I wanted to do this one last posting regarding the election, but specifically about polls. It was proven in this election cycle that nearly every poll was wrong in predicting the outcome of the election. But a close examination of why the polls failed is not any more fundamental than looking at what polls are.
Polls are based on the principle of quality control sampling in factories. In a large production factory where many units are being produced, checking each unit for quality is time consuming and costly. So the QA engineers thought of the concept of sampling. Randomly choose a certain number of units and examine them carefully and that should give you an idea of the quality of the entire batch. The larger the sample size, the more likely you will find any quality problems until you reach a theoretical max at 100% sample size, which is what we wanted to avoid in the first place.
QA sampling in factories depend on the uniformity of the products being produced. Its also assumed that the products are identical in physical characteristics ( i.e. weight, size, shape, etc) and that the measuring criteria of the 'perfect' item has been validated and calibrated. Get all those right and you will get a very accurate and predictable model for predicting the production output quality of your factory's run. Essentially you are predicting the future outcome of any production run based on previous data (the measuring criteria) and the sample data. The combination yields the 'model' for that specific product line run.
When pollsters applied the same principle to elections, all seemed well for a while. Polls were pretty accurate and they were easy to do. Select the sample based on what ever criteria you want or think is important and make up some lists. Get a bunch of people on the phone and start calling. Yeah..instant poll. It was such a simple process that hundreds and hundreds of companies used this model to set up their businesses. Many were "pay to play" and they saw big dollar signs where any result you wanted could be produced by selective sampling. When the data is collected and computed, the math is correct and all seems scientifically sound.
What the pollsters didn't recognize was that people are not like bottles of lotion coming off the assembly line. They are ever changing and morphing into different entities. Imagine an assembly line where the products can change shape, size, and other physical characteristics on their own. Additionally, new products are also rolling off the assembly line that were never there before. Sampling these people with a flawed set of measurement criteria and measurement tools only added to the error. Lastly, what if the people didn't tell you the truth about their vote ? Why ? because they wanted to screw up the polls that have been calling them day and night for 18 months. The bottom line is that they were using the wrong tool to predict future outcome of the voters. It was as if they were using a ruler to measure the volume of a blob of Jello.
The end result was that 95+ percent of the 'professional' polls were wrong - devastatingly so for the Clinton campaign and the media. Millions of dollars poured into useless polls that did not properly forecast the winner. So now the pollsters and media are finding rocks to crawl under until the next election. Good luck - polls are likely doomed.
The real winners were those that had insight into the voters based on criteria other than poll sampling. For example, an AI program (MogIA) has maintained its unbroken record of predicting every winner for the last three elections. Then there is the college professor, Allan Lichtman, who has correctly predicted the elections for the last 30 years. They all used different tools and analytical methods for their predictions. And finally, you can't forget the Chinese monkey king who correctly selected Trump a few weeks ago - I think Hillary's face was just way too scary to kiss...555
This guy better get a business agent quick. I think he is going to be busy.