Damn that Data! Where did polls go wrong?
A couple of days have now passed since the election. We all need to sit back relax and reflect on what has just happened. The polls were obviously wrong. Or were they? The polls are a reflection of the questions being asked and who they are being asked to. Just maybe the polls were correct based upon what was being asked and to whom.
So if that's the case before we start saying they were all off base let's examine how the polls were structured and conducted. I believe in data. I live for data, I spend my life helping organizations capture, manage, process, analyze, store and retrieve data. The data does not lie. Or does it?
Many years ago, I took a class in high school called "How to lie with Statistics" written by Darrell Huff way back in 1954! That course had a lasting impact on me and how I look at data. Statistics just as with data can say what ever we want. The reality is it still prone to our own biases. That is the problem of polling.
Over the past few years we have focused on "Big Data" to help make decisions or find trends in healthcare, elections, education, government, manufacturing--you name it, there is data everywhere! If you have seen any of my other posts about "Big Data" you know I also believe that "Big Data" is only useful if we break it down into meaningful pieces of data using analytics and tools to dig into the data to find the trends.
What happened to the polls though? Well, I propose that it was a combination of human bias in creating the questions and in who they were being asked of. That created a false sense of accuracy in the polls because it was across the board. Remember, garbage in garbage out? Well that is exactly what we saw this election cycle. Let's not repeat our mistakes next time.
Scott McCabe
www.integraecm.com
Digital IT
8 年I believe its as simple as they were flat out lying to manipulate people and hopefully discourage some and give hope to others. I wouldn't look deeper into than that.
Chief Executive Officer at TECHEAD
8 年Jim Ball predicted the win correctly with his String Logic polling tool: https://twitter.com/realtimepolls
Operations Accounting Supervisor at Cooper Tire & Rubber Company
8 年The polls used selective groups of people so their data was incorrect. They wanted to project a certain viewpoint and was very biased and therefore was not a good predictor of what would happen. Maybe next time they won't try to mislead the public by only going after people that agreed with their viewpoints. I personally know of at least two of the polls where they did not include my data simply because they didn't like my answers. I knew every time they reported their polls they were incorrect. I love data and research so I took my own polls and had quite the opposite results...mine were a lot closer to the truth.
Consultant on finance and operations. Driving business success through finance, data, and technology.
8 年Some data issues. Algorithm issues as well for "adjustments". Correlations change, and small changes can have a large impact. Particularly nasty when done before a large public audience.