|Don't Believe the Big Data Hype|
|Matthew Dowd||Feb 1, 2013, 1:58 PM|
Opinion by Matthew Dowd:
Confirmation bias is the sociological condition where we, as human beings, seek out information that confirms what we already believe or think and ignore (or don't seek out) information that disputes our opinions.
It makes us more comfortable to find "facts" that agree with our already held opinions or theories, and it makes us uncomfortable to discover information that might prove our theories wrong. But the path to the truth is most often an uncomfortable one where we must confront our own false narratives or myths.
As I have often said in the past, myths consistently get set in stone in politics. Why is this? First, the winners most often write the history, and campaign operatives have an incentive to repeat the story that what they did was the reason a campaign won. Many times this just isn't true, but it serves their purposes.
Second, confirmation bias by the media plays a part because as they explain what happened, they look for information that confirms their theory of the race. Again, many times this only confirms a false theory or a misguided rationale.
There has already been much written and discussed about why President Obama won this race. Various theories have been presented for the reason the president beat Mitt Romney. Let's take a little pop quiz at this point to see what has settled into the pundit population.
What made the difference in this 2012 presidential race and gave Obama the victory?
The correct answer is "E." None of the above.
Yes, these things had some effect, but not the impact that has now become part of the common wisdom. Let's start with a premise I think everyone should agree with: If these three Obama advantages from Big Data to advertising to presidential visits all were determinative of victory, then the difference between the results in target states vs. non-target states should be profound.
A major point on confirmation bias that has fed this is the statistic that while turnout from 2008 to 2012 dropped in the non-target states, turnout was actually up in the 12 target states. And so, therefore, the Big Data technological edge of the Obama campaign made the difference or the targeted advertising or campaign visits. What this cause-and-effect attribution error ignores is that nearly every bit of both campaigns (Obama and Romney) was concentrated in these 12 target states. The 12 states, of course, got all the attention in this past year's presidential race.
The more telling statistic to look at is what happened to President Obama's vote percentage from 2008 to 2012. President Obama's national vote percentage dropped 1.73 percent between his first campaign and his re-election. One would expect that his drop would be smaller in the target states because of the three advantages mentioned above.
Interestingly, there is no significant difference between the president's drop in the 12 target states and all the other non-target states where these three factors were not a play. Indeed, the president's margin in the 12 target states dropped by 1.85 percent and in the non-target states by 1.67%. Surprisingly, Romney did slightly (only slightly) better in the target states than he did in the rest of the country. All the advantages the Obama campaign had in the target states made no real difference in the vote totals when you compare it to all the other non-target states.
If I had to pick the three factors that mattered most in this election, it would be: 1) A flawed opponent (Romney) from the beginning who had serious issues with authenticity and connection with middle-class voters, and one where the Obama campaign did an effective job nationally on big message exploiting this (the success of the Obama national convention vs. the failure of the Romney convention was a major part of this); 2) A macro environment of a recovering economy that helped drive up President Obama's job approval number; and 3) Hurricane Sandy, which showed Obama as an effective strong leader amid a terrible disaster.
Although the winning consultants want us to believe it was millions in target-state advertising or technological advantages on voter turnout or numerous campaign offices located in target states, that really is only a bias that, unless we look at the data ourselves in a discerning way, we will repeat as a myth and move further from a clear view of what happened.
And this is why many people might be overestimating the ability of the president's advocacy arm, Organizing for America - stocked with former campaign staff, target-data lists and tactics honed in 2012 - to effect change in Washington.
Further, as the Republicans examine what went wrong, they would be wise to understand exactly what happened. They might just be replicating a myth.