Yougov changes down to their own recalculation or change in voting intention?

Last week I asked how did Yougov exactly weight the Don't knows and how it brought up a couple of queer results, and as such did it effect the overall result.

Link to last weeks blog post.

Link to this weeks data >> 8-9 Aug
Link to last weeks data >> 1-2 Aug

Now funnily enough in last weeks raw data they gave a comparison with the weeks before in July yet in this weeks they don't. Yougov are wildly heralding a leap of 3% for Labour and a fall of 4% for the Conservatives.

Yet when you look down the figures there has been little change

Whilst including Don't Knows & Won't Votes

If there were a General Election tomorrow how would you vote? This weeks figure (last weeks change)

CON 26% (nc)
LAB 21% (+1)
UKIP 9% (-2)
LD 6% (-1)
NAT 4% (nc)
WV 11% (nc)
DK 18% (+2)
OTH 3% (-1)

The headline rate changed quite a bit

CON 38% (-4)
LAB 31% (+3)
UKIP 13% (+1) Funny that UKIP should go up when the numbers actually go down from last week in the question which includes Won't vote/Don't Knows. Are UKIP voters getting shy?
LD 8% (nc)
SNP/PC 7% (+1)
GREEN 4% (+1)
OTHER 0 (-1)

So yougov due to rounding find a 1%

Now last week I looked at 2 sub samples Scotland and Lib Dem voters from 2015, as both had anomalies.

Scotland managed to have the Labour vote share go down when they took out the DK/WV Quite funny when you have less people sharing from the pie that any ones share should go down it should all go up in proportion.

In this weeks data Labours vote share does now go up when DK/WV are taken out has this contributed to a Labour rise in the headline rate?

Then lets have a look at the Lib Dem voter of 2015 Yougov had them go from 8% when including don;t knows and Won't votes up to 28% who would now vote Conservative when used for the headline rate.

This week much different It is 6% of Lib Dems from 2015 who would now vote Conservative and when removing the Don't Knows and Won't votes it doesn't leap to 28% as it did last week it goes to a more sensible 9%

Just noticed another section which doesn't leap up as much as it did last week for the Conservatives. North of England voters.

Last week Yougov converted 22% of those voting conservative including DK/WV and when excluding the DK/WV it became 38%

This week Yougov had 20% for the Conservatives in the North of England including Don't Knows and Won't vote just down 2% from the week before but when they exclude the DK/WV it 32% a 6% downward change from the week before.

All this adds up to the headline rate for the Conservatives which goes down by 4%

So is it coincidence or Yougov correcting their work which has created this weeks figures? Would love your thoughts.