Using the treasuries statistics, the only accurate thing to say is 6.2% smaller compared to staying in by 2030.
Doesn't make such a good headline though.
Thats without going into the flaws in the calculations itself, which are beyond complicated.
LSE data has estimates ranging from 1.5% to 9.5% smaller, and doesn't attempt to estimate the effects of a number of changes which the brexit campaign says will benefit the UK economy.
Most calculations assume that leaving the EU would lead to lower productivity growth, which forms a major part of the drop in GDP.
This is because they estimate that foreign investment would fall by a rather large amount. However there isn't much evidence to suggest that this is guaranteed to be the case. The effects on productivity and wages that come from having far fewer unskilled migration is never mentioned either.
Depending on what variables you plug into these models, you end up with GDP changes ranging from small-medium positive, to large negative. All major institutions assume a similar set of variables, as they come from the same schools of thought.
My take is that we'll see a small short term drop, and after a number of years we'll be growing faster than we would be inside the EU.