Great interview with Nate Silver on the Daily Show
Great interview with Nate Silver on the Daily Show
Plenty of ink has been spilled as of late talking about the potential virtues of Twitter and polling. Last week Twitter CEO Dick Costolo announced that new interactive features in the pipeline for the micro-blogging platform will include some sort of polling. Salon mentioned possible correlations between tweets and polls during the RNC and DNC. We blogged on the Twitter Political Index, which Adam Sharp who manages government and news at Twitter, said is not supposed to replace traditional polling but rather build on it.
I fully anticipate the mainstream media playing up the role of the new media as much as possible this Election season. There is nothing easier to fill in the gaps in a 27-hour-a-day news cycle than reading random tweets from unknown Tweeters (see @ladybigmac, @shuulace, @mrswisscheese, @piratedave, @dogfart).
But the folks over at @civicscience were kind enough to share some information about their Twitter-related polling (for people who think Twitter is an unbiased source of data, this might be shocking):
Nate Silver at FiveThirtyEight mentioned (again) the impact of cell phone bias in presidential polls. Cell phone bias was a known issue in 2010 according to a Pew study, and in 2008 (another Pew study linked here) and even in 2004 where pollster Jim Zogby identified some bias among young voters.
When we started Votifi we knew that the problem of Cell Phone bias is only going to get worse and that many polling companies and campaigns will have trouble absorbing the cost of calling cell phones. Data that is being used to develop important strategies and reported in the news will be skewed. We’re trying to come up with a better way to understand audience insights in a mobile and social world.
The clincher from Nate:
But if Mr. Obama over-performs the polling averages on Nov. 6, we’ll have to reconsider things for 2014 and 2016, by which time number of Americans who rely solely on cellphones may have grown to almost half of the population.
It’s very unlikely that polling in 2016 will look like it does today, when we’ll all have iPhone 8s and probably have computer chips embedded under our skin or in our brains.
By Sam Pauken
Last Wednesday, Twitter released a new feature on their website – the Twitter Political Index. According to Buzzfeed, by analyzing 400 million tweets from 140 million active users, they are able to provide a daily assessment of how the Twitterverse feels about President Obama and Candidate Romney.
In coordination with the parties’ polling companies The Mellman Group and North Star Opinion Research, this looks like the most serious effort to gauge voter sentiment via analysis of data on social networks this election cycle.
For Lou Aronson, inspiration struck four years ago during the heat of the 2008 campaign. He listened as his neighbors traded complaints about the overwhelming number of robocalls streaming into their phones—all except one who proudly proclaimed that since he no longer had a landline, he hadn’t endured a single political call all year.
“I turned to one of my neighbors, who’s a political consultant, and suggested he open a mobile phone based polling company,” recalls Aronson. “He laughed and told me it was the dumbest idea he’d ever heard.”
Four years later, Aronson has left behind a career as an attorney to launch Votifi—a company he hopes will eventually fill a void left by traditional survey research as smartphone use continues to soar. [READ MORE]
By Jeremy Merkel
Lynn Frazier, North Dakota, 1921
Gray Davis, California, 2003
This is the list of governors in US history who have been removed from office through a recall election. Today we could see an addition to this list, when Wisconsin Governor Scott Walker faces off against Milwaukee mayor Tom Barrett.
Walker drew national headlines for his decision to end collective bargaining rights for a majority of the state’s public employees. Opposition from labor unions and Democratic leaders resulted in weeklong protests outside the state capital building. Petitions circulated by United Wisconsin, the Democratic Super PAC that led the recall effort, received over one million signatures.
INFLUENCE OF OUTSIDE MONEY
Public opinion about Walker is certainly polarized. However, it is clear that cash donations from out-of-state players have fueled the campaign on both sides. Spending has also been lopsided. The nonpartisan Wisconsin Democracy Campaign reported that over $60 million has been spent on the race. Half of that came from Walker. Democrats on the other hand spent $4 million. An additional $21.5 million was spent by outside groups that reported their spending. The unreported contributions could add many more millions to the counter. Walker’s top three donors combined gave more than Barrett’s campaign had raised overall. Four of Walker’s top seven donors are out-of-state billionaires, including Amway founder and former Michigan gubernatorial candidate Dick DeVos, and casino magnate Sheldon Adelson, who each contributed $250,000.
Money aside, referendums on public servants are nothing new in Wisconsin, with Scott Fitzgerald, Wisconsin Senate Majority Leader and Walker’s closest ally, also facing recall. Lori Compas, a 41-year-old mother of two became so enraged with Fitzgerald’s attempt to steamroll a pro-Walker Budget Bill through the Senate without a quorum she filed for a recall herself. After creating a website, Facebook page, and Twitter account, Compas spoke at a rally in front of the capital building attended by 60,000 people, and countless town hall meetings after that. Oh, the power of social media….
This more or less speaks for itself on why polling methodologies need to evolve.
It should not be a surprise that the outcomes of landline and cell phone surveys are vastly different given the basic assumptions about age and demographic of cell phone vs. landline users. That’s what polling is supposed to disentangle for us.
However, the harder it gets to run polls by calling people, the more the public is going to have to scrutinize the results of polls to make sure they are telling us an accurate story.
Also, the assumption that cell phone users are all young voters on the East and West coast is changing. Those young, early adopters of cell phones and social networks are all getting older. Senior citizens are migrating to mobile and social in order to stay in touch with their kids and grandkids.
Read the rest of the story here.
Pew just released some new data on the declining response rate for telephone-based opinion surveys.
Everything from contact rates to cooperation rates to response rates are on the decline. From 2009 to 2012 the drop in response rates was particularly significant at 40%, the highest drop since they presented data over a span of 15 years. Even when Pew set out to really try and persuade people to respond by 1) using interviewers “particularly skilled at persuading reluctant respondents” 2) calling non-respondents 15-25 times over a period of two months 3) sending advance letters 4) and offering financial incentives of up to $20, the response rates were still only 22%.
Said Pew of these developments:
It has become increasingly difficult to contact potential respondents and to persuade them to participate.
Despite the low response rates, which might lead many to question the validity of results derived from such surveys, there is some good news. Pew said:
…telephone surveys that include landlines and cell phones and are weighted to match the demographic composition of the population continue to provide accurate data on most political, social and economic measures.
Edwin Rios, a journalism and psychology student at Northwestern University, just posted his feature story of me at the TechCocktail Spring Startup Showcase. Eddie found us online when he was researching tech startups from Chicago that are working on issues related to politics. We had a great conversation at TechCocktail and met again for hot chocolate on a chilly day in Hyde Park. The article generally quite flattering even though I’m not thrilled about the hairline part.
Couple of useful bits:
Ahmad describes Votifi as a peer-to-peer recommendation engine where users find others that share their political view and engage with others on the other side of an issue. The idea arose in 2008. Votifi CEO Lou Aronson was waiting at his children’s bus stop with other parents. He heard them complaining about the automated calls they would receive during the upcoming election season. One neighbor wasn’t because he didn’t have a landline, just his cell phone.
And my favorite:
But when he winds up for his pitch to the people, the delivery is smooth. He changes up his explanation of what the company is with each person he talks to, staring into their eyes through his thick glasses. A young woman in a track jacket returns a glare. After a 10-minute conversation with Ahmad, she leans on her back heel and crosses her arms. “I’m sold,” she says.
It has never been harder to get a person to answer questions over the phone, but millions are eager to share their opinions on Facebook, Twitter, and other emerging social media sites. Figuring out how to analyze this new massive source of information could make conventional polling based on lengthy telephone surveys obsolete.
According to a Gigaom article, IBM is so high on social media analytics software that they think it could generate $16 billion in revenue by 2015. That number is astounding, and if accurate could mean that in a few short years the opinions expressed online through social media might shape the marketing and advertising campaigns of major corporations.
The commercial value of mining social media for consumer information is already being realized, but using tweets and posts as a stand in for political polling is another intriguing possibility. Recently the Wall Street Journal examined the prospect of Tweets as Poll Data. The article points out that public opinion research has “never been easier, or never been harder, depending on how you look at it.” More information has advantages, and the prospect of being able to harness millions of social media data points is intoxicating for political pollsters. Part of what makes social media so appealing is that the data comes in instantaneously. Traditional polling can’t take a new poll every hour, but on the Internet, feed back from say a Presidential State of the Union can be tracked in real time. Social media also gives a better sense of what people are actually doing as opposed to what they think a pollster wants to hear.
An explosion of new data also presents new problems. For example, another WSJ article on the same topic warned that posts are often “ambiguous or even sarcastic.” While computers are making progress in language recognition software, nothing can change the fact that while twitter and facebook are extremely diverse, those who opine politically in public are likely to lean far to the left or far to the right. There is no shortage of data on the people who passionately voice their positions on a myriad of issues, but it is the thoughts of those reluctant to publicly spill their ideology that matter most. Independents and undecided voters are unlikely to tweet about candidates or politics.
Social media analysis certainly has a part to play in modern political polling and strategy, but there is a selection bias inherent in using social media activity as a substiture for traditional polling. A hybrid between the two could be the answer. A social media platform that gathers implicit data while employing specific polling questions that provide explicit data could be the future of political polling.