Six predicted questions with the reasons behind each pick, a plan, the concepts and examples to know, and practice. Everything stays folded until you open it.
We rebuilt the predictions from the full 2019 to 2025 question history with three rules. The picks copy the short, direct style of the 2024 and 2025 papers. Topics are weighted by recency: anything examined in 2025 is treated as very unlikely to return, 2024 topics as unlikely, and so on back through the years. And where two topics were equally credible we chose the one most worth revising anyway, because its material works across the widest range of questions.
For US, that pushes out federalism and Congress against the President (both 2025) and the Supreme Court, which carried a 30 marker three years running from 2023 to 2025, and pulls in elections (the Electoral College last ran in 2020), parties (no straight essay in the whole real series) and the amendment process (never asked).
The President's powers are codified in Article II and capped by the 22nd Amendment, whereas the PM's rest on uncodified convention and prerogative, stretching or shrinking with circumstance.
Separation of powers means the President sits outside Congress, cannot introduce bills and bargains for every vote, whereas fusion of powers puts the PM inside Parliament, normally commanding a disciplined majority and the timetable.
The President serves a fixed term and can only be removed by impeachment, whereas a PM can fall in days to their own party or the Commons: two PMs were removed by their parties in 2022 alone.
| Separation v fusion | The structural difference that explains most of the rest |
| Executive orders | Presidential rule-making without Congress, reversible by the next President or the courts |
| Royal prerogative | PM powers exercised without statute, including patronage |
| Examples | Article II; 22nd Amendment (1951); Johnson and Truss removed by their own party (2022); veto power against the PM's control of the Commons timetable |
In both systems groups seek influence without seeking office, and the types map onto each other: sectional groups (US Chamber of Commerce, AMA; CBI, BMA) and promotional groups (ACLU, Sierra Club; Liberty, RSPB).
Both lobby the executive and legislature, mobilise members, run media campaigns and, importantly, both use the courts: the NAACP litigated Brown, just as Liberty and campaigners have used judicial review against UK governments.
In both systems money and expertise buy access: insider groups enjoy routine consultation (the revolving door in Washington, charter status and Whitehall consultation in the UK), while outsider groups protest from the street.
| Sectional v promotional | Member interest against cause, in both countries |
| Insider v outsider | Access through consultation against pressure through protest |
| Judicial route | Litigation as group strategy in both systems |
| Examples | US Chamber of Commerce and CBI; AMA and BMA; ACLU and Liberty; NAACP in Brown v Board (1954); UK judicial review campaigns |
US parties are decentralised, fifty state parties with national committees that matter mainly at election time, whereas UK parties are centralised machines with permanent national headquarters, a single leader and central control of the brand.
US parties cannot reliably whip: members answer to primary voters, not leaders, whereas UK whips control careers and can suspend rebels, so cohesion is structurally enforced.
Primaries hand US candidate selection to voters, weakening the party's gatekeeping, whereas UK parties select and deselect their own candidates centrally or locally under party rules.
| Structural theory | Institutions shape behaviour: separation against fusion |
| Rational theory | Politicians follow incentives: primary voters against whips |
| Cultural theory | Shared ideas shape behaviour: candidate-centred against party-centred politics |
| Examples | Primary challenges driving US behaviour; the UK whip system and suspensions; McCarthy needing 15 ballots to become Speaker (2023) as discipline failure |
FORTwice in living memory the College has installed the popular vote loser: 2000 (decided by 537 votes in Florida) and 2016 (Clinton won by nearly three million). Even 2020 sat 43,000 votes across three states from inversion.
AGAINSTIn most elections College and country agree, and 2024 showed it: Trump won 312 to 226 and the popular vote together. The misfires are rare exceptions.
JUDGEMENTA system that only usually agrees with the voters is a system that cannot be trusted to, and two failures in six cycles is not rare.
FORIn 2024 the campaign was fought almost entirely in seven battleground states; voters in safe states were spectators in their own election, ignored by visits, spending and policy.
AGAINSTDefenders argue it forces geographically broad coalitions and stops candidates camping in the big cities, protecting smaller states as the Founders intended.
JUDGEMENTSwapping rule by big states for rule by swing states protects nobody; it just moves the distortion.
FORA Wyoming voter carries several times the College weight of a Californian, and winner-takes-all in 48 states erases every losing voter in the state: millions of Republican votes in California count for nothing.
AGAINSTThe College mirrors the federal design, states as units, exactly as the Senate does, and states can reform themselves: Maine and Nebraska already split their votes by district.
JUDGEMENTFederalism explains the design; it does not justify unequal votes in the one election the whole nation shares.
FORThe machinery itself has become an attack surface: faithless electors in 2016, and the certification process targeted on 6 January 2021.
AGAINSTCongress answered with the Electoral Count Reform Act 2022, and no agreed replacement exists: the National Popular Vote compact has stalled at 209 electoral votes, short of the 270 it needs.
JUDGEMENTPatching the pipework after an attempted breach concedes the point: the system is a vulnerability, not a safeguard.
| Electoral College | 538 electors; 270 to win; chosen state by state |
| Winner takes all | 48 states give every elector to the state's plurality winner |
| Faithless elector | An elector voting against their state's result (seven in 2016) |
| NPVIC | Interstate compact to award member states' electors to the national winner; 209 electoral votes signed up |
| ECRA 2022 | Reform that clarified certification and the VP's purely ceremonial role |
| Key examples | Bush v Gore 2000 (537 votes); 2016 inversion (minus 2.9 million); 2024: Trump 312 to 226 and the popular vote; 6 January 2021; Maine and Nebraska district method (the NE-2 blue dot) |
FORTwo thirds of both chambers plus 38 states is unreachable when the parties are polarised: thirteen small states can block anything, and no substantive amendment has passed in over half a century.
AGAINSTDifficulty is the design, not a defect: Madison built a high bar to protect the system from transient majorities, and stability is the reward.
JUDGEMENTA bar set high in 1787 has become a wall in a 50-50 nation; protection has hardened into paralysis.
FORThe ERA passed Congress in 1972 and still is not in the Constitution; DC residents remain without voting representation; the 27th Amendment took 203 years.
AGAINSTWhen consensus is real, amendment is fast: the 26th Amendment was ratified in about 100 days in 1971. The process works; the consensus is what is missing.
JUDGEMENTIf the process only works when nobody disagrees, it is not a working process for a divided country.
FORBecause Article V is blocked, nine unelected justices do the amending: Brown, Roe, Obergefell and then Dobbs rewrote constitutional rights in both directions without a single ratifying vote.
AGAINSTInterpretation is flexibility: a living reading lets an eighteenth century text govern a modern state without constant formal change.
JUDGEMENTDobbs proves the cost: rights that swing with the Court's membership are evidence the formal channel has failed, not that it is unneeded.
FORChange is forced down to the states, producing a patchwork: abortion law now varies wildly state by state, which is constitutional inconsistency, not constitutional change.
AGAINSTState constitutions amend easily and often, so federalism supplies the adaptability the national document lacks, including the state abortion ballot measures since 2022.
JUDGEMENTFifty answers to one constitutional question is the clearest sign the national amendment process no longer functions.
| Article V | Two thirds of Congress proposes; 38 states ratify |
| Substantive amendment | Real constitutional change, none since the 26th (1971) |
| Interpretive amendment | The Court changing meaning without changing text |
| Polarisation | The modern blocker: no cross-party supermajorities exist |
| Key examples | 26th Amendment 1971; 27th Amendment 1992 (proposed 1789); ERA saga 1972 to 2024; DC Voting Rights Amendment failure; Brown 1954; Roe 1973; Obergefell 2015; Dobbs 2022; state abortion measures 2022 to 2024 |
FOR (decline)Primaries stripped parties of their core power: Trump took the 2016 nomination against the party establishment, and primary challenges now discipline politicians more than leaders do.
AGAINST2024 showed the party machine alive: when Biden withdrew, Democratic elites coordinated behind Harris within days, no contested process, and endorsements and money still police most races.
JUDGEMENTSelection power has weakened, but 2024 proves the party can still act decisively as a collective when it matters.
FOR (decline)Since Citizens United (2010), super PACs and candidate operations raise and spend outside party control, making campaigns candidate-centred.
AGAINSTThe national committees, joint fundraising and the party-aligned platforms (ActBlue, WinRed) move enormous sums, and ground operations still run through party infrastructure.
JUDGEMENTThe money flows around the parties but overwhelmingly along party lines; the network is the party in modern form.
AGAINST (renewal)Party identification is the single best predictor of the vote, party-line voting in elections runs at historic highs, and ticket-splitting has collapsed: barely a handful of districts now split their House and presidential votes.
FOR (decline)Rising numbers self-describe as independents and trust in both parties is low.
JUDGEMENTMost "independents" vote like loyal partisans; on the measure that matters, behaviour, the party label has rarely been stronger.
FOR (decline)Factions humiliate leaders: McCarthy needed 15 ballots to become Speaker in January 2023 and was removed by his own side that October.
AGAINSTSet against the mid twentieth century, when both parties were ideological coalitions that crossed lines constantly, today's parties vote with near-perfect unity on major legislation.
JUDGEMENTNoisy factions inside historically disciplined parties are a sign of life, not decline.
| Party decline thesis | Parties losing selection, money and voter loyalty (1970s to 1990s account) |
| Party renewal | Polarised, disciplined, nationalised parties since the 2000s |
| Invisible primary | Elite endorsements and money before any vote is cast |
| Ticket splitting | Voting different parties on one ballot, now near historic lows |
| Polarisation | The engine of renewal: distinct, cohesive, hostile parties |
| Key examples | Trump 2016 hostile takeover; Biden to Harris handover July 2024; McCarthy 15 ballots and removal 2023; Citizens United 2010; record party-line voting and the shrinking set of split districts after 2024 |