I often see how people point out here and other places that a poll over sampled the D’s
but the fact is that there are more registered D’S then R’S in the USA so a random sample should have more D’s then R’s right?
Yes but I believe for the last presidential election the electorate was D+3.
DjT’s rally surveys show huge numbers attending are Ds, or have never voted.
Bodes well.
“I often see how people point out here and other places that a poll over sampled the Ds
but the fact is that there are more registered DS then RS in the USA so a random sample should have more Ds then Rs right?”
___________________________________________________________
No. Sampling isn’t based on registration, but instead on “turnout models”, ie., what you expect the turnout percentage by party to be. John Zogby just criticized his fellow Dem pollsters for using models that inflate Dem Turnout. Zogby basis his on the 2016 model, which gives Dems a 2% advantage, as I recall. Some of the media polls give Dems a 9% or more advantage.
“I often see how people point out here and other places that a poll over sampled the Ds
but the fact is that there are more registered DS then RS in the USA so a random sample should have more Ds then Rs right?”
Exactly right. The definition of “oversample” is NOT as simple as 38 (D) minus 33 (R) = 5 therefore “ZOMG!!! DEMOCRAT OVERSAMPLE BY 5 POINTS!!!11!”
...that there are more registered DS then RS...
___________________________________________
Not any longer.
Recent Gallup pol has Rs up _nationally_ +1%. (no link; just saw this this morning)
Subtext: In some places with partisan primaries, straight R voters register D to influence the opposition in the primaries. In those and other places, people must register D in order to vote in municipal/county elections. This is without factoring in that there are workplaces that check registration and jobs can be lost or never gained if one is registered R. Finally, Ds pay for registrations and workers (think ACORN-types) falsify registrations.