Local council by-elections are not a perfect guide to national electoral performance. Each area has its own issues, by-elections don’t necessarily all happen in the most representative or swing areas, and some people do vote differently at local level than they would in national elections.

Those caveats apply to the contests that have taken place since the General Election – particularly as a few of them were sparked by newly-elected MPs resigning their council seats. But while we shouldn’t convert local vote shares to a national vote share, such election results can still be useful to spot trends. Getting an idea of which parties are up, and which are down, is particularly useful given the cloud of doubt which still cloys around opinion polls.

I’ve had a look through the by-election results as reported by the very useful BritainElects Twitter account, and totted up the rise of fall in the vote of each of the four largest parties (Conservative, Labour, Lib Dem and UKIP) in the seats where they had stood previously. (The rise or fall in a party’s share of the vote where it hadn’t stood the time before would obviously be both distortionary to the overall result and irrelevant to the question of how a party’s vote share is changing, so I left those numbers out). There are a couple of by-elections not included due to there being multi-member contests, in which it is harder to accurately gauge how vote share has changed. But I still ended up with a decent amount of information: 23 results for both Labour and the Conservatives and 15 each for UKIP and the Lib Dems.

Here are the headline average changes in vote share:

Conservative: +4.0 percentage points

Labour: -0.8 percentage points

Lib Dems: +4.7 percentage points

UKIP: -10.9 percentage points

That is an interesting set of numbers. In a small set, subject to the potential issues I’ve already mentioned, I wouldn’t get particularly excited about the Conservative, Labour or Lib Dem figures. But UKIP’s fall in vote share is much larger than the other parties – and tellingly it is based on a far more consistent trend in the data.

The other parties saw a wide degree of variation in their results. Of the 23 Conservative results, seven saw us lose vote share, 15 saw us gain share, and one was no change. Labour lost vote share in ten wards and gained vote share in 13. The Lib Dems gained vote share in eight wards, and lost vote share in seven. By contrast, UKIP saw their share of the vote decline in every single one of the 15 wards.

Similarly, the results are also more clustered for UKIP – the gap between their best and worst changes in vote share is 19.8 percentage points. For Labour that figure is 38.3, for the Lib Dems it is 39 and for the Conservatives it is 46.6.

In other words, the slippage in the UKIP vote appears to be across the board and subject to less regional or local variation than the other parties. If you consider that as a smaller party an area with a UKIP candidate is likely to be in their better territory, which should make the figures better than the national average, it looks even more serious for the People’s Army.