On the practicalities of voter modeling in elections, for the purpose of influencing how they vote, with special reference to Australian elections. Marketing psychology for governments, and for those who wish to have control of governments. This also relates, in these increasingly polarised times, to the difficulties of getting along.
For now, a smattering of links.
TODO: mention problems of ecological inference, Simpson’s paradoxes in electoral demographics, “Symbolic data analysis” for census data, response bias, survey design etc. Finish up with practical tips.
How it’s being done
Dan O’Sullivan, Inside the RNC Leak
In what is the largest known data exposure of its kind, UpGuard’s Cyber Risk Team can now confirm that a misconfigured database containing the sensitive personal details of over 198 million American voters was left exposed to the internet by a firm working on behalf of the Republican National Committee (RNC) in their efforts to elect Donald Trump. The data, which was stored in a publicly accessible cloud server owned by Republican data firm Deep Root Analytics, included 1.1 terabytes of entirely unsecured personal information compiled by DRA and at least two other Republican contractors, TargetPoint Consulting, Inc. and Data Trust. In total, the personal information of potentially near all of America’s 200 million registered voters was exposed, including names, dates of birth, home addresses, phone numbers, and voter registration details, as well as data described as “modeled” voter ethnicities and religions.[…]
“‘Microtargeting is trying to unravel your political DNA,’ [Gage] said. ‘The more information I have about you, the better.’ The more information [Gage] has, the better he can group people into “target clusters” with names such as ‘Flag and Family Republicans’ or ‘Tax and Terrorism Moderates.’ Once a person is defined, finding the right message from the campaign becomes fairly simple.”
Trump’s ascendancy is far from the first material consequence of Facebook’s conquering invasion of our social, cultural, and political lives, but it’s still a bracing reminder of the extent to which the social network is able to upend existing structure and transform society — and often not for the better.
The most obvious way in which Facebook enabled a Trump victory has been its inability (or refusal) to address the problem of hoax or fake news. Fake news is not a problem unique to Facebook, but Facebook’s enormous audience, and the mechanisms of distribution on which the site relies — i.e., the emotionally charged activity of sharing, and the show-me-more-like-this feedback loop of the news feed algorithm — makes it the only site to support a genuinely lucrative market in which shady publishers arbitrage traffic by enticing people off of Facebook and onto ad-festooned websites, using stories that are alternately made up, incorrect, exaggerated beyond all relationship to truth, or all three.
Businessweek, which published a major look into the campaign this morning, explains how the Trump team has quietly organized a data enterprise to sharpen its White House bid. According to the magazine, the campaign is meanwhile attempting to depress votes in demographics where Hillary Clinton is winning by wide margins.
Parscale was given a small budget to expand Trump’s base and decided to spend it all on Facebook. He developed rudimentary models, matching voters to their Facebook profiles and relying on that network’s “Lookalike Audiences” to expand his pool of targets. He ultimately placed $2 million in ads across several states, all from his laptop at home, then used the social network’s built-in “brand-lift” survey tool to gauge the effectiveness of his videos, which featured infographic-style explainers about his policy proposals or Trump speaking to the camera. “I always wonder why people in politics act like this stuff is so mystical,” Parscale says. “It’s the same shit we use in commercial, just has fancier names.”
It is often asserted that friends and acquaintances have more similar beliefs and attitudes than do strangers; yet empirical studies disagree over exactly how much diversity of opinion exists within local social networks and, relatedly, how much awareness individuals have of their neighbors’ views. This article reports results from a network survey, conducted on the Facebook social networking platform, in which participants were asked about their own political attitudes, as well as their beliefs about their friends’ attitudes. Although considerable attitude similarity exists among friends, the results show that friends disagree more than they think they do. In particular, friends are typically unaware of their disagreements, even when they say they discuss the topic, suggesting that discussion is not the primary means by which friends infer each other’s views on particular issues. Rather, it appears that respondents infer opinions in part by relying on stereotypes of their friends and in part by projecting their own views. The resulting gap between real and perceived agreement may have implications for the dynamics of political polarization and theories of social influence in general.
A central idea in marketing and diffusion research is that influentials— a minority of individuals who influence an exceptional number of their peers— are important to the formation of public opinion. Here we examine this idea, which we call the “influentials hypothesis,” using a series of computer simulations of interpersonal influence processes. Under most conditions that we consider, we find that large cascades of influence are driven not by influentials but by a critical mass of easily influenced individuals. Although our results do not exclude the possibility that influentials can be important, they suggest that the influentials hypothesis requires more careful specification and testing than it has received.
I wanted to excerpt a thing from that, but they want to charge me money for even that, and as I don’t actually like Wired magazine per se, I cannot be arsed.
[…] every product, brand, politician, charity, and social movement is trying to manipulate your emotions on some level, and they’re running A/B tests to find out how. They all want you to use more, spend more, vote for them, donate money, or sign a petition by making you happy, insecure, optimistic, sad, or angry. There are many tools for discovering how best to manipulate these emotions, including analytics, focus groups, and A/B tests. Often times people aren’t given a way to opt out.
FWIW I can’t imagine what “opting out” means for Facebook. It’s already a front-and-centre emotional-manipulation-machine. Can you opt out of the curves on a rollercoaster?
Attention conservation notice: 2700 words on a new paper on causal inference in social networks, and why it is hard. Instills an attitude of nihilistic skepticism and despair over a technical enterprise you never knew existed, much less cared about[…]
How do you do psephological graphical models? causalimpact a la BGKR15? Or is a straight-up PC-algorithm causal study sufficient? How about when the data is a mixture of time-series data and one-off results (e.g. polling before and election and the election itself) How do you integrate external information such as population mobility?
Peter Ellis, on his Australian Federal Election 2019 forecasts post introduces his useful ozfedelect package for R, for a bit of easy modelling. More from that author. eechidna, by Carson Sievert, Rob Hyndman, Diane Cook and Heike Hofmann and many others, has a temporo-spatial electorate data from 2001-2016, and I think ozfedelect builds upon that?
On voters strategically changing electorates
What are the marginal benefits of treating politics like the porkbarrel machine it seems to be and behave accordingly? I’d like it to be otherwise, but let’s work with what we have.
Optimal electoral marginalness, inverse gerrymandering etc. Invading marginal electorates. Organised opposition means one would be are more likely to claim council seats as a side benefit.
How well could you do this? How static are the preferences of the voters?
- BFJK12: Robert M. Bond, Christopher J. Fariss, Jason J. Jones, Adam D. I. Kramer, Cameron Marlow, Jaime E. Settle, James H. Fowler (2012) A 61-million-person experiment in social influence and political mobilization. Nature, 489(7415), 295–298. DOI
- CBDL00: Justin Cheng, Michael Bernstein, Cristian Danescu-Niculescu-Mizil, Jure Leskovec (n.d.) Anyone Can Become a Troll: Causes of Trolling Behavior in Online Discussions.
- BaPe16: Elias Bareinboim, Judea Pearl (2016) Causal inference and the data-fusion problem. Proceedings of the National Academy of Sciences, 113(27), 7345–7352. DOI
- YPHS16: Pranjul Yadav, Lisiane Prunelli, Alexander Hoff, Michael Steinbach, Bonnie Westra, Vipin Kumar, Gyorgy Simon (2016) Causal Inference in Observational Data. ArXiv:1611.04660 [Cs, Stat].
- Bail16: Christopher Andrew Bail (2016) Combining natural language processing and network analysis to examine how advocacy organizations stimulate conversation on social media. Proceedings of the National Academy of Sciences, 201607151. DOI
- Lerm17: Kristina Lerman (2017) Computational Social Scientist Beware: Simpson’s Paradox in Behavioral Data. ArXiv:1710.08615 [Physics].
- ShMc16: Cosma Rohilla Shalizi, Edward McFowland III (2016) Controlling for Latent Homophily in Social Networks through Inferring Latent Locations. ArXiv:1607.06565 [Physics, Stat].
- GhGe13: Yair Ghitza, Andrew Gelman (2013) Deep Interactions with MRP: Election Turnout and Voting Patterns Among Small Electoral Subgroups: DEEP INTERACTIONS WITH MRP. American Journal of Political Science, 57(3), 762–776. DOI
- RuWa06: Donald B Rubin, Richard P Waterman (2006) Estimating the Causal Effects of Marketing Interventions Using Propensity Score Methodology. Statistical Science, 21(2), 206–222. DOI
- WRGG15: Wei Wang, David Rothschild, Sharad Goel, Andrew Gelman (2015) Forecasting elections with non-representative polls. International Journal of Forecasting, 31(3), 980–991. DOI
- FeWi15: Matthew Feinberg, Robb Willer (2015) From Gulf to Bridge When Do Moral Arguments Facilitate Political Influence? Personality and Social Psychology Bulletin, 41(12), 1665–1681. DOI
- KMDB19: Edward H. Kennedy, Jacqueline A. Mauro, Michael J. Daniels, Natalie Burns, Dylan S. Small (2019) Handling Missing Data in Instrumental Variable Methods for Causal Inference. Annual Review of Statistics and Its Application, 6(1), 125–148. DOI
- ShTh11: Cosma Rohilla Shalizi, Andrew C. Thomas (2011) Homophily and Contagion Are Generically Confounded in Observational Social Network Studies. Sociological Methods & Research, 40(2), 211–239. DOI
- Deni16: Benoit Denizet-lewis (2016, April 7) How Do You Change Voters’ Minds? Have a Conversation. The New York Times.
- KiPR00: Gary King, Jennifer Pan, Margaret E. Roberts (10000) How the Chinese Government Fabricates Social Media Posts for Strategic Distraction, not Engaged Argument. American Political Science Review.
- BGKR15: Kay H. Brodersen, Fabian Gallusser, Jim Koehler, Nicolas Remy, Steven L. Scott (2015) Inferring causal impact using Bayesian structural time-series models. The Annals of Applied Statistics, 9(1), 247–274. DOI
- WaDo07: Duncan J. Watts, Peter Sheridan Dodds (2007) Influentials, Networks, and Public Opinion Formation. Journal of Consumer Research, 34(4), 441–458. DOI
- YLSS11: Shuang-Hong Yang, Bo Long, Alex Smola, Narayanan Sadagopan, Zhaohui Zheng, Hongyuan Zha (2011) Like Like Alike: Joint Friendship and Interest Propagation in Social Networks. In Proceedings of the 20th International Conference on World Wide Web (pp. 537–546). New York, NY, USA: ACM DOI
- KoKS19: Ulrich Kohler, Frauke Kreuter, Elizabeth A. Stuart (2019) Nonprobability Sampling and Causal Analysis. Annual Review of Statistics and Its Application, 6(1), 149–172. DOI
- ACKM05: Dimitris Achlioptas, Aaron Clauset, David Kempe, Cristopher Moore (2005) On the Bias of Traceroute Sampling: Or, Power-law Degree Distributions in Regular Graphs. In Proceedings of the Thirty-seventh Annual ACM Symposium on Theory of Computing (pp. 694–703). New York, NY, USA: ACM DOI
- BGHH13: John G. Bullock, Alan S. Gerber, Seth J. Hill, Gregory A. Huber (2013) Partisan Bias in Factual Beliefs about Politics (Working Paper No. 19080). National Bureau of Economic Research
- Degr74: Morris H. Degroot (1974) Reaching a Consensus. Journal of the American Statistical Association, 69(345), 118–121. DOI
- GoMW10: Sharad Goel, Winter Mason, Duncan J. Watts (2010) Real and perceived attitude agreement in social networks. Journal of Personality and Social Psychology, 99(4), 611–621. DOI
- BaTP14: Elias Bareinboim, Jin Tian, Judea Pearl (2014) Recovering from Selection Bias in Causal and Statistical Inference. In AAAI (pp. 2410–2416).
- ZURG17: Ali Zarezade, Utkarsh Upadhyay, Hamid R. Rabiee, Manuel Gomez-Rodriguez (2017) RedQueen: An Online Algorithm for Smart Broadcasting in Social Networks. In Proceedings of the Tenth ACM International Conference on Web Search and Data Mining (pp. 51–60). New York, NY, USA: ACM Press DOI
- AcOP10: Daron Acemoglu, Asuman Ozdaglar, Ali ParandehGheibi (2010) Spread of (mis)information in social networks. Games and Economic Behavior, 70(2), 194–227. DOI
- CrSo82: Vincent P Crawford, Joel Sobel (1982) Strategic Information Transmission. Econometrica: Journal of the Econometric Society, 50(6), 1431–1451. DOI
- Gelm07: Andrew Gelman (2007) Struggles with Survey Weighting and Regression Modeling. Statistical Science, 22(2), 153–164. DOI
- RAWF16: Arnout van de Rijt, Idil Akin, Robb Willer, Matthew Feinberg (2016) Success-Breeds-Success in Collective Political Behavior: Evidence from a Field Experiment. Sociological Science, 3, 940–950. DOI
- Kulk16: V. Kulkarni (2016) Temporal Evolution of Social Innovation: What Matters? SIAM Journal on Applied Dynamical Systems, 1485–1500. DOI
- BrKS16: David E. Broockman, Joshua Kalla, Jasjeet S. Sekhon (2016) The Design of Field Experiments With Survey Outcomes: A Framework for Selecting More Efficient, Robust, and Ethical Designs (SSRN Scholarly Paper No. ID 2742869). Rochester, NY: Social Science Research Network
- Evan17: David S. Evans (2017) The Economics of Attention Markets (SSRN Scholarly Paper No. ID 3044858). Rochester, NY: Social Science Research Network
- Lyon11: Russell Lyons (2011) The Spread of Evidence-Poor Medicine via Flawed Social-Network Analysis. Statistics, Politics, and Policy, 2(1). DOI
- Gran73: Mark S Granovetter (1973) The Strength of Weak Ties. The American Journal of Sociology, 78(6), 1360–1380. DOI
- Gran83: Mark Granovetter (1983) The strength of weak ties: A network theory revisited. Sociological Theory, 1(1), 201–233.
- NoNy11: Hans Noel, Brendan Nyhan (2011) The “unfriending” problem: The consequences of homophily in friendship retention for causal estimates of social influence. Social Networks, 33(3), 211–218. DOI
- TNDL16: Chenhao Tan, Vlad Niculae, Cristian Danescu-Niculescu-Mizil, Lillian Lee (2016) Winning Arguments: Interaction Dynamics and Persuasion Strategies in Good-faith Online Discussions. In Proceedings of the 25th International Conference on World Wide Web (pp. 613–624). Republic and Canton of Geneva, Switzerland: International World Wide Web Conferences Steering Committee DOI