After much delay, CNN amended the criteria for who will be included in their Republican primary debate in two weeks. It became clear in recent weeks that the initial criteria would make it nearly impossible for former lower-tier candidates to ascend to the main stage, even if their poll numbers improved. The practical impact of the initial rules would have been to keep former business executive Carly Fiorina off the stage, despite her clear rise to into the top 10  in recent polls.

CNN debate organizers, like Fox News before them, are contending with the challenge of holding a debate with a historically large field of Republican candidates. Rather than crowding all 17 or so candidates onto the same stage, CNN and Fox each decided to split their debates in two and to use national primary poll averages to determine who would make the main event and who would be left on the undercard. Fox used the last few polls before their event in August, while CNN initially took the opposite route, saying they would average nearly eight weeks of polling stretching back to mid-July.

But by including the old polls conducted before Carly Fiorina’s rise, CNN made it nearly impossible for her to shake off her old anonymity and reach the top tier. Facing criticism from all quarters (including this one) CNN announced yesterday that they will add any candidate to the main stage who “is polling in the top 10 in an average of approved national polls released between August 7 and September 10, even if those candidates did not poll in the top 10 in an average of approved national polls between July 16and September 10.”

The catch is, they will still apply their old and strange timeframe to pick the top 10, but will introduce a new side door to add yet more candidates to the stage. The top 10 will be chosen using the same odd method, but anyone who is improving (such as Fiorina) will also be included. The most likely outcome is that the now-top 11 will include someone who used to be doing better but has fallen lately. This appears most likely to benefit either Chris Christie or Rand Paul, who will be invited despite no longer polling in the top 10, simply because they used to be more popular.

It is not clear how including someone who is not in the top ten is less objectionable than it would have been to exclude someone who is the top ten. The most clear and logical approach would be to include anyone doing well in fairly recent polls and not anyone else. But that’s still not what CNN is looking at right now.


The Crosstabs


Organizers of the LA Olympic bid say support for the idea is sky high among local residents. To evaluate their claim, we analyzed all of the independent polling to date. See?Polls2024

2016 Primary Check

Democrats: Bernie Sanders leads the NH Democratic Primary, according to the last two polls. In Iowa, one of the state’s top pollsters finds Sanders within 7 points, a closer race than others have seen so far. Nationally, it still looks like a laugher.

Republicans: Donald Trump.

From the 2016 Motivated Reasoning Desk, this remarkable report from HuffPollster.  Partisans are much more likely to support a policy if the poll suggests their preferred leaders support it, and vice versa-even when the policy is exactly the same.

People think President Barack Obama is a Muslim. No they don’t. Yes they do. No they don’t.


Pew finds more and more Catholics are open to non-traditional families and other things too.

– – – – – – – – – – – #NerdAlert Tearline  – – – – – – – – – – –

Steve Koczela and several co-authors have published a new article on survey data fabrication. The article shows a serious problem has developed in international survey research that goes beyond what would be found with typical quality control measures. These concerns don’t only apply to small research outfits without the resources to address them, but large institutions and governmental organizations as well. Datasets from major institutions showed signs of fabrication extensive enough to call findings from these datasets into question.

The site is paywalled, but a free copy of the paper will be posted soon.