Study Shows Google was Biased Toward Clinton

Written by Robert Eno

Which came first, the bias or the search results? That’s a question worth asking after reading a (PDF) by a group of researchers looking at how search results could have affected the 2016 election.

In a short paper presenting their preliminary work, the social scientists showed that search results from Google, in particular, showed bias toward Hillary Clinton in the month leading up to the election. The same scientists, in 2015 (PDF), that biased search results may sway the results of elections. A further question in this case is, “Were the results skewed, or was the media reporting from which they sprang?”

Late last month Robert Epstein, of the American Institute for Behavioral Research and Technology (AIBRT), Ronald E. Robertson, of Northeastern University, with Samantha Shepherd and Shu Zhang of AIBRT presented a paper that showed there was systemic bias in favor of Hillary Clinton in search results about the 2016 election. Here are some of the findings of the study:

The authors answer another question regarding the bias: Were the native Google and Yahoo results, on their respective websites, equally biased in favor of Clinton? According to the research, the level of bias toward Clinton on Google, “was more than twice as high as the level of pro-Clinton bias we found on Yahoo.” This data seemingly proves what those in conservative circles have been saying about Google —that it overwhelmingly shows pro-leftist results in search. These criticisms predate the 2016 election.

The authors go on to show that the level of bias was dependent on a number of demographic factors, including gender; whether the search was in a blue, red, or swing state; age; and voting preference. While there were changes in the level of bias, all groups saw a pro-Clinton bias in their search results.

The researchers didn’t go into what could have caused the biased search results. That, of course, is the central question. Once bias is established, the next step is trying to figure out how that bias came about. There are a number of ways the bias could have come about. Here are a couple:

• • Algorithm programming … Search engine algorithms are ultimately programmed by people. Those algorithms could have filters which weigh non-mainstream news sources with a lower degree of importance. When Donald Trump accused Google of bias in their auto-complete in June of 2016, Google said they do not program bias into that feature. However, giving left-leaning or mainstream sources preference over conservative outlets would account for the bias.

• • Underlying bias of the media … Another, more probable, cause is the underlying bias of the mainstream media. Leading up to the election, the media reported that it was inevitable that Hillary Clinton would win.  There have been numerous self-reflections on the election by the media, including this in The New York Times. Many of those reflections talk about group-think and confirmation bias. Based on how the media have treated now-President Donald Trump, however, it is doubtful they learned their own lessons.

In the upcoming months, the group of scientists will be expanding upon their findings. It will be interesting to watch what they report. The important thing they have shown is that there is a problem. The next step is watching to see if Google and Yahoo will admit they have a problem in order to correct it and identify the underlying source.

The mainstream media, and technology companies, have great power to shape the national narrative. As Epstein et al. have shown, that power can sway elections as much as anything the candidates themselves do. It is important to hold those who wield such power accountable. Bravo to the authors for attempting to do so.

Article originally published at ConservativeReview.com.