As Benjamin Disraeli or Mark Twain or possibly someone else said: Существуют три вида лжи: ложь, наглая ложь и статистика.

And it's always worth remembering that the phrasing of questions in a poll can influence the answers. And, too, you should remember that in polls like this, the wording of the question is standardized and it's often impossible for the participant to give a more nuanced or qualified answer. Instead, it's like this:

QUESTIONER: "I tend to trust Russia less than I did 10 years ago." Do you strongly disagree, slightly disagree, slightly agree, strongly agree, or have no opinion?

PARTICIPANT: Well, gosh, it depends! Are you asking me whether I trust the Russian people, or whether I trust Putin? And are we talking about Russia's relationship with NATO, or Russian oil policy?

QUESTIONER: [repeating the question with almost identical wording] Do you strongly disagree with, slightly disagree with, slightly agree with, strongly agree with, or have no opinion about the statement "I tend to trust Russia less than I did 10 years ago"?

PARTICIPANT: Sheesh! Put me down for "slightly agree," I guess...

The "Pew Center" is as well-respected as any other polling organization, but that doesn't mean their surveys are completely free of accidental bias.