MRP Seat predictions - not as easy as you'd hope

Perhaps they can be consistent with a simple 2 party race constituency (especially Lab v. Con), but throw in a 3rd party (and significant boundary changes, why not), and it's all a hopeless muddle.

Poll 1: Electoral Calculus, poll taken 20-27 May, about 10,000 asked.

Countrywide numbers:
Lab 46%, Con 19%, LD 10%, Reform 12%, Green 8%

For my new Eastleigh Constituency:
LD 37%, Con 24%, Lab 22%, Reform 13%, Green 4%

Poll 2: YouGov, poll taken 24 May-1 June, about 50,000 asked

Countrywide numbers:
Lab 46%, Con 21%, LD 8% , Reform 15%, Green 6% (YouGov's latest general polling figure - not actually in MRP page or spreadsheet)

Eastleigh:
Con 33%, LD 31%, Lab 19%, Reform 10%, Green 6%

So, Lib Dem lead of 13%, or Tory lead of 2%? Quite a difference, even though the national numbers are similar. Meanwhile, Lib Dem party leaflets show them comfortably ahead of the Tories, with Labour a long way behind, and Labour leaflets show the Lib Dems with a narrow lead over Labour, and the Tories way behind! Lies, damned lies, and election leaflets ...