Those polls were wrong in just about everything.

那些民調幾乎都是錯的。

2016 U.

2016 U.

S.

S.

Election polls got it all wrong, didn't they?

選舉民調都搞錯了，對吧？

Well, not quite.

嗯，不完全是。

The national polls were actually pretty accurate.

全國的民調其實很準確。

They showed Hillary Clinton ahead by around 3% on average, and she did win the popular vote by a margin of just over 2%.

他們顯示，希拉里-克林頓平均領先3%左右，而她確實以2%多一點的優勢贏得了普選。

So pretty close to the polling.

所以很接近民調。

But in America, that's not what wins you.

但在美國，這不是你的勝利。

The presidency.

擔任總統的。

Donald Trump won in MAWR key States and ultimately walked into the White House.

唐納德-特朗普在MAWR關鍵國家獲勝，最終走進白宮。

So can we trust the polls to predict the winner in 2020?

那麼，我們可以相信民調來預測2020年的勝負嗎？

If you want to trust them to predict the winner in a razor thin election, that's a mistake.

如果你想相信他們能在剃刀般的選舉中預測勝負，那就錯了。

Pulls even the best ones are really not up to.

普拉即使是最好的，也確實達不到。

That task pulls our snapshot in time about what the public is thinking or feeling at a particular moment when we collect the data.

這個任務可以拉出我們的時間快照，瞭解我們收集數據時，公眾在特定時刻的想法或感受。

So how does polling work specialist polling groups, news organizations and some universities carry out polls over the phone online or a combination of these methods?

那麼，民調工作是如何進行的專業民調小組、新聞機構和一些大學通過電話在線或結合這些方法進行民調呢？

Election polls ask people a variety of questions, but the main one is if the election was held today, who would you vote for?

選舉民調會問人們各種各樣的問題，但最主要的問題是如果今天舉行選舉，你會投給誰？

The number of people taking part varies from pole to pole, but for election polls to be accurate, they need to survey a representative sample of likely voters.

參加的人數因杆而異，但選舉民意調查要想準確，就需要對可能的選民進行有代表性的抽樣調查。

But that could be hard to dio to do polling in the US Well, you have to recognize that some groups are more likely to take polls than others, older folks, whites and college graduates.

但是，這可能很難dio在美國做民調嗯，你必須認識到，有些群體比其他群體更有可能參加民調，老年人，白人和大學畢業生。

And so you have to statistically weight down the college graduates, so they're proportional to what they are in the population.

所以你要從統計學上對大學畢業生進行加權，讓他們與人口中的比例相稱。

Under representing people without college degrees is what some state polls did in 2016 which meant there were some surprise wins for Trump.

低於代表沒有大學學歷的人是2016年一些州的民意調查，這意味著特朗普有一些意外的勝利。

It's the upper Midwest.

這是中西部的上層。

It's Pennsylvania, Michigan and Wisconsin, where the states that were the shockers they used to be known as the Blue Wall for Democratic candidates because they were just consistently cycle after cycle would would support the Democrat for president.

這是賓夕法尼亞州，密歇根州和威斯康星州，這些州是震驚者，他們曾經被稱為民主黨候選人的藍牆，因為他們只是一致的週期後會支持民主黨的總統。

Using polling data to predict the final result doesn't account for voters being influenced by the candidates campaigns.

用民調數據來預測最終結果，並不能說明選民受候選人競選活動的影響。

There was a really late swing in Donald Trump's favor among the 10 to 15% of voters who waited until the last days to make up their mind.

在那些等到最後幾天才下定決心的10到15%的選民中，唐納德-特朗普真的出現了遲到的搖擺。

Getting it right means people accurately predicting their own future behavior about whether they're going to vote in whom they're going to vote for.

做對了，意味著人們準確地預測了自己未來是否要投給誰的行為。

And even if the pollster does everything perfectly, they can change their minds especially, you know, in the long campaigns that we have in the U.

即使民調機構做得再完美，他們也會改變主意，尤其是在美國漫長的競選活動中。

S.

S.

A lot can happen.

很多事情都可能發生。

This'll year.

今年。

More voters than normal say they've made up their minds.

比正常人更多的選民說他們已經下定決心了。

But there's still uncertainty due to the pandemic and a big increase in postal voting.

但由於疫情和郵遞投票大增，仍存在不確定性。

So while Poles can be trusted to provide a good guide to the public mood at a given time, they shouldn't be relied on to predict the final outcome.

是以，雖然波蘭人可以信賴，他們可以很好地引導特定時間的公眾情緒，但不應該依靠他們來預測最終結果。