LogFAQs > #945919815

LurkerFAQs, Active DB, DB1, DB2, DB3, DB4, DB5, DB6, Database 7 ( 07.18.2020-02.18.2021 ), DB8, DB9, DB10, DB11, DB12, Clear

Topic List	Page List: 1
Topic	Politics Containment Topic 329: There's No Debate
Corrik7 10/16/20 9:36:12 AM #95:	xp1337 posted... rock as someone who has been in many arguments over how 538 works with corrik trust me this isn't going to go anywhere 538's "skews" are adjustments based on how a polling firm has historically over/underrated a party. If a firm consistently puts out results showing one party a point or two ahead of the actual results... it accounts for that. Like how we laugh at Rassmussen putting out absolutely crazy numbers that in no man's land until they hurriedly start releasing polls closer to what everyone else is saying in the last week or two so that they can try and point to those polls and claim they're a very serious pollster. Silver isn't out there going "i think clinton should be up a point or two in this poll; just a gut feeling" lmfao. And his skews were wrong. He was adding +1 or +2 to polls that had Clinton up already +10 and shit. He had polls that were oversampling the electorate for Democrats already to create their polls and adding an additional skew. Almost 7 out of 10 polls if not more always skewed to Clinton. No matter how clearly a Clinton outlier it was. The polls are adjusted every year so a historical skew in favor of someone never made sense. And, it is likely why he stopped posting skews. If he hasn't skewed in 2016, he would have had a more accurate model. Simple as that. People from this year to then are always pointing to the skewed poll numbers for Clinton to why this year's polls don't matter. And it's why it is exhausting trying to explain to them they are insane to think Trump will win this year. They either call you a liar when showing you the skewed numbers or if they get it explain this is why polls can't be trusted because people can skew them. --- Xbox Live User Name - Corrik Currently playing: Spider-Man (PS4), Quantum Break (X1) <cite>Corrik7 posted [p:945919815] in Politics C... [t:79042418]...</cite> <quote>xp1337 posted... </cite> <quote>rock as someone who has been in many arguments over how 538 works with corrik trust me this isn't going to go anywhere 538's "skews" are adjustments based on how a polling firm has historically over/underrated a party. If a firm consistently puts out results showing one party a point or two ahead of the actual results... it accounts for that. Like how we laugh at Rassmussen putting out absolutely crazy numbers that in no man's land until they hurriedly start releasing polls closer to what everyone else is saying in the last week or two so that they can try and point to those polls and claim they're a very serious pollster. Silver isn't out there going "i think clinton should be up a point or two in this poll; just a gut feeling" lmfao.</quote>And his skews were wrong. He was adding +1 or +2 to polls that had Clinton up already +10 and shit. He had polls that were oversampling the electorate for Democrats already to create their polls and adding an additional skew. Almost 7 out of 10 polls if not more always skewed to Clinton. No matter how clearly a Clinton outlier it was. The polls are adjusted every year so a historical skew in favor of someone never made sense. And, it is likely why he stopped posting skews. If he hasn't skewed in 2016, he would have had a more accurate model. Simple as that. People from this year to then are always pointing to the skewed poll numbers for Clinton to why this year's polls don't matter. And it's why it is exhausting trying to explain to them they are insane to think Trump will win this year. They either call you a liar when showing you the skewed numbers or if they get it explain this is why polls can't be trusted because people can skew them. --- Xbox Live User Name - Corrik Currently playing: Spider-Man (PS4), Quantum Break (X1)</quote> ... Copied to Clipboard!
Topic List	Page List: 1