2. Using the table, evaluate whether the redistricting battle is likely to help or hinder the Republican Party’s chances of retaining the House in 2026. |
6. Case study: the November 2026 midterms
Going into the 2026 midterms, Republicans hold a narrow 218–214 House majority (with three vacancies) and a 53–47 Senate majority (counting the two independents who caucus with Democrats). This means Democrats need a net gain of only three House seats and four Senate seats to win control of each chamber respectively, while Republicans can afford to lose no more than two seats in either chamber.
| Indicator | Figure | As of | Interpretation |
| Trump approval | 38–40% | late July–August 2026 (Gallup, Silver Bulletin) | Firmly in the “below 50%” band that Gallup associates with an average 37-seat House loss for the president’s party. |
| Generic ballot | Democrats +3 to +7 | July–August 2026 (Decision Desk HQ, other trackers) | Consistent, if variable, Democratic lead; comparable in scale to the leads seen before the 2018 and 2006 Democratic wave years. |
| 2026 shutdowns | Two shutdowns, Jan–Feb and Feb–Apr 2026 | resolved by 30 April 2026 | Both stemmed from disputes over federal immigration enforcement following the killing of Alex Pretti by Customs and Border Protection agents; contributed to a decline in Trump’s approval. |
| Forecast models | Democratic gain of roughly 20–35+ House seats; Senate a toss-up leaning Republican | mid-2026 model projections (USPollingData.com) | A modelled projection, not a certainty; the Senate map (which states are up) structurally favours Republicans regardless of national mood. |
Source: Gallup, Silver Bulletin, Decision Desk HQ and USPollingData.com polling and forecast trackers, accessed 1 August 2026; figures are subject to change before polling day in November 2026.
It is worth being precise about what this evidence can and cannot show in August 2026.
Approval ratings and generic ballot polling are strong historical predictors of the direction and rough scale of midterm losses, but they are not the result itself, and forecast models built from them carry real uncertainty, particularly for the Senate, where the specific seats being contested in any given cycle can matter more than the national mood.
What the evidence does show clearly is that every major indicator, approval, the generic ballot, and the precedent of two damaging government shutdowns, points in the same direction: towards a difficult midterm for the president’s party, consistent with the historical pattern rather than an exception to it.
| Check your understanding 1. Using the table, explain why the Senate is described as a “toss-up leaning Republican” even though Trump’s approval rating is low. 2. “Polling evidence in August 2026 makes a Republican defeat in November 2026 a certainty.” To what extent do you agree with this statement? Use evidence from this section in your answer. |
7. Evaluation: how far are midterms beyond a president’s control?
The evidence points overwhelmingly in one direction. In 80 years of midterm elections since Franklin D. Roosevelt, only two presidents (Clinton in 1998 and Bush in 2002) have avoided losing House seats, and both did so only because of extraordinary, short-lived circumstances that pushed their approval above 60%. Trump’s approval in 2026, at 38–40%, sits squarely within Gallup’s “unpopular president” band, associated historically with an average House loss of 37 seats, more than enough to end Republican control of the chamber given their three-seat margin.
The Democratic generic ballot lead of 3–7 points in mid-2026 is consistent with, rather than contradicting, this historical pattern; leads of this size preceded both the 2006 and 2018 Democratic wave elections. Set against this, the Six-year itch literature and the two 2026 government shutdowns both point towards accumulated, rather than fresh, discontent, exactly the conditions historically associated with larger, not smaller, midterm losses.
The genuine complicating factor is redistricting. Because roughly nine Republican-leaning seats have been engineered in Texas, Missouri, North Carolina and Ohio, partly offset by a similar number of Democratic-leaning seats in California, Utah and potentially Virginia, the relationship between national vote share and seats won is less predictable in 2026 than in a normal, map-stable midterm such as 2010 or 2018.
This means that even if Democrats win the national popular vote for the House by a margin similar to 2018 (roughly 8–9 points), the number of seats that changes hands could diverge more than the historical pattern alone would predict, in either direction, depending on which side’s redrawn maps hold up in court and at the ballot box.
Weighing this evidence, Trump’s claim that Republicans “will never lose the midterms” in 2026 runs directly against 80 years of precedent, an approval rating in the range historically associated with a roughly 37-seat loss, and a consistent Democratic generic ballot lead. It is very difficult to reconcile with the evidence available in August 2026. What is genuinely uncertain, and what the historical pattern alone cannot settle, is not whether Republicans will lose ground but how much ground they will lose, given that the 2025–2026 redistricting wave means the 2026 map is not the same map that produced the historical pattern in the first place.
The nearest UK equivalent, using local and by-elections as a barometer of a governing party’s mid-term popularity (for example the May 2025 local elections), is developed in more detail elsewhere on histpology.co.uk.
Where this page began . . .
“If we do it, we will never lose the midterms, and we will never lose the general election, because we will have produced so many different things for our people, for the people, for the country, that it would be impossible to lose.”
Tested against the evidence available as this page was written, in early August 2026, Trump’s May 2026 claim does not hold: historical precedent, Trump’s own approval rating, and a consistent Democratic generic ballot lead all point towards Republican losses in November 2026, in line with the pattern that has held in 36 of the last 38 midterm elections.
The genuinely open question, deliberately left unresolved by a single opinion poll or historical average, is whether the unprecedented mid-decade redistricting fight of 2025–2026 will cushion the scale of those losses enough to let Republicans cling on to one or both chambers of Congress, a live test of the historical pattern that only the November 2026 results themselves can finally settle.
| Check your understanding 1. Drawing on the whole document, construct a two-paragraph answer to the question: “Explain and evaluate the extent to which US congressional midterm elections are a referendum on the sitting president.” 2. Identify one piece of evidence in this document that supports Trump’s May 2026 prediction, and one that contradicts it. Which do you find more persuasive, and why? 3. To what extent has mid-decade redistricting weakened the usefulness of historical midterm patterns as a predictive tool? Use specific figures from Section 5 in your answer. |
Sources
Primary sources
Rucho v. Common Cause, 588 U.S. 684 (2019).
Secondary sources
Axios, “The number of districts that split tickets for Congress and president is plummeting” (2022).
Ballotpedia, “Redistricting ahead of the 2026 elections” (2026).
Ballotpedia, “United States Congress elections, 2026” (2026).
Britannica, “United States midterm elections” (2026).
Brookings, “President-elect Trump’s short coattails” (2024).
Congressional Research Service, “The 2025 (FY2026) Government Shutdown: Economic Effects”, Congress.gov, R48832 (2026).
Decision Desk HQ, “Where the polls stand as Decision Desk HQ launches its 2026 forecast” (2026).
Democracy Docket, “Live Redistricting Tracker” (2026).
Gallup, “Midterm Seat Loss Averages 37 for Unpopular Presidents” (2018).
Pew Research Center, “Split-ticket districts, once common, are now rare” (2016).
Pew Research Center, “Voter turnout in US elections, 2018–2022” (2023).
Pew Research Center, “Voter turnout in the 2020 and 2024 elections: A detailed analysis” (2025).
The Conversation, “For 80 years, the president’s party has almost always lost House seats in midterm elections…” (2026).
The Downballot, “The Downballot’s 2024 presidential results for all 435 House districts” (2024).
USPollingData.com, “2026 Midterm Elections Forecast” (2026).