Over the last few weeks several individuals kindly shared their insights and polling data with me about the Veronica Escobar race. All of them swore me to secrecy because they were affiliated with the campaigns. Eddie Holguin, as you know, recently launched his El Paso Votes APP from which I spent some time discussing how big data can be leveraged in politics. Last weekend, Holguin shared with me a telephone poll he had concluded that day. I’ve been thinking about his polls numbers since Tuesday night.
When Eddie Holguin told me that his numbers showed Veronica Escobar at over 60%, I questioned his sanity, oh no, I mean his methodology because, although, I felt it was likely that Escobar would win, it was not going to be by 60%. Or, so I thought. I posted on my social media channels that I expected Escobar to win by 53%. As the early voting numbers trickled, I instantly realized how Holguin’s poll numbers were right on the money.
I reached out to Holguin and asked him about how his polling methodology. Holguin kindly ignored my previous comments about how I thought his numbers were way off and told me what he had done. He gave me permission to share this with you today.
Holguin told me that he created a dataset of likely voters using his APP based on those who had voted in the last four primary elections. He then ranked his voter list by using data metrics such as estimated household income, work place and how high his targeted voter’s family members ranked in terms of propensity to vote. Holguin added, “I then sprinkled some pixy dust,” actually, he told me that he applied a ranking metric he wasn’t ready to share with me just yet. I persisted but he refused to divulge more details about his strategy.
Eddie Holguin then told me that from his ranked list, he initiated a robocall to the top 12,000 voters on his list. His robocall succeeded in connecting with about 2,200 voters, which answered the question of who they voted for or were likely to vote for on Tuesday.
We both understand that his sample was too small to be scientifically sound. It is what I had challenged him on when he shared his numbers with me on the weekend. But his numbers were almost dead on.
According to his poll, Veronica Escobar was going to get 60.95% of the vote, while Dori Fenenbock was at 18.25%. The actual votes were Escobar at 61.42% and Dori at 22.04%. Clearly Holguin’s poll numbers were close, albeit a little low for Veronica and a little high for Dori. For Chavez, Holguin’s data showed her at 7.66%, a little higher than her actual result of 6.67%. Holguin had Enrique Garcia at 5.11%. Garcia received 5.34% of the vote.
I had seen other polling data that showed Escobar winning with 51 to 54% of the vote. That is why I had predicted an Escobar win at 53%.
But I kept coming back to Eddie Holguin’s poll numbers as the results were sinking in. David Karlsruher had posted on his blog that Veronica Escobar was going to win with “60 or 70 percent of the vote.” When I read that, I laughed out loud and made a screenshot of his post because I thought I might have some fun with him after the results were out.
Karlsruher and Eddie Holguin were right.
Everyone else was wrong.
None of the polling data that was shared with me and the conversations I had with political operatives did not show Escobar at above 54%. Most thought it would be close, but Escobar would win nonetheless.
In the case of David Karlsruher, although I do not know it to be a fact, I suspect that he made his guess based on polling data from Dori Fenenbock’s campaign. Fenenbock and David’s parents are close, at least politically.
But in the case of Eddie Holguin, his insights of the potential outcome came from number crunching using big data analytics. That got me excited. Not only is polling expensive but access to poll data is difficult, for obvious reasons. But Eddie Holguin achieved an almost perfect prediction based on generating a “likely voter” list using big data metrics and robocalling them to get an idea of the election outcome.
That, right there excites me, not only because of the technology nexus, but because I see it as an equalizer in El Paso politics.
