Selzer & Company, a gold-standard pollster that is rated "A+" by statistics aggregator FiveThirtyEight, released their latest poll at 5am on Wednesday. Because of Selzer's sterling reputation, I am inclined to accept their results at face value. However, this survey particularly shows the importance of looking beyond the headline numbers and digging into the data.
In the poll, former Vice President Joe Biden nominally beats President Donald Trump by 47-43% in the national popular vote. However, only 55% of Biden voters say their minds are made up, compared to 82% of Trump voters. Framed in different terms, 43% of declared Biden voters say they "could be persuaded"; only 17% of declared Trump voters say the same.
Let us do some math. For the sake of simplicity, we will ignore the voters who are "not sure" on the question of persuadability.
We can multiply Biden's total proportion, 0.47, by 0.43 (the proportion of Biden voters who are flappable). That yields a result of 0.2021, though Trump will certainly not get all of those votes. Let us assume that just a quarter of these voters decide to cast their ballot for Trump. 0.2021 times 0.25 is 0.051, which we can add to Trump's proportion -- 0.43 -- to yield an "adjusted" Trump proportion of 0.481.
Let us also assume that 25% of Trump's "persuadables" go for the former vice president. 0.17 times 0.43 times 0.25 equals 0.0183, which we can add to Biden's 0.47 to yield 0.488. From there, we have to subtract from Biden the 0.051 who flocked from Biden to Trump. We also have to subtract from Trump the 0.0183 who went to the former vice president.
That calculation gives our final results: 46.3% for Trump compared to 43.7% for Biden, provided that 25% of each side's "persuadables" change their affiliation.
In another scenario, if half of the Biden "persuadables" go to Trump without a reciprocal shift in Trump's support, the president would win 53.1% of the national popular vote (0.2021 x 0.5 plus 0.43). Considering America's polarization at present, a majority for the president would be a modern-day landslide and it would all but guarantee him the Electoral College.
My goal here is not to say that Trump will win half of Biden's persuadable voters, or vice versa. I am not a political scientist and I am not qualified to make that judgement. What I am saying, however, is that headline numbers are a very poor substitute for raw data. Looking at headline numbers could convince you that Biden will be our next president even though a Trump victory in 2020 is statistically possible (or even probable).
Look at all polls with caution and analyze the sub-tabs before rendering judgement.
All unsigned FDL Review content is the product of Declan M. Hurley. Hurley dispatches a digital newsletter several times a week with political news and analysis, a free product that is called the Hurley Report. You can sign up at HurleyReport.weebly.com.
 |
| Trump & Biden (Reuters) |
In the poll, former Vice President Joe Biden nominally beats President Donald Trump by 47-43% in the national popular vote. However, only 55% of Biden voters say their minds are made up, compared to 82% of Trump voters. Framed in different terms, 43% of declared Biden voters say they "could be persuaded"; only 17% of declared Trump voters say the same.
Let us do some math. For the sake of simplicity, we will ignore the voters who are "not sure" on the question of persuadability.
We can multiply Biden's total proportion, 0.47, by 0.43 (the proportion of Biden voters who are flappable). That yields a result of 0.2021, though Trump will certainly not get all of those votes. Let us assume that just a quarter of these voters decide to cast their ballot for Trump. 0.2021 times 0.25 is 0.051, which we can add to Trump's proportion -- 0.43 -- to yield an "adjusted" Trump proportion of 0.481.
Let us also assume that 25% of Trump's "persuadables" go for the former vice president. 0.17 times 0.43 times 0.25 equals 0.0183, which we can add to Biden's 0.47 to yield 0.488. From there, we have to subtract from Biden the 0.051 who flocked from Biden to Trump. We also have to subtract from Trump the 0.0183 who went to the former vice president.
That calculation gives our final results: 46.3% for Trump compared to 43.7% for Biden, provided that 25% of each side's "persuadables" change their affiliation.
In another scenario, if half of the Biden "persuadables" go to Trump without a reciprocal shift in Trump's support, the president would win 53.1% of the national popular vote (0.2021 x 0.5 plus 0.43). Considering America's polarization at present, a majority for the president would be a modern-day landslide and it would all but guarantee him the Electoral College.
My goal here is not to say that Trump will win half of Biden's persuadable voters, or vice versa. I am not a political scientist and I am not qualified to make that judgement. What I am saying, however, is that headline numbers are a very poor substitute for raw data. Looking at headline numbers could convince you that Biden will be our next president even though a Trump victory in 2020 is statistically possible (or even probable).
Look at all polls with caution and analyze the sub-tabs before rendering judgement.
All unsigned FDL Review content is the product of Declan M. Hurley. Hurley dispatches a digital newsletter several times a week with political news and analysis, a free product that is called the Hurley Report. You can sign up at HurleyReport.weebly.com.
Comments
Post a Comment