Politics, Uncategorized

538 White Liberals

When political reporters and bloggers talk about a pending election, we can usually expect little more than a recitation of the latest poll. The great ones might throw in a margin of error.

Since 2008, Nate Silver brought a whole new brand of analysis using reason and actual statistical analysis. He didn’t make predictions; he calculated probabilities. In 2012, while many pundits were wildly wrong, Nate accurately predicted the outcome of all but one election, suggesting that his probability forecasts might have even been too conservative.

I’ve been disappointed by FiveThirtyEight’s 2016 coverage and change in vision for a number of reasons, but I’ll focus on one here. Many of their posts have argued that Bernie Sanders has a big (yoooge) demographic advantage in Iowa and New Hampshire, since those two states have some of the highest percentages of self-described “white liberals”. Though consistently trending up, Bernie Sanders is still trailing Hillary Clinton nationally and in states following Iowa and New Hampshire. FiveThirtyEight argues that demographics are to blame. They bring it up again, and again, and again, and again, and again, and again, and again, and again, and again… ok, you get the point. FiveThirtyEight can’t get enough of the “white liberals” argument. Here are 2016 polling aggregates from RCP in Iowa, New Hampshire, and nationally:

Screen Shot 2016-01-30 at 2.50.50 PM

Iowa, 2016

Screen Shot 2016-01-30 at 2.50.07 PM

New Hampshire, 2016

Screen Shot 2016-01-30 at 2.51.13 PM

National, 2016

The problem is that Bernie’s higher poll numbers in Iowa and New Hampshire could be due to demographics, or more likely, are simply due to Iowa and New Hampshire being first, more engaged, and therefore most likely to show interest in a truly insurgent campaign. The rest of the country is obviously going to be slower to take notice. Obama wasn’t nearly the insurgent that Sanders is, nor was he even anti-establishment, yet you saw a similar trend in their numbers in the early states. Keep in mind that in 2008, the Iowa Caucus was on January 3.

Screen Shot 2016-01-30 at 2.56.27 PM

Iowa, 2008

Screen Shot 2016-01-30 at 2.41.42 PM

New Hampshire, 2008

Screen Shot 2016-01-30 at 2.59.18 PM

National, 2008

Clinton led Obama by 20 points until Iowa and New Hampshire. He continued to trail her by 10 points nationally all the way until Super Tuesday, after which he edged out a narrow lead. Sanders is actually doing much better nationally than Barrack Obama did at this point in the race.

I think that it’s simply a matter of time rather than race and labels. The results in Iowa and New Hampshire move national opinion, which is why the candidates put so much time and effort into them. Later states are going to be naturally delayed in paying attention, and naturally hesitant to support an insurgent.

The “white liberals” argument really comes from the demographics of South Carolina where there is a significantly larger black vote, especially among democratic primary voters. Sanders has long trailed Clinton in South Carolina, especially among African Americans. It’s important to realize that very little polling has been done so far in South Carolina, including only two this month. Let’s look at the latest CBS poll in South Carolina, since it asks some interesting questions beyond your typical tracking poll.

Clinton is ahead of Sanders 60/38. Clinton leads Sanders 76/22 among African Americans and Sanders is winning among Whites 60/38.

In my opinion, the most important question from this poll was “How much attention have you paid to the 2016 election so far?”. 46% of likely Democratic voters have paid “some” or “not much” attention to the campaign, so there are plenty of minds to be changed, and that will definitely help Bernie. More important is who those open-minded people are. They are disproportionately young and African American. It’s well known that Bernie is huge (Yooooogge) among young people, and it explains the tepid support from African Americans. Those who haven’t paid attention to this campaign are much more likely to support the well-known, establishment candidate… at least for now.


Conclusion? There is none. That’s why FiveThirtyEight should stop acting like they’ve found the secret decoder ring that will keep Bernie Sanders from the nomination. Maybe African Americans will reject the candidate who marched with MLK, got arrested for protesting segregation, fights for $15/hr minimum wage in favor of the pro-death penalty candidate who pushed the crime bill, mandatory minimums, TPP, and the Iraq war. Maybe… but extraordinary claims require extraordinary evidence.

Politics, Research

ReasonBound Research: Answer the Question!

The latest in my series of highly biased, non-peer reviewed, low-impact, original research…

I have been consistently frustrated with the mainstream media’s surprise at Hillary Clinton’s falling polls and the success of Bernie Sanders. What’s more frustrating is the common consensus that this is a result of GOP attacks with respect to her emails and the attack on the Benghazi consulate. When pundits discuss her poor perception of trust and likeability, phony comedy routines or anecdotes about Clinton’s mother are suggested as good strategy. For me, all this babble misses the mark by a wide margin. I think it’s a simple matter of answering straightforward questions.

I wanted to study how eager candidates have been to answer an interview question. By quickly and directly answering questions, the audience perceives that the candidate is comfortable being asked anything and has nothing to hide. By focusing on the direct question, the candidate surrenders the choice of topic to the journalist, the representative of the people. I believe that the perceptions of candor and honesty can be more precisely analyzed and even corrected when quantified in this way.

The problem is that objectively determining whether or not a candidate has answered a question is problematic. Who gets to decide that someone did or did not answer the question? How do we distinguish style from substance? Often, questions must be challenged on their premise, and doing so is routine, smart politics as well as logic.

On the other hand, “yes or no” questions are much more simple. Objectively, their answers can only be categorized as “yes”, “no”, or “didn’t answer”. The answer is, by definition, not reliant on an accurate premise. What follows is my analysis of Bernie Sanders and Hillary Clinton’s approach to yes or no questions in an interview setting.

If my impression is correct that Bernie Sanders demonstrates greater candor and honesty than Hillary Clinton by directly answering questions, it should be clear that he answers a greater percentage of yes/no questions than Clinton, and takes less time to do so. Maybe I’m wrong. If this is a misperception of Clinton, merely having to do with her gender, political history, etc, the data should show that she answers questions at reasonably the same rate as Bernie.


I collected the top 5 interviews on YouTube from the search “Bernie Sanders Interview” and “Hillary Clinton Interview” respectively. Fortunately, these interviews were comparable in length, content, and both searches even included two interviews from the same journalists, Andrea Mitchell and John Dickerson.

I analyzed only the “yes or no” questions from the interviewers and categorized the candidate’s respective answer into “answered yes/no” or “didn’t answer”. I also recorded the number of seconds it took to get to an answer, (in the first case) or the number of seconds spent talking without giving an answer. (in the second case) I did not record the amount of time talking after giving an answer. This was because, in an interview format, yes/no questions are meant to be open-ended opportunities for elaboration. Otherwise, it would be a very short interview. I don’t think that an explanation of a yes/no answer after the fact detracts from the candidate’s candor, but long explanations before giving an answer do so detract.

For the most part, a “yes” or “no” is required to constitute an answer, but in a few clear cases, answers such as “I agree with that” or “I wouldn’t do that” constituted a clear and direct answer to the question. I watched the interviews on a single pass, and I determined whether or not a question was a yes or no question before analyzing the answer.

I initially included a third category, which was “refused to answer”, because I consider this a non-answer, but still directly addressing the question. However, it only occurred twice with each candidate, and did not require much time to do so. Often this refusal to answer was implicit rather than explicit, so difficult to categorize. I included all these answers in the “non-answer” category for the sake of simplicity.

The clock starts when the candidate starts speaking and ends when the candidate begins the yes/no answer. It was common for an answer to be given before the interviewer was done with the question, in which case, zero seconds is recorded.

The raw data is at the bottom of this page, including a synopsis of the question being asked. The interviews as well as their respective questions are in chronological order so you can look for evolution in the candidate’s style. I encourage you too look it over if you are interested.

Results and Discussion:

I had no idea that the data would be so clear. Bernie Sanders appears consistently desperate to answer the interviewer’s question. Hillary Clinton appears allergic to a straight answer.

Bernie Sanders answers yes/no questions at a much higher rate than Hillary Clinton

In a more holistic, subjective analysis of the questions’ content, I tried to categorize the non-answers into types of question, but found it very difficult to parse. However, there were obvious trends in the types of questions that were not answered. The questions that Bernie Sanders did not answer were almost entirely composed of questions about political speculation or attempts to get him to attack Hillary Clinton. There were only 1-2 questions about policy about which Sanders did not give a straight answer.

For Clinton, on the other hand, about half of her non-answers were about policy. About 40% of the non-answers were concerning her negative perception or controversy concerning her email or the Clinton Foundation. Sanders wasn’t asked as many questions about comparable controversies concerning him, but he answered all of them. The idea that Clinton’s non-answers were merely a result of the controversy surrounding her political standing are not supported by these data.

chart 3

Above are histograms describing the time required to get to an answer or non-answer. I was surprised to see that the time taken for non-answers was roughly the same for both candidates. Of course, Sanders did not give that many non-answers to analyze. Watching the interviews, I had the perception that Clinton was droning on for a long time without answering, suggesting that the mere quantity of time spent without answering gives this impression. When a candidate answers the question and then elaborates, I think the audience is more relaxed and listening to the explanation. When the candidate goes on and on without answering, there is a tension around whether or not the candidate will come around to the answer we all want to hear. Perhaps this is just me.

When it comes to answers, both candidates tended to give their answer right away before elaborating. Clinton gave long explanations before answering with slightly greater frequency then Sanders, but it was still rare. This is accentuated by the fact that Sanders often answered the question before the interviewer was finished asking the question.

Without clear trends in question content driving the candidate’s candor discrepancy, it appears that Clinton simply doesn’t feel the need to answer direct questions to the degree that Sanders does. I believe that this is a product of old school politics of the 1990’s. When media became fast paced 20 years ago, campaigns became concerned with winning the news cycle and producing sound bites. Staying on message meant answering the questions you wanted to answer the way you wanted to answer them.

Now that online media and social networks have dominated our communication, campaigns aren’t limited to sound bites. Free media is unlimited. Furthermore, people used to the constant exposure of social media expect their candidates to be more exposed than ever. The standards by which we judge exposure and candor have shifted, especially among Millennials. Perhaps that accounts for the large age discrepancy in the polls. We are not as likely to forgive candidates like Clinton who feed us bullshit answers. Unfortunately, we have not yet developed the vocabulary to describe this evolution. Hopefully this research can shed a bit more precision on a previously described intuition concerning political perceptions.

Raw Data

RB interview data


Where Were We At This Time In the 2008 Primary?

Using prior election narratives as predictions of the current election season should make you roll your eyes. They are fun ideas to play with, comparing Bernie Sanders to Dean and McGovern, but not very informative. Even comparing 2007 Clinton to 2015 Clinton is greatly misleading, so don’t mistake the following data for that. I do think that it is useful to look back at this time in the last, open presidential primary as a measure of the overall presidential campaign season. It is also valuable to remember how different the environment was as a reminder to avoid such juxtapositions.

So, take a ride with me back to 2007, when George W. Bush was president, and Barack Obama had a babyface. The economy had yet to fall to pieces, and no one outside Alaska had yet ever heard of Sarah Palin. Gay marriage was opposed by most major Democratic candidates and only legal in Massachusetts. It was way, way back when we were 4 years into a war in Iraq and Afghanistan…

National Polls

Barack Obama announced his candidacy for president on February 10, 3 months earlier in the calendar than Bernie Sanders would do. I was there on that freezing, seriously fucking freezing day that Obama announced his run. Obama had faced speculation about a run since his kick-ass DNC speech in 2004, so his name recognition was pretty high going in. It’s interesting to see that he hovered around the low 20%’s for so long into the campaign. He entered the 30%’s after winning the Iowa Caucus, but didn’t catch up to Hillary until Super Tuesday, halfway through the actual voting.

2008 National Polls, RCP

2008 National Polls, RCP

With Sanders’ late announcement and the wide gulf in name recognition, the current national polling is to be expected, no matter what the eventual fate of the Bernie Sanders campaign will be. Though he is ~10 points behind where Obama was at this time, Sanders has actually made greater gains since announcing than Obama did. There’s a lot of time left until people actually vote; It’s still very early.

2015 National Polls, RCP

2015 National Polls, RCP

Iowa Polls

At this time in 2007, Obama had been consistently in 3rd place behind Edwards and Clinton, though they were all close enough for it to be anyone’s game. The first debate was April 26, but the polls in Iowa didn’t really start to move until the fall and winter. It’s surprising to me how many debates passed before the field evolved toward what would be a decisive win for Obama and a disappointing 3rd place for Clinton: 38% Obama, 30% Edwards, 29% Clinton

2008 Iowa Polls, RCP

2008 Iowa Polls, RCP

Considering that Bernie’s name recognition was nil going in, and with only one front-runner in competition, there was no place to go but up for him. What is compelling in all of these 2016 polls is that Biden and O’Malley have not made any gains. This is a legit, pro-Sanders momentum that is showing a true 2-person race. However, it is worth repeating this with every analysis of the 2016 polls: It’s way too early to be sure.

2015 Iowa Polls, RCP

2015 Iowa Polls, RCP

New Hampshire Polls

New Hampshire was by far the most interesting story in the 2008 primary campaign. The polls proved to be wildly inaccurate, as an Obama lead and even momentum out of Iowa turned into a loss to Clinton in the Granite State, foreshadowing a long slog of a primary. Edwards support began draining to the two frontrunners, and his hope for a resurgence in South Carolina would not come.

As for the polls, at this point in the year, Edwards was slowly losing support, and wouldn’t really recover. During the fall, Obama’s support was actually falling in NH, despite gains in Iowa. A late surge in NH had many convinced he was in for a big win. Of course, we were reminded that we have to let people vote, and that New Hampshire voters don’t give a fuck about what anyone outside the state thinks.

Results: 39% Clinton, 36% Obama, 17% Edwards

2008 New Hampshire Polls, RCP

2008 New Hampshire Polls, RCP

Though New Hampshire revitalized Clinton’s campaign, New Hampshire polls are the ones that most have Bernie’s supporters doing happy dances. Bernie has gained 15 points in a couple weeks, which is corresponding to a drop in Clinton’s support. It’s easy to imagine Bernie even passing her in this state before the first debate. But again, it’s too early to tell.

2015 New Hampshire Polls, RCP

2015 New Hampshire Polls, RCP

So What?

FiveThirtyEight did a great piece on Iowa and New Hampshire’s demographics misleading us in their affinity for Bernie Sanders. Read it. It emphasizes the need for Sanders to campaign and show progress beyond the early states, so take these results with that in mind. Unfortunately, there is hardly any polling in the Carolinas, where I really want to see how Sanders performs. I believe that once Sanders is in the debates and gets to tell stories of his role in the Civil Rights movement, his support will broaden a great deal. We also need to hear more of his personal narrative for him to be for real. If anything is clear it is that we’ve only just gotten started. Despite the compelling story of the Sanders Surge of Berniementum, the GOP primary is still the circus to watch. I mean, fucking Trump can be called one of the frontrunners. I’m popping the popcorn already.

GOP National Poll, RCP

GOP National Poll, RCP