"Polling suggests Donald Trump will win the presidential election"
It says nothing of the kind.
What model are you running to reach this conclusion?
You surely don't think that you can just look at the top-line numbers, which represent a less-than-1% sub-sample of all the people asked, and draw any conclusions from it. That would be, not to…
"Polling suggests Donald Trump will win the presidential election"
It says nothing of the kind.
What model are you running to reach this conclusion?
You surely don't think that you can just look at the top-line numbers, which represent a less-than-1% sub-sample of all the people asked, and draw any conclusions from it. That would be, not to put too fine a point on it, idiotic. No one so innumerate as to look at top-line numbers and think they represent the electorate should be allowed near a prognosticator's podium.
Even a relatively simple model--which are often the best kind, as they are robust against a diversity of assumptions--based on recent electoral and primary outcomes, and accounting for top-line polling results-- shows Biden with a five-to-seven point advantage over trump. And that's today, before his first criminal conviction and before his slide into dementia becomes so extreme that even the paid incompetents of the mainstream press are forced to acknowledge it.
So please stop with the doom-saying nonsense based on a profoundly innumerate response to raw polling data. It serves no one, other than click-hungry ghouls trading on fear and outrage, and the worst elements of the anti-democratic right.
I'm not sure I'd go so far as to use the term "Innumeracy" to describe the prognosticators you're criticizing, much less the general population or adult electorate's capacity for understanding news reports; but you're certainly right about the unfortunate consequences of the widespread ignorance of the most basic concepts of statistics and statistical reasoning. What I often find myself wishing in vain is that a basic course in statistics would be required for any four year degree, even, say, French Literature or Grievance Studies. Publishers could use that standard in hiring decisions, of course. At the least, style sheets could prohibit reports of changes in any data that were not statistically significant as if they were newsworthy. We're entering the season when we'll see reports of truly inconsequential, tiny changes in political polls treated as if they were truly reliable harbingers of an important trend.
Curious: what do you mean when you say "top-line numbers"? I'm not a poll-watcher at all (I only know that the polls don't look as promising as they did in 2020, which doesn't mean a lot to me). I do have basic literacy in the concept of statistical power and the significance of random sampling (though I'm hardly a statistician). I have no problem believing the media will draw the most dramatic conclusions it can from the available data (even if the available data doesn't say anything particularly dramatic). I simply haven't heard of "top-line numbers" before.
"Polling suggests Donald Trump will win the presidential election"
It says nothing of the kind.
What model are you running to reach this conclusion?
You surely don't think that you can just look at the top-line numbers, which represent a less-than-1% sub-sample of all the people asked, and draw any conclusions from it. That would be, not to put too fine a point on it, idiotic. No one so innumerate as to look at top-line numbers and think they represent the electorate should be allowed near a prognosticator's podium.
Even a relatively simple model--which are often the best kind, as they are robust against a diversity of assumptions--based on recent electoral and primary outcomes, and accounting for top-line polling results-- shows Biden with a five-to-seven point advantage over trump. And that's today, before his first criminal conviction and before his slide into dementia becomes so extreme that even the paid incompetents of the mainstream press are forced to acknowledge it.
So please stop with the doom-saying nonsense based on a profoundly innumerate response to raw polling data. It serves no one, other than click-hungry ghouls trading on fear and outrage, and the worst elements of the anti-democratic right.
I'm not sure I'd go so far as to use the term "Innumeracy" to describe the prognosticators you're criticizing, much less the general population or adult electorate's capacity for understanding news reports; but you're certainly right about the unfortunate consequences of the widespread ignorance of the most basic concepts of statistics and statistical reasoning. What I often find myself wishing in vain is that a basic course in statistics would be required for any four year degree, even, say, French Literature or Grievance Studies. Publishers could use that standard in hiring decisions, of course. At the least, style sheets could prohibit reports of changes in any data that were not statistically significant as if they were newsworthy. We're entering the season when we'll see reports of truly inconsequential, tiny changes in political polls treated as if they were truly reliable harbingers of an important trend.
Curious: what do you mean when you say "top-line numbers"? I'm not a poll-watcher at all (I only know that the polls don't look as promising as they did in 2020, which doesn't mean a lot to me). I do have basic literacy in the concept of statistical power and the significance of random sampling (though I'm hardly a statistician). I have no problem believing the media will draw the most dramatic conclusions it can from the available data (even if the available data doesn't say anything particularly dramatic). I simply haven't heard of "top-line numbers" before.