Politics

This article does talk about "When a Harvard poll meets Fox News", but it's criticizing Fox News's distortion of a Harris poll, not the poll itself (i.e., "How Fox News and conservative media outlets are using a recent Harvard Poll to support their own election narrative.")

And Penn is actually the one complaining about those who "cherry pick to advance agendas". He specifically objected to Fox cherry picking his poll to say that voters prefer a "law and order" candidate like Trump.

Mark J. Penn ’76, a visiting lecturer at Harvard University and leading pollster for the Harvard CAPS/Harris Poll, explains that Fox News’s claim that poll results reflected a positive response to “law and order” messaging “is not the full context of the story.” “Look, articles like this take things out of context,” he continues. “They’re written to make a political point. That’s not the whole picture.”

Penn, a former Crimson news editor, believes that the correct analysis of the poll is that “BLM and the police, frankly, have much better images than” both Biden and Trump. The poll finds that 67 percent of respondents view the police either favorably or very favorably, and 51 percent of respondents view Black Lives Matter favorably or very favorably. In comparison, 44 percent and 48 percent of respondents had a favorable or somewhat favorable view of Trump and Biden, respectively.

“I didn’t cooperate with that article,” Penn says when asked about his thoughts on the Fox News story. “It’s unfortunate that people cherry pick [the poll] and use it to advance agendas.

I replied to a different comment linking that article, so I'll copy it here:

Thank you for showing where that phrase was used in writing, but that is not the only time he has been pointed out for the irony of his juxtaposition. He is a former pollster for the Clintons that became very "trumpy" (to use Politico's word) and instead of being on all news shows the only one that would bring him on is Fox.

The thing about polling is that one can write the questions in order to get the answers they want or need and data can be extracted to portray what is needed. Without the raw data, we really don't know what was asked or how the data portrayed was pulled.

56% of Americans think former President Donald Trump should drop out of the 2024 presidential race

https://www.usnews.com/news/national-news/articles/2023-06-16/poll-majority-says-trump-should-drop-out-of-presidential-race

That's how all reputable election polling was done in 2020. For example, if you take a random sample that happens to be 52% men and 48% women, it is completely appropriate to overweight the women's responses to match their actual percentage in the US, 50.5%.

In fact, in the 2020 election there was a bunch of Trump supporters who had the same doubts as you, and they would "unskew" polls with 52% men responding to give them 52% of the final weighting. Lo and behold, their "unskewed" polls showed Trump in the lead. But the proof of the method is in the election results...

Yeah, that admission kind of makes me pause when considering the results. There should have been a page of the published poll that better described how this was taken. For instance, doing just a LAN line poll skews poll results considerably.

But it's only the beginning of the fed case against Trump, so I'm sure opinion will change.

Most polls are conducted with a political agenda, so they don’t really care…. Especially because it’s incredibly uncommon for people to actually look.