Populus March Poll

Populus's monthly poll for the Times has topline figures, with changes from last month, of CON 42%(nc), LAB 30%(+2), LDEM 19%(+1).

The last couple of polls have shown Labour staging a slight recovery. This particular poll took place after Gordon Brown's visit to the USA and his address to Congress, and after a period where the Conservatives have been largely silence while David Cameron mourns his son. Populus asked specifically about Brown's visit to Washington - 25% said it made them feel more positive about Brown, 13% less positive - but as usual with questions like this, most of those people saying it made them more positive about Brown were people who are Labour voters anyway.

Cameron and Osborne are now ahead on being the most trusted team to run the economy in the months or years ahead. We've seen them regain a lead with other pollsters already, but this is the first time Populus have put them back ahead since October.

The Times's coverage highlights the difference in voting intention between public and private sector employees. The Conservatives are doing much better amongst private sector employees than amongst the public sector (45% and 38% respectively). Labour too do slightly worse amongst the public sector (26% to 29%). The parties that do better amongst state employees are the Liberal Democrats and Greens. The Times says that the Conservatives are ahead amongst NHS and local government workers and neck and neck amongst teachers, but I expect the samples sizes for occupations are too small to be really meaningful.

This does raise an interesting methodological aside though. Long time readers might remember last year when Ipsos MORI reviewed their methodology after wrongly showing Ken Livingstone ahead in the mayoral race. They discovered their samples were including too many public sector workers, and have since then weighted by public sector or private sector employment. Taking the most recent MORI poll as an example, their raw sample was 25% public sector workers and MORI needed to weight this down to 12%, a huge reduction.

Now, Ipsos MORI these days carry out their polling using quasi-random phone sampling, so if their raw samples are skewed towards the public sector, it's quite possible that other companies will be too - and MORI are alone in weighting by public sector employment.

In Populus's sample, they classed 18% of their respondents as public sector employees - so they've got substantially more public sectors workers than MORI do in their weighted samples. MORI say that their weighting target of 12% is drawn from the ONS's Economic & Labour Market Review, so their target at least should be correct. Does this mean the other pollsters are including too many public sector workers? Well, not necessarily - it could be that the questions they are using to find out employment sector are different, or they are classifying different jobs differently (where, for example, do Northern Rock or Post Office employees go? Or employees of private companies doing contracted out local government work? The question isn't necessarily black and white). Nevertheless, it's something that might be worth looking at.

UPDATE: Andrew Cooper replies in the comment section here about how Populus classified people as public sector.