Given the response to my regular Scottish feature I've been looking for other crossbreaks to analyse. There seems to have been increased interest in the youth vote recently (including a poll of first-time voters from Opinium) which makes this an obvious candidate.
For YouGov’s 18-24 crosstabs, the weighting bias – the difference in voting intention taken from an aggregation of crosstabs of a poll weighted at GB level and that of a bespoke, weighted poll of that age group – seems quite a lot larger than it is for geographic crossbreaks. But since YouGov recently published monthly aggregates for January to November 2014, we can measure the level and stability of this bias.
From July to November, we find an average (aggregated – weighted) of : CON +5.1 LAB -0.8 LIB +0.9 UKIP -2.5 GRN-3.8. It also turns out that 88% of the time, the difference is within a point of its average for the whole period. By comparison, the margin of error for the approximately 5,000 monthly sample sizes is ±1.4 points, and due to poor response rates among under-25s (and the consequent need to upweight those that do respond), the effective margin of error is slightly higher. So we can use this bias as an adjustment with reasonable confidence.
From these numbers, we can see that December saw a continuation of the trend in prior months for all parties, with (on a like-for-like basis) declining VI shares for Labour (33%, down 2 points) and the Lib Dems (down a point to 5%) and gains for the Greens (up a point to 19%), UKIP (up a point to 12%) and the Tories (up less than a point – their rounded VI remains 24%).
The broader trend over the second half of last year seems to have been a swing from Labour to the Greens, with the other main parties broadly stable.
Going forward I’ll provide periodic updates. Watch this space!