The only way to evaluate the accuracy of a poll is to have the actual questions and responses before anyone has "curated" them.
Just as a trivial example, occasionally a question like "Who do you think would make the best PM ?"
Kier Starmer
Kemi Badenoch
Ed Davey
Nigel Farage
Now if you were told on the basis of that question alone that "52% said Nigel Farage would be the best PM", you may form a particular series of thoughts.
However if you then learned that the question was (a) mandatory and (b) only have the options I presented, your thoughts may be somewhat different.
All of this was known 45 years ago at least. If it's a shock to anyone then they've not been paying attention.