When all the changes in party support in an opinion poll are within the margin of error you can’t report anything as having definitely moved.
Pretty obvious stuff you’d have thought, but take a look at the Reuters headline for its new marginal seats poll:
Poll shows late surge for Conservatives in swing seats
Now look at the actual vote shares:
Conservative 36% (+1)
Labour 36% (-2)
Lib Dem 20% (-1)
In other words, all the changes are within the margin of error. This is not a poll result that justifies the “surge” headline, even though curiously Reuters does rightly describe the Lib Dem share as “steady” which is a fair description for a change that’s within the margin of error.
UPDATE: Just to add to the misreporting of this opinion poll, although initially it was reported as pointing to an overall Conservative majority, MORI then issued a press release pointing out that’s not the case. Hence this juxtaposition of stories over on Politics Home:
UPDATE 2: MORI’s correction has been missed by the Daily Telegraph, which today (Tuesday) has reported (wrongly) that, “An IPSOS/Mori poll of marginal constituencies for Reuters on Monday suggested the party was on course for a two-seat majority”.