Most polling appears to show that the AV referendum is headed for a solid defeat but is the polling wrong for a odd reason. England might well vote strongly against but Wales, Scotland and NI are having regional assemblies elections and given their local voting systems are forms of PR they may well be elected to vote strongly in favour and with a considerably higher turnout rate than for the English local council elections.
In a worked example using some not unreasonable numbers.
33% turnout of 45 million would be 15 million or so. a 60/40 split for No would equate to 9 / 6 million.
The likely number of the potential voters in the regional assemblies would be around 3.5 million. If they went 3.0 million to 0.5 million (which is a big ask in fairness) in favour of AV it would lead to a much closer result than expected of 9.5 million Yes to 9 million No. Were the Yes vote to make that England result much closer that 60/40 then the higher turnout and margin of victory from the regions could actually save the result for the YES campaign. I suspect a 2:1 win in the regions for AV and if the YES vote in England can get to 45% then I think they can still win.
Between 1979 and 1996 the average turnout in local elections in England was 41 per cent but it is much closer to 30% over the last decade and a half.