Sometimes a longer historical perspective can be useful. Electoral Calculus has records of opinion polls going back to 1983. These charts show the well-known stories of the Thatcher years, the rise and fall of New Labour, the "Get Brexit Done" election, and the Labour landslide of 2024.
But they also contain deeper insights about British politics – the popularity of the left and the right overall, and the importance of unity. These deep lessons are particularly relevant now, even as the surface of politics is in such flux.
The first chart shows a monthly poll-of-polls, taken over each month from June 1983 to December 2025. In it, we can see:
Back in 1983, the Conservatives had just won a landslide victory under Margaret Thatcher, and the opposition was split between Michael Foot's Labour Party and the new SDP-Liberal Alliance. The SDP itself, founded two year's earlier, was a breakaway party from Labour.
At the 1983 election, the Conservatives won with 42pc of the UK vote, Labour got 28pc, and the Alliance 25pc. The Conservatives were the largest party, and won the most seats, but the two centre-left parties together were more popular.
This is a theme that reoccurs over the last forty years.
Let's start by looking at the overall support for left-wing and right-wing parties from the public. The second graph shows the total support, in the monthly poll-of-polls, for the parties of the right (Conservative and Reform) and left (Labour, Lib Dem and Green) respectively.
The graph shows that the left parties are usually ahead of the right parties. In fact, the left parties are more popular for 89pc of the time, and the right parties are more popular for only 11pc of the time. The left parties were particularly popular from 1993 to 2003 (the peak Blair Years) and around the 2024 Labour landslide election victory. But the left was also more popular in the mid-eighties (the Thatcher years), and at other times where the right parties won general elections.
In fact, the left parties were only in government for one third (34pc) of this forty-two year period, and the right parties were in government for two-thirds of the time.
To see why this happened, we are going to borrow a scientific concept from information theory, called entropy. Entropy is a measure of the disorder or diversity in a system. In this case, it shows how divided or united a political side is. It's useful in a multi-party (aka multi-state) system to quantify the effective number of parties, giving less weight to smaller parties.
If there were, say, three equally popular parties on the left, then the effective number of parties would be three. But if one party were much more popular than the other two, then the effective number of parties would be close to one.
The third chart shows the entropy on the left and right respectively of British politics. The vertical axis represents the effective number of parties on either side.
On the left, we see that there have been effectively two parties since the formation of the SDP-Liberal Alliance. This declined a little under Blair, as Labour eclipsed the Lib Dems, but has increased to around three parties with the decline of Labour and the rise of the Greens.
On the right, there was only one party (the Conservatives) for a long time, except for the blip of the Referendum Party in 1997. But the popularity of Nigel Farage parties (UKIP, Brexit and then Reform) has increased the effective number of right parties to two by 2025.
A key insight from this chart, is that the left are usually more divided than the right. There are some exceptions, such as the push for Brexit in 2013-16, but often the left has about one more party than the right.
These divisions on the left are not a dull fact of purely academic interest. They are a key reason why the left can enjoy a plurality of popular support in the polls, but lose general elections. Under the first-past-the-post electoral system, a group is much more successful if it is united than divided.
The current outlook for the left in these terms is bleak. Not only are the left behind the right in total vote share – which is unusual, though it may be reversed. But they are more divided than ever with entropy showing that the left-wing vote is split fairly evenly three ways.
That division is more entrenched. The left needs to face some hard decisions if it is to avoid electoral catastrophe. Options which would help (in decreasing order of electoral effectiveness) are:Options 1 and 2 would be the most successful electorally, but are unlikely to happen. (See our electoral pact poll from 2021.) Option 3 is also difficult because local parties will resist standing down their candidates. Options 4 and 5 are much discussed currently but have smaller effects. Option 5 will have some effect, and we have already polled on tactical voting and included it in our predictions. It makes a difference in about seventy seats, but is not enough to stop Reform UK from leading the government.
The arc of the last forty years is that the left has usually had the numbers but the right has more often held power, because unity has outweighed raw popularity. The polling and entropy charts underline the same message. Whenever one party has managed to consolidate its side of the political spectrum, it has been rewarded with a long spell in office, even when its policies were less popular than its opponents. Today, the left faces another decision point, more fragmented than at any point in the polling record, with three parties competing for the same pool of voters. If that divide isn't resolved, the left could face another period in which it watches a more united right govern with or without a majority of public support.
We use the effective number of parties, also called the perplexity or first-order true diversity, as our measure. This is equal to the exponential of the Shannon information entropy, which is why we call this an entropy measure.
In symbols, suppose party $i$ is a party of the left (for $i=1,\ldots,m$). We define the relative support among the left for party $i$ as its share of national support divided by the overall support for all left-wing parties. We call this $p_i$.
The effective number of left-wing parties, is then defined as $$ E_L = \exp\left( -\sum_{i=1}^m p_i \log(p_i) \right). $$
The definition for the effective number of right parties is similar.
For example, in December 2025, Labour has 18.5pc support, the Lib Dems 12.9pc and the Greens 12.9pc. Then the total support for left parties is 44.3pc, and their relative supports are $$ p_{LAB}={18.5\%\over 44.3\%}=0.42,\quad p_{LIB}={12.9\%\over 44.3\%}=0.29,\quad p_{Green}={12.9\%\over 44.3\%}=0.29. $$
The effective number of left parties is then $$ E_L = \exp\left(-\left( 0.42\log(0.42) + 0.29\log(0.29) + 0.29\log(0.29) \right) \right) = 2.95. $$
So the three left-wing parties are split as much as 2.95 equally-popular parties. That is quite close to the maximum possible score of 3.00, indicating that the left is currently evenly split three ways. If 'Your Party' were included, the effective number of left-wing parties would be even higher, at around 3.4.
Similar calculations are performed on the right-wing parties. In December 2025, Reform had 28.0pc support in the polls and the Conservatives 19.2pc. That translates into total right support of 47.2pc, with relative supports of $$ p_{Ref}={28.0\%\over 47.2\%}=0.59,\quad p_{CON}={19.2\%\over 47.2\%}=0.41. $$
The effective number of right parties is then $$ E_R = \exp\left(-\left( 0.59\log(0.59) + 0.41\log(0.41) \right) \right) = 1.97, $$ which is also close to the maximum value of 2.0, showing that the right is split nearly exactly into two halves.