Disclaimer: The opinions of the columnists are their own and not necessarily those of their employer.
C. Warren Axelrod

Campaign Lessons Learned—Part 2: Big Data vs Polls

As children, we were frequently admonished by irate adults to “Do as I say, not as I do!” whenever we questioned why we couldn’t do what they themselves did. It was often difficult to reconcile in our own minds why there should be this dichotomy. Well, examining the results of the recent presidential campaign’s polls and big-data analytics, it seems that we are confronted with a similar contradiction.

Until the 2016 presidential campaign, it was usually a matter of who was better at designing polls, and collecting and analyzing the data. In the 2012 campaign, thanks to Nate Silver, Obama’s campaign received better insights into the likely results, with Romney’s team thinking, up until the eleventh hour, that its candidate was going to win. Silver’s methodology is described in the November 10, 2012 FiveThirtyEight blog “Which Polls Fared Best (and Worst) in the 2012 Presidential Race” at https://fivethirtyeight.com/features/which-polls-fared-best-and-worst-in-the-2012-presidential-race/ The revealing first paragraph of the blog is as follows:

“As Americans’ modes of communication change, the techniques that produce the most accurate polls seem to be changing as well. In [the 2012] presidential election, a number of polling firms that conduct their surveys online had strong results. Some telephone polls performed well. But others, especially those that called only landlines or took other methodological shortcuts, performed poorly and showed a more Republican-leaning electorate than the one that actually turned out.”

This analysis had within it a hint of the next advance in divining voter sentiment, namely, Big Data analytics. The lesson that the 2016 U.S. presidential campaign might be that Big Data analytics can provide much more useful and arguably more accurate information about the electorate than does traditional polling. In the first place, the collection of Big Data avoids such problems as contacting those who don’t have landlines. Secondly, Big Data analytics are more objective and avoid the subjective views that many share in a phone call or face-to-face meeting—the “do-as-I-say” syndrome.

While some have attributed Donald Trump’s success to the British Big Data analytics firm, Cambridge Analytica, it is not clear to what extent Big Data analytics actually played a part. I suspect that they did, but so many contributing factors were involved that we may never really know the root cause of the election results, as it were.

In a December 8, 2016 column with the title “No, Big Data Didn’t Win the U.S. Election,” Leonid Bershidsky, a Bloomberg View columnist and founding editor of the Russian business daily Vedomosti and the founder of the opinion website Slon.ru, claimed that:

“[The] legend [around London-based Cambridge Analytica, which advised Trump’s campaign using “big data”] should be taken with a grain of salt … Trump didn’t really win because he ran a smarter, more tech-savvy campaign than Hillary Clinton.”

The article can be found at https://www.bloomberg.com/view/articles/2016-12-08/no-big-data-didn-t-win-the-u-s-election

The lingering cybersecurity issue that arises from the Big Data approach, which will most certainly become the analysis du jour going forward, is that of privacy. It is no longer as necessary to conduct surveys to understand people’s behavior and predict their actions. So much of the data that are available for increasingly-accurate analysis are taken from us without our even being aware or giving consent. Since so many, if not all of us, will say one thing, yet do another, why not just cut out the middleman and go directly to the data? That’s what is surely happening and, unless we adjust our minds and our rules to this new world of privacy (or lack thereof), we are going to continue feeling discomfort as Big Data analytics companies demonstrate that perhaps they know us better than we know ourselves.

This latter assertion was further supported in a brief piece titled “Why you click on those cat videos” on page17 of the February 2017 issue of Fortune magazine, which talked about Facebook’s News Feed algorithm. The takeaway was this:

“The social media giant [Facebook] has found that every user has two distinct taste profiles: what they like, and what they say they like. For example, many people tell Facebook in surveys that they like news stories, but then they almost exclusively click on baby photos and funny videos.”

Yes, I do as I say … or not.

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*