4 ms·

>you and I and most of our peers know that the polls weren't wrong They were wrong though, multiple key polls in battleground states were far outside their mar

by 0xy 5y ago

>you and I and most of our peers know that the polls weren't wrong They were wrong though, multiple key polls in battleground states were far outside their margin of error, indicating methodology failures or bias. As an example, the national ABC/WaPo poll on November 3-6, 2016 claimed a +4% Clinton result, with a 2% margin of error. [1] That's an indictment of their methodology, and it's an indictment of Nate Silver who weights them as an "A+" pollster. I also don't think anyone needs to misrepresent research and science, since huge swathes of it are fraudulent anyway. From the reproducibility crisis to outright fraud and lies, trusted institutions have been desecrating their reputation for years. The CDC now enjoys a trust rating around 50%, because of incompetence and lies. News media is worse, because they lie more than they tell the truth. [1] https://www.washingtonpost.com/news/the-fix/wp/2016/11/07/post-abc-tracking-poll-clinton-47-trump-43-on-election-eve/ https://www.washingtonpost.com/news/the-fix/wp/2016/11/07/po...

InitialLastName 5y ago
>As an example, the national ABC/WaPo poll on November 3-6, 2016 claimed a +4% Clinton result, with a 2% margin of error The end result was roughly Clinton: 48%, Trump: 46%, which is right at the margin, not "far outside".
slg 5y ago
That poll has it 47% to 43% Clinton when the actual results were 48.2% to 46.1% Clinton. The only real discrepancy is they overestimated the vote totals for 3rd party candidates. The numbers they had when excluding 3rd parties was 49% to 46% Clinton. Do some blending of those two and they basically nailed it. Is this really the poll you want to call out as fundamentally flawed?
belltaco 5y ago
>As an example, the national ABC/WaPo poll on November 3-6, 2016 claimed a +4% Clinton result, with a 2% margin of error. [1] That's an indictment of their methodology, and it's an indictment of Nate Silver who weights them as an "A+" pollster. Your link says 2.5% margin of error and not statistically significant. What am I missing? Ironically the comment appears to be misinformation. >Clinton’s edge in the Post-ABC poll does not reach statistical significance given the poll’s 2.5 percentage-point margin in sampling error around each candidate’s support

by 0xy 5y ago

InitialLastName 5y ago

>As an example, the national ABC/WaPo poll on November 3-6, 2016 claimed a +4% Clinton result, with a 2% margin of error The end result was roughly Clinton: 48%, Trump: 46%, which is right at the margin, not "far outside".

slg 5y ago

That poll has it 47% to 43% Clinton when the actual results were 48.2% to 46.1% Clinton. The only real discrepancy is they overestimated the vote totals for 3rd party candidates. The numbers they had when excluding 3rd parties was 49% to 46% Clinton. Do some blending of those two and they basically nailed it. Is this really the poll you want to call out as fundamentally flawed?

belltaco 5y ago

>As an example, the national ABC/WaPo poll on November 3-6, 2016 claimed a +4% Clinton result, with a 2% margin of error. [1] That's an indictment of their methodology, and it's an indictment of Nate Silver who weights them as an "A+" pollster. Your link says 2.5% margin of error and not statistically significant. What am I missing? Ironically the comment appears to be misinformation. >Clinton’s edge in the Post-ABC poll does not reach statistical significance given the poll’s 2.5 percentage-point margin in sampling error around each candidate’s support