There have been a lot of questions and finger pointing in the aftermath of Majority Leader Eric Cantor’s historic loss to the little-known Dave Brat. People have been asking: How did it happen? How did so many people not see this coming? As the polling firm that conducted the lone independent survey during the final weeks of the election, Vox Populi Polling has become part of that story.

The irony of the situation is that the media can’t seem to get its story straight. Some have hailed our firm’s poll for The Daily Caller as the canary in the coal mine, but others have claimed we were wrong. So where does the truth lie? The honest answer is both, and here is why.

Given Cantor’s previous landslide victory and the lack of any serious media attention on the race we assumed turnout for this election would be like the others. And because we were interested in seeing how close this race was among past Republican primary voters, we didn’t pick up on the momentum of new voters that Brat was able to mobilize. We were shocked to see how close the race was and surprised that our poll didn’t get picked up more widely.

But even among past primary voters, our survey showed that Cantor was in trouble. Despite winning the 2012 primary with 79 percent of the vote, Cantor was polling at just 52 percent more than a week out from his election, indicating the growing momentum Brat was building. We also had Brat winning big with independents and Democrats, and also had Cantor winning with self-described Republicans by a much slimmer margin than his own polling.

In the end we only missed Cantor’s final number by 7.5 points on a survey with a margin of error of 4 percent, not the grossly exaggerated numbers some have reported. In hindsight, we wish we had gone back into the field the following day and done a survey that included a larger population of eligible voters. If we were Cantor’s pollsters instead of a firm conducting a public poll, we would have. However, just like every politico, pollster, pundit and reporter, we didn’t believe Brat could pull of one of the biggest upsets in political history.

Another influential factor was turnout. Most people expected turnout to be around 45,000, but in reality it was 65,000. And of those 20,000 new voters they broke 3 to 1 for Brat. So, in the end, Cantor likely received about 35,000 votes from traditional Republican primary voters, but only 9,500 from the new voters. Brat likely received about 30,000 votes from traditional primary voters and an astounding 25,500 from the new voters.

Regardless of what has been written, one thing I agree with is pushing pollsters to review and update their survey methodology, which is something I have been doing. Collectively, we need to review and improve everything including survey design, sampling, data collection, and voter screening. The problem is that these things are difficult and cost money, but they are necessary. I am proud of the adaptations and innovations Vox Populi Polling has incorporated. We are open and honest about our methodology and we will acknowledge flaws. We will continue to study, innovate and improve our methodology and will push others to do so as well. As a team of pollsters with a collective 35 years in the business, the reality is that you don’t always get it exactly right. There are always different variables and factors in play, and this particular example was exceptionally unique. But as a result of the things we missed, Democrats want to discredit our work and we don’t blame them. Every political prognosticator and pollster has painted a dire picture for both House and Senate Democrats this election cycle.

As we head into November, we will take the lessons we learned from the Virginia race, but we will continue to conduct accurate and open polls. Everyone has losses, but we hang our hat on our many victories.

Brent Seaborn is a partner of Vox Populi Polling.