There was once the era when:
Every boy and every gal, That’s born into the world alive, Is either a little Liberal, Or else a little Conservative
By the 1950s and early 1960s there politics was basically the Conservatives and Labour. Yes, there might be a small group of Liberals, but really we had a two-party system.
So, to all intents and purposes you knew that if boundary changes moved a group of voters from constituency A to constituency B they would vote the same way regardless of which constituency they were in. No tactical voting.
When we draw up notional results, we make the assumption that people would vote the same way under proposed constituencies as they did in the real constituencies. But these days there is tactical voting and targetted seats.
To take the extreme example- the proposed seat of Brighton Pavilion which I looked at when I considered the proposed South East England constituencies.
The real seat of Brighton Pavilion was narrowly won from Labour by the Greens' Caroline Lucas. With the new proposed seats, Brighton Pavilion is altered by the transfer of two wards of Brighton & Hove City Council. Firstly, Regency is moved into an expanded version of the current Hove seat. And secondly, Moulsecoomb & Bevendean is picked up from the current Brighton Kemptown.
When we draw up notional results for Brighton Pavilion we effectively assume that if the proposed constituencies for the May 2015 general election had been in place for the May 2010 general election then the Greens there would have fought an odd campaign. In Hove they would be extensively focussing on just the Regency ward in a seat they wouldn't stand a chance of winning, while in Brighton Pavilion it'd be all hands to the recycled pumps- except for the Moulsecoomb & Bevendean ward, which could be the ward that makes the difference between a Labour MP and a Green MP.
There is another example, very locally, that is connected with tactical voting. And this is the constituency I live in, Southampton Test, which picks up one Southampton City Council ward (Swaythling) from Romsey & Southampton North to form a new Southampton Test, which is only 32 voters short of the maximum allowed.
The new Southampton Test is drawn from two very different constituencies. Romsey was a safe Conservative seat when created for the May 1997 general election, but the Liberal Demcorats won it in a by-election in May 2000, and it was a close Liberal Democrat/Conservative marginal at the June 2001 and May 2005 general elections. Probably many Labour-leaning voters went tactically for the Liberal Democrats to keep the Conservatives out.
Southampton Test is often considered a bellwether seat, oscillating between the Conservatives and Labour. However, in May 2003 the Liberal Democrats overtook Labour to be the largest party on Southampton City Council- a situation which remained until May 2006 when the Conservatives, Labour and the Liberal Democrats tied on 16 seats each. With the large student population, and a council led by the Liberal Democrats, at the 2005 general election I was seriously thinking Southampton Test would be another of those Liberal-Democrats-gain-from-third-place seats. At the very least I expected them to push the Conservatives into third place (which they nearly did) and to then pick up enough votes from disillusioned Labour supporters and tactically-voting Conservative supporters to go on and win Southampton Test at the next election (at the time, elections were normally 4 years apart, so I expected that one in June 2009 to coincide with the European Parliament ones).
But what about Swaythling? Well, drawing up notional results tells us that this ward would have brought in plenty of Conservative and Liberal Democrat voters, and not many Labour ones. But this is in part due to Romsey & Southampton North being a seat Labour can't dream of winning.
The notional results give a proposed Southampton Test that would have seen a smaller Labour majority over the Conservatives, and a larger Liberal Democrat vote, than the real Southampton Test. It is possible that there were Labour voters who might be natural Liberal Democrat supporters but voted tactically to keep the Conservatives out but might have voted Liberal Democrat if they felt that party was in with a chance of winning.
Suppose that it were indeed the proposed Southampton Test in existence at the 2010 general election, then what would be of use to the parties and voters? The notional results for the 2005 general election would be used.
Note that I am not talking about what the 2005 general election would have been if the 2010 constituencies had been used, but what the 2005 general election would have been if the 2015 constituencies had been used.
Colin Rallings and Michael Thrasher drew up the notional results of the 2005 general election based on the real 2010 constituencies. For Southampton Test they got:
For Romsey & Southampton North they got:
Remember these are the notional results of the 2005 general election if the current (not proposed) constituencies had been in place.
Now look at the actual result of Southampton Test of the 2010 general election;
The notional results for Southampton Test (the 2010 result projected onto the 2015 constituency) we have are:
The difference between the notional 2010 results and the real 2010 results are:
What does this difference represent? The only thing that has changed is the inclusion of Swaythling, so these are, as near as we can get them, how the good people of Swaythling voted in the 2010 general election.
The next step is to look at the actual result in Romsey & Southampton North (including Swaythling):
We can now calculate the percentage of each party's supporters who were in Swaythling, and then multiply that by the Rallings & Thrasher notional result for Romsey & Southampton North- this gives us the number of Swaythling voters for each party at the 2005 election:
This shows us something interesting about the Labour vote- just under one-third of the Labour vote in Romsey & Southampton North comes from Swaythling.
We now add these to the Rallings & Thrasher notional results for Southampton Test:
These are the 2005 general election results projected onto the proposed constituencies for the 2015 general election. Suppose that the 2010 general election had been fought on the proposed 2015 boundaries, then these would be the notional 2005 figures used. Every issue of Focus in the run-up to the election, every piece of campaign literature from David Callaghan, would urge Labour supporters disillusioned with the Government to vote for the second-place Liberal Democrats and would tell Conservatives that the best way to defeat Labour is to vote for the second-place Liberal Democrats.
My gut feeling is that if the 2010 general election had been held on the proposed 2015 constituency boundaries, then the Liberal Democrats would have done better in Southasmpton Yest, and the other two main parties worse, than the notional results give.
Maybe a way forward in producing more accurate notional results is to look, not at the most recent election, but the one before that, and draw up notional results (e.g. 2005 election on proposed 2015 boundaries) and somehow work forward from those.