What's so great about this writeup is that it's the gift that keeps on giving as you keep reading:
- the weakness of arguments from authority
- how replication and not peer review uncovers fraud & mistakes
- the importance of informative priors in evaluating results
- basic numeracy ('how did he afford such an expensive survey?')
- the value of open data
- real data is messy and noisy, undermining clean stories and post hoc narratives
- the weakness of arguments from authority
- how replication and not peer review uncovers fraud & mistakes
- the importance of informative priors in evaluating results
- basic numeracy ('how did he afford such an expensive survey?')
- the value of open data
- real data is messy and noisy, undermining clean stories and post hoc narratives
Absolutely fascinating. It's rare that I can't tear myself away from a story, but this one had the charm.Jun 3, 2015
/subJun 3, 2015
I've just thoroughly enjoyed reading Brookman's paper over breakfast on how misleading the lib-cons 1D ideological scale can be (the average is the meanest information measure) which, iiuc, provides evidence that partisans and more knowledgeable voters have cross-issue consistency whereas the majority support a mix of lib and cons policies: an arbitrary selection of which can be extreme.
I found it a convincing refutation of what is apparently a current political statistical conclusion from ideological scale studies that legislators are out of step with the electorate and that the electorate is less extreme than their representatives.
The paper's referenced in the link as one of Brookman's current research interests.Jun 3, 2015
