Polls and Statistics

Written By:  Andrew Eide

I’m here to tell you not to trust what the news organizations give you as polling data and for sure you cannot trust the statistics you see or hear. I served as a Statistician for one year at General Dynamics Corporation in New London, Connecticut from 1981 to 1982. I assure you I can take raw statistical data and make it state whatever you want to make it state so I know statistics are tweaked to make them say what the organization, person, news agency, or politician wants them to say.

Why would you not trust polling data especially when they conduct exit polls when you leave the voting place? Simply put the people running the polls and asking the question skew the data heavily to promote their candidate of political party. I’ve had times when I voted and on my way out someone doing polling asked me for my opinion. They would ask me who I voted for and when I told them the name of the politician, who happened to be a Republican, they refused to tally my vote. Therefore if they ask 100 people and 45 state they voted for the Republican candidate, but their vote wasn’t counted, then instead of claiming the Democrat had a 55 percent to 45 percent advantage, or only 10 percent, they would take the 55 votes for Democrats and maybe count only 10 of the Republican responses. When you divide 55 by 65 (because the poll taker refused to count the other 35 Republican responses) you now come up with them claiming their Democrat candidate has an 85 percent to 15 percent advantage over the Republican. See how, if they don’t take the raw data and only give you statics based on the raw data, that the results are incorrect and skewed?

When I worked as a Statistician my job was to pull completed welding job orders and then compare the work reported as completed against what the blueprint called for. Anything less than what the blueprint called for was an error.

At the end of the week my raw data of comparing completed work orders against what the blueprints required came to 11 percent. My supervisor, the Director for the Quality Assurance Engineering Department, told me that having an error rate of 11 percent isn’t enough to justify the existence of the Quality Assurance Engineering Department since the work is being done with a very low error rate. My supervisor then told me to factor the results in some other way to see what I come up with.

Since the welds I reviewed on the completed work orders came in various requirements, such as a well all-around, a weld half way around, a weld on one side only, and depth and width of welds from one-eighth inch to one-half inch I broke down the wells by size and how far around the object they were required to go.

Using the exact same raw data from errors made in the welding, but broken down into different categories based on type of weld and the size, I came out with error rates in each category which ran from 11 percent to 60 percent with an overall error rate of around 35 percent which justified the existence of the Quality Assurance Engineering Department. How in the world did I take raw data that gave an 11 percent error rate and make it come out to an average error rate of 35 percent? Not magic...just tweaking the numbers.

When you vote always remember the exit polling data is skewed, mostly by Liberals and Democrats, to try to get people to think their candidate is winning when, in fact, they are not.