In the last Holyrood election in 2011, the SNP swept to majority government with 45% of the constituency vote.

On September 18, 2014, 45% of Scots voted Yes in the referendum.

But these were not the same 45%.

There was overlap, of course, but it was far from complete. There were already clear signs of this on referendum night, when SNP strongholds like Moray gave a resounding 'No' while West Dunbartonshire, one of the few places not airbrushed yellow in 2011, went Yes. Data from the ESRC-funded Scottish Referendum Study confirm the point both ways round:

The difference between these two 45%s is not surprising because they were achieved in very different ways. At the time of the 2011 election, a large majority of Scottish voters were opposed to independence, so the SNP's strategy was to downplay the constitutional issue and to try to persuade voters that it was the best governing option under the current arrangements. (There were only 35 references to independence in the party's manifesto, compared with 79 references to health and 90 to education.) The irony of 2011 is that it was exactly the easing of fears among many future No voters that brought the SNP its majority and, in turn, brought Scotland to the brink of independence. But then it was the Yes voters who delivered the SNP's 2015 general election landslide, as referendum voting and Westminster voting moved smoothly into alignment.

So who will vote SNP in 2016? Will the party win back No voters by reminding them that they still see it as the best option for Holyrood government? Or will this be another 2015, in which SNP success is built on a near-monopoly - and another high turnout - among Yes supporters? Of course, the party's preferred answer is 'both'. But riding two horses at once is tricky. Nicola Sturgeon's conference announcement of "a new initiative to build support for independence" was intended to rally the Yes troops while reassuring No voters that a re-run of the referendum was not imminent. Equally, though, the mere mention of independence could scare off the No side while talk of long-term initiatives leaves many Yes supporters champing impatiently at the bit.

The fact that the SNP is polling over 50%, at least on the constituency vote, suggests some success at riding both horses. But there is a limit to what we can infer from overall poll shares. To examine the SNP vote more directly, I return to the Scottish Referendum Study's long-term panel. This has tracked around 600 voters from 2011 onwards, and asked most recently about their Holyrood voting intention in October 2015.

There are two groups of particular interest here, representing the two horses that the SNP is seeking to ride in 2016. The first group voted SNP in 2011 but then No in 2014. The second did not vote SNP in 2011 but voted Yes in 2014. These groups' 2016 voting intentions, for both the constituency and regional ballots, are shown in the graph below.

At this point, then, the SNP's 2016 vote has more in common with the 2014 Yes vote than with the party's own 2011 support. Only around one-third of those who backed the SNP in 2011 but then voted No are planning to return to the party in 2016. That horse may not have bolted completely but it is tugging hard at the reins. By contrast, the referendum continues to be a gateway into SNP-voting for Yes voters who had not previously supported the party. Even on the regional ballot, which offers the Greens as well as some more radical options for independence supporters, the SNP is still by far the most popular destination for referendum-driven switchers.

Why does all this matter? After all, the SNP's poll rating makes a repeat of its 2011 majority seem almost inevitable. Even if built on different electoral foundations, it is still a majority. And the SNP can afford to dip to around 42% of the regional vote and still, thanks to the not-so-proportional-after-all electoral system, win a working majority. So a near-monopoly of Yes voters is enough.

There are two reasons why the tall blue columns above matter. For one thing, they argue against the kind of late swing that might deny the SNP that majority. The party's opponents at Holyrood perceive vulnerability in the Scottish Government's record on health and education, and indeed voters themselves often deliver a much more mixed verdict than in 2011. So a relentless focus on that record might indeed apply downward pressure to the yellow columns in the graph. However, a poll rating built on the referendum is much less vulnerable given the strength of many Yes voters' commitment to independence and given that, at least in the constituency contests, these voters have nowhere else to go.

The second reason is more about longer-term politics within the SNP. In the past, the party's leaders were able to defend their more 'gradualist' approach to the constitutional question on the grounds that electoral success required backing from opponents of independence. That all changes if a 2016 victory is instead delivered by Yes voters impatient for progress. Given that Nicola Sturgeon would clearly prefer such progress to be decidedly unhurried (and understandably so given the continuing No lead in independence referendum polls), she has a strategic interest in broadening the party's 2016 electoral base as far as possible. The SNP's campaign thus has to revive the conditions of 2011 and to re-extend the party's reach among No voters. To do that while keeping supporters of independence happy will not be a straightforward task. Still, as James Maxton put it: "If you can't ride two horses at once, you shouldn't be in the circus".

Author's note: Thanks are due to the Economic & Social Research Council for funding the Scottish Referendum Study, and to James Mitchell for his insights and equine metaphor.

Rob Johns is a reader in politics at the University of Essex. He has written several books on Scottish public opinion and voting and is STV's studio psephologist on our overnight results programmes.