Predicting the US Presidential Election Results

Google's Election Center Explains the Rigor in Their Consumer Surveys

Four business people review qualitative and quantitative information.
Real-Time Insights and Trend Tracking in Presidential Elections. Getty Images Tom Merton | CaiaImage

The techniques that Google Consumer Surveys employ are applicable to more than everyday consumer insights. Consider the opinion of Nate Silver, who writes the  FiveThirtyEight blog for The New York Times and is considered to be a master - and expert extraordinaire - in statistical polling. Nate Silver has commented that, 

"Perhaps it will not be long before Google, not Gallup, is the most trusted name in polling."

This is not an empty phrase. A look back at the 2012 United States Presidential Elections puts this claim in perspective. And it highlights how far Google Consumer Surveys have come with regard to utility, application, accuracy, and reliability.  

Google used their brief online surveys to measure the reactions of people following the election debates. One aim of the research was to gain a deeper understanding of the political positions and opinions that people held regarding the major platform issues of the candidates. Google also boldly used their consumer surveys to predict the election outcome.

The outcome of the election is history. The outcome of the Google Consumer Survey venture is historical. Put to the ultimate and very visible test, the results from the Google approach to political polling were actually better than those of other more well-known, more conventional polls. In fact, Nate Silver, the statistical polling guru, declared at the time that, 

"Google Consumer Surveys was the #1 most accurate poll online and the #2 most accurate poll overall."

What Makes Google's Brief Online Surveys So Trustworthy? 

Google has designed a broad and detailed process for validation research that provides evidence that, among other things, the samples used in their research is truly representative.

Their validation methods include comparing statistics taken from government sponsored studies with the results of more conventional approaches to collecting data. For instance, telephone surveys have been a stalwart of the surveys research industry for decades, and internet market research panels are a well-developed aspect of data collection in the digital age.

These data sources - one more traditional and the other more innovative - are compared with the government data. The comparative cross-walk of the research types showed that Google Consumer Surveys were superior in three distinct measures: 

  1. The percent of data that fell within 3.5 percentage points of a designated benchmark;
  2. The largest absolute error of the results obtained;
  3. The average absolute error, which is essentially the distance of the results obtained from the benchmark. 

It is apparent from these measures that online brief surveys do not necessarily have to sacrifice the reliability and validity of their results when they pursue speedy outcomes. Indeed, the Pew Research Center, a preeminent global research organization with a sterling reputation, has employed Google Consumer Surveys in their own research.

Pew Research Center Scrutinized the Google Consumer Survey Tool

One nonpartisan study conducted by the Pew Research Center investigated the reactions of the public to the outcome of the 2012 Presidential election.

The post-election study was conducted nearly immediately - the very morning after the election returns.

One aspect of the interest the Pew Research Center had in the Google Consumer Surveys was to examine the validity of the tool. The focus of this deep dive into the validity of the "instrument" was to gauge how well it performed when the research focus is immediate measure of public reactions to events. In addition, the Pew Research Center was interested in whether the Google Consumer Surveys would do a good job of identifying and tracking any changes to those public responses over time. 

Here is what the Pew Research Center reported following their investigation:

"The Google Consumer Surveys sample appears to conform closely to the demographic composition of the overall internet population...In addition, there is little evidence so far that the Google Consumer Surveys sample is biased toward heavy internet users."

Now That You Know

From this work with its endorsing partnerships and its own rigorous validation studies, a market researcher can conclude that the use of well designed and well executed brief online surveys can be at once accurate and fast.