評論
打印
Starting with the national fundamentals, our model's expectation for Mr Biden (before seeing a single horse-race poll) is that he should win 50.5% of the two-party vote-a bit above his current 49.4% share in national polls, though below the 52.3% he won in 2020.
從全國基本面開始，我們的模型預計拜登（在未看到任何選舉民調之前）應該贏得50.5%的兩黨選票，略高于他目前在全國民調中的49.4%，但低于他在2020年贏得的52.3%。
The model thinks he is slightly more likely to gain ground on Mr Trump during the next five months than to lose it: on average, it expects Mr Biden to pick up half a percentage point, yielding a tie in the national popular vote.
該模型預測，他在未來五個月內領先于特朗普的可能性略高于落后于特朗普的可能性：平均而言，它預計拜登將獲得半個百分點的支持率，從而在全國普選中打成平手。
Unfortunately for the president, state-level polls do not suggest that the electoral-college advantage Mr Trump enjoyed in 2016 and 2020 has eroded materially.
不幸的是，州級民意調查并未表明特朗普在2016年和2020年享有的選舉人團優勢已大幅減弱。
Mr Biden trails by around five points in the Sun Belt battlegrounds of Arizona, Georgia and Nevada, all of which voted for him four years ago.
拜登在亞利桑那州、佐治亞州和內華達州這三個陽光地帶戰場州落后約五個百分點，然而這三個州四年前都投票支持他。
Our model gives him just a 24% chance of holding on to Georgia, where his lost popularity with black voters is most damaging, and 31% and 36% shots in Arizona and Nevada, where his losses among Latinos hurt him.
我們的模型顯示，他在喬治亞州保持優勢的幾率只有24%，在那里他在黑人選民中的支持最為薄弱，在亞利桑那州和內華達州，由于拉丁裔選民的支持流失，他分別只有31%和36%的機會。
Mr Biden's polling has held up better in the relatively white Rust Belt swing states of Michigan, Pennsylvania and Wisconsin.
拜登的民調結果在密歇根州、賓夕法尼亞州和威斯康星州這三個白人占多數的“鐵銹地帶”搖擺州表現更好。
Mr Trump leads narrowly in all three states. Our model considers all three close to coin flips.
特朗普在這三個州都以微弱優勢領先。我們的模型認為這三州的結果都接近拋硬幣的概率。
State-level vote shares in the Great Lakes region have tended to fluctuate in tandem from one election to another.
大湖區的州級選票份額往往會在每次選舉中同時波動。
It would take only a tiny improvement in Mr Biden's numbers or polling error in his favour to hand him the trio, and a second term.
拜登的支持率只需稍微提高，或者民調出現對他有利的誤差，就能讓他贏得三州的選舉，并成功連任。
However, there are two reasons why Mr Biden's true chance of winning the electoral college is lower than the average chance of victory that the model gives him in these three states.
然而，拜登贏得選舉人團的實際機會低于模型給出的在這三州的平均勝率，原因有二。
First, his slide in other regions makes it less likely that he can make up for a loss in the Midwest with wins in the South or West.
首先，他在其他地區的支持率下滑使得他不太可能通過贏得南部或西部的勝利來彌補中西部的失利。
If he loses Arizona, Georgia and Nevada, he then needs all three Rust Belt battlegrounds (plus one electoral-college vote from Omaha, Nebraska) to muster exactly the magic 270.
如果他輸掉亞利桑那州、佐治亞州和內華達州，那么他需要獲得所有三個“鐵銹地帶”戰場州（加上內布拉斯加州奧馬哈的一張選舉人票）才能湊齊270 票的勝選門檻。
Our model gives Mr Trump a 43% chance of winning them all and Mr Biden 31%, leaving 27% for a split decision that would probably also return Mr Trump to the White House.
我們的模型給特朗普在這三州獲勝的幾率為43%，拜登為31%，剩下27%的幾率是“分歧判定”，這也可能會讓特朗普重返白宮。
Second, in both 2016 and 2020 Mr Trump did far better in the Midwest than surveys implied. Our model assumes that both candidates are equally likely to benefit from polling errors.
其次，無論是在2016年還是 2020 年，特朗普在中西部的表現遠超民調預測。我們的模型假設兩位候選人同樣有可能從民調誤差中受益。
But if you compare current surveys with state-level polling averages from late 2020, rather than with the actual election results, the Rust Belt no longer looks like a Biden-friendly outlier.
但如果將當前的調查結果與2020年底的州級民調平均值進行比較，而不是與實際選舉結果進行比較，鐵銹地帶不再顯得對拜登有利。
Instead, it is consistent with a national trend, in which Mr Trump appears to have gained three to five points of vote share.
相反，它與全國趨勢一致，其中特朗普似乎已經獲得了三到五個百分點的選票份額。
If that were the case, the former president could be on track for a decisive victory, flipping light-blue states like Maine, Minnesota, New Hampshire or Virginia.
如果這種情況成立，這位前總統可能會取得決定性的勝利，在緬因州、明尼蘇達州、新罕布什爾州或弗吉尼亞州等淺藍州翻盤。
Recent polls showing a tied vote in Virginia, or Mr Biden's lead in the Democratic bastion of New York dwindling to a modest nine-to-ten points, bolster that possibility.
最近的民意調查顯示，弗吉尼亞州的選票持平，或者拜登在民主黨堡壘紐約的領先優勢縮小到九到十個百分點，這一點強化了前述可能性。
Mr Biden's situation is far from desperate. Good news for him or bad news for Mr Trump-such as the former president's recent conviction in New York, which has been reflected in only a handful of state-level surveys so far-could shift their chances.
拜登的形勢尚未絕望。對他來說好消息或對特朗普來說壞消息--例如前總統最近在紐約被定罪，只反映在少數幾次州級調查中--可能會改變他們的勝算。
So could the presidential debates. Moreover, the Democratic Party's new base of white voters with college degrees is far more likely to turn out than Mr Trump's less-educated coalition is, a disparity that pollsters' likely-voter screens may not fully reflect.
總統辯論也可能產生影響。此外，民主黨的新基礎：擁有大學學位的白人選民，比特朗普的低學歷支持者更有可能投票，這種差異可能沒有完全反映在民調的選民篩選中。
Our model gives Mr Trump better chances of victory than do other public quantitative forecasts. A model by Decision Desk HQ and The Hillputs him on 56%.
我們的模型認為特朗普的勝率高于其他公開的量化預測。Decision Desk HQ和The Hill的模型認為他的支持率為56%。
The forecast released on June 11th by ABC News's FiveThirtyEight expects Mr Biden's position in swing states to improve enough by November to give him an edge, with a 53% win probability.
ABC News旗下的FiveThirtyEight于6月11日發布的預測顯示，拜登在搖擺州的地位將在11月之前得到充分改善，從而為他帶來優勢，獲勝概率為53%。
All three forecasts agree that the contest is close, and that a victory for either major candidate is easily plausible. But by our reckoning, Mr Trump is the favourite.
所有三項預測都認為競選接近，兩位主要候選人都有可能獲勝。但根據我們的估計，特朗普是最有可能勝出的。
