The Parties | 27th October 2014

As expected, Greens fall back to 5 per cent in Ashcroft's weekly poll

Last Monday we looked at how Lord Ashcroft’s weekly poll was probably putting the Greens too […]

By

Last Monday we looked at how Lord Ashcroft’s weekly poll was probably putting the Greens too high at 8 per cent.

“…we should be wary of the finding: Ashcroft’s data suggests 28 per cent of 18-24 year olds are now planning to vote Green.

“There are only 56 weighted 18-24 year olds in Ashcroft’s poll – so we can’t say 28 per cent of them are actually going to vote Green.

“…this is only one poll. We should hesitate to say 8 per cent of the population are now converts. … We can only take sub-breaks seriously over time. And YouGov’s have shown the party’s youthful support at far lower levels than today’s poll implies.”

This didn’t stop the Guardian declaring a poll “surge” for the party, and talking of multiple “polls” putting them on 8 per cent. John Harris’ anti-establishment roadshow was duly dispatched to Bristol West, which the party is targeting.

Now Ashcroft’s weekly poll has put the party back at 5 per cent.

So much for the surge.

The Greens shouldn’t be ignored, and perhaps should be in the leaders’ debates, but last week was a case study in the way some pundits can blow up the modest findings of pollsters.

The full data tables for each poll can be found here (last week) and here (today).

This is how the percentage of each age group supporting the Greens changed over the two polls:

This explains why the Greens have dipped back down to 5 per cent.

Last week’s poll gave a headline of “8 per cent Green” because 42 of the 517 weighted respondents backed the party. 16 of them were 18-24 year olds (out of 56 weighted respondents – i.e. 28 per cent – but this is based on 38 actual replies. Only 11 18-24 year olds would have actually said they would vote Green but the age group’s poor response rate meant they were made into 16).

This compares to 4 out of 44 18-24 year olds this week (35 actual replies), and 28 out of 511 respondents overall (i.e. 5.4, or 5, per cent).

These are clearly small numbers, and scarcely ones that should be shaping column inches. The margin of error for 500 respondents is nearly 5 per cent. The data is useful over time. The fault isn’t with Ashcroft or any other pollster (we looked at how YouGov’s Scottish sub-polls were distorted by the Mail and Breitbart last week), but the way polls are reported.

For a more in-depth look at the Green vote, Peter Kellner today looked at how three weeks of YouGov samples have put the party on around 5, rather than 8, per cent.