Two referendums in the UK achieved high turnout – the Northern Ireland Good Friday Agreement in 1998 (81%) and the vote on Scottish Independence in 2014 (85%). Elsewhere, only Quebec had higher. At those levels, roughly seven out of every eight people voted. To reach this degree of democratic engagement, voters on both sides need to care a lot about the issue and better still know it is neck and neck.

There is not the slightest prospect of a similar turnout in the UK’s EU referendum. The EU simply doesn’t matter to voters to the same degree. It is not that EU is unimportant; there just isn’t the affection for or affinity with the EU that comes anywhere near that expressed for the British Union by No voters in the Scottish referendum. If this referendum was about joining the EU, there would be no contest.

On Thursday, the turnout will not be high, as defined as more than the last British General Election (66.4%). Normal turnout in referendums averages 15% less than the closest general election in both Europe and the US.

Of course, the US figure contains more minor issue referendums, but although there are big variations in these averages, many important referendums follow the trend. For example, the turnout of Scots authorising the creation of the Scottish Parliament in 1997 was 61%. The Swiss joined the UN on a turnout of 58%. Irish referendums repeat the pattern. It is called selective participation.

The 1975 EU referendum fits too. Though it was held in a more deferential age, in a less media rich culture, it was well contested with Remain losing 4% of its vote share during the official campaign. The 64% turnout was at least 10% below the two closest elections, of October 1974 and May 1979. Although more people voted in general elections then, if the 10% formula is repeated, on the same relative basis, Thursday’s turnout will be 56% to 59%.

Today’s pollsters predict turnout differently. They ask voters to self-certify their intention to vote on a scale of 0-10 where 10/10 is rated certain to vote. For this referendum, certain to vote is between 59% and 62%.

But human nature being what it is, the certain to vote figure is always optimistic. In the Scottish Independence referendum it was discounted by 5%, in UK general elections by 15%, in the AV referendum by 20%, in the last EU elections by 50%. If the general election discount is applied the EU referendum turnout falls to 55%.

Turnout is beginning to look not “normal” but low for a major issue referendum. Damned by a lack of enthusiasm for the EU, turnout poses problems for both sides.

Around 20 million people look likely to abstain in ways that will decide the referendum. For a variety of reasons, abstentions will hit Remain harder than Leave but it is not a one way street.

First, there is the status quo advantage. Much nonsense is talked about this. There is no strict rule but a tendency towards caution in the late stages. It doesn’t always apply. Change still wins out in referendums, though less emphatically.

What tends to happen is a late wilt to caution rather than a flight, a flight would have been already apparent in a dramatic move. Instead the only detectable momentum has been to Leave. The wilder hopes of Remain for a 60/40 win will not be fulfilled.

Harold Wilson went into the official 1975 referendum campaign with a vote share in the polls of 71% which fell to 67% in the actual vote. Cameron and Remain entered this campaign with a vote share of 55% and they appear to have lost that early lead weeks ago.

In response, Project Fear has tried its best to play on the risks of change and the safety of the status quo. I expect it to be rewarded with a wilt to Remain in the last few days – perhaps of 3% or 4%.

However, transferring that shift in opinion to votes in the ballot box is a huge task for Remain. Beyond the Europhile middle classes, the civic duty voter and the political and business elites, many ordinary Remain voters must be motivated to go to the ballot.

In contrast the Leave vote looks more determined on several measures. The certain to vote figures for the two sides is already showing 6% to 12% in favour of Leave.

In terms of the Get Out The Vote (GOTV) operation, the party machines are something of a wasting asset in electoral politics as new ways of reaching voters open up. In Scotland, and elsewhere campaigners and funds are exhausted by elections only six weeks ago. GOTV is likely to be a weak advantage for Remain. The counter from Leave is likely to be stronger when it comes to utilising social media to reach voters.

What about complacency? Campaigners know the result is close. But among the public, there is a widespread expectation that Remain will win.

A similarily complacent expectation in the first Irish Lisbon treaty referendum in 2008 led almost 9% of Yes voters not ‘bothering’ to vote on the grounds they personally wouldn’t be missed. Given the distinct lack of enthusiasm for EU on the Remain side, there is a near certainty that some of this complacency will reduce the Remain vote.

Then there are referendum only voters to consider. The Good Friday Aggreement and Scottish Independence referendums brought to the polls 16% and 12% of their electorates who were referendum only voters (people who don’t vote in elections). In the Scottish referendum of 2014, there was an active canvassing as well as much higher engagement of this group.

Leave’s strongest constituency is the C2DE voter with the poorest record of voting in general elections. These electoral non-voters often say “voting doesn’t make any difference”. Of course in a referendum, every vote counts and it does make a difference. There is no evidence that a referendum only voter cannot be just as active in a low turnout referendum. Leave must get across the importance to this group their vote counts. The weight of coverage in the popular newspapers such as The Sun and the Daily Mail will help turnout.

They may also be aided by the Government and its international cast list, who have allowed a government v the people character to develop in the referendum instead of a people v the people. This could easily introduce an anti-elitism aspect to to the vote.

Remainers are also clinging to the idea that there are silent Remainers. This is another misreading from Scotland. In 2014, the Yes campaigners tended to be noisier in public meetings and online. It caused a small proportion of No voters to keep their heads down until the vote. Voters have not been engaged to the same degree in this referendum. Silent Remainers are probably no more than wishful thinking.

What of deliberate political abstention? It is often forgotten that in a binary referendum that not voting is the third option. A sliver of Remainers who have lost faith in the EU but can’t go the whole hog to voting against it will abstain or spoil their ballot paper. Again, I see no counter trend among Leave voters.

Then there is the question of length of time in office. Mads Qvortrup of UCL produced evidence that the longer a government was in office, the more likely it was to lose a referendum. Given the high profile, central at times, role the PM has played in the campaign, towards the end of his premiership, this may be a small factor aiding Leave.

So, what is my forecast. I correctly forecast the turnout and result for the 2014 Scottish Independence referendum. I don’t forecast this one with the same confidence though I may be lucky.

The difference between the two sides will not get into double figures. Barring a massive and unique loss of nerve, Brexit will win perhaps with a vote of 54% on a 56% turnout. I put an absolute cap on turnout at 61%.

And what of the aftermath? Legitimacy in a referendum comes from the size of the majority and the turnout but also from how fairly the referendum was conducted and the quality of the debate.

If the government does somehow win by a few points on a low turnout, with a referendum campaign seen by th other side as unfair, the result will lack legitimacy. The Government will find the problem has not been solved.

But if Leave wins, they too will be affected by the implications of a low turnout and a small majority.

And in Scotland, it could be more of a problem for the SNP than is appreciated. A split Brexit vote where Scottish turnout is 55% or even dips below 50% will prove a problem for Nicola Sturgeon. Why should a referendum so lacking in legitimacy override the 85% turnout of the independence referendum in which the Scots voted to stay in the Union? If it happens, Brexit looks unlikely to break up the UK.

Nigel Smith is a former industrialist who was chairman of the Yes campaign in the 1997 Scottish devolution referendum. He is a leading expert on referendums.