The other day I saw another post by Scott Rasmussen of The Rasmussen Reports daily Presidential Tracking Poll and I thought, oh crap, more crap from Scott. Then I started to wonder about polling in general and thinking back to my college days when I ran into statistics…oh boy, what fun. I try not to wonder and ponder too much because I usually end up in trouble.
So I did what I usually do when I wonder and ponder too much; I researched opinion polling in Wikipedia and other sources. This paper represents some of what I found out from Wikipedia, and Bubba, the guy down the street.
Wikipedia tells us, “Opinion polls are usually designed to represent the opinions of a population by conducting a series of questions and then extrapolating generalities in ratio or within confidence intervals.” OK, that’s clear.
So when did opinion-polling start? It seems that the first “official” example was a straw poll by The Harrisburg Pennsylvanian in 1824. Their poll showed Andrew Jackson leading John Quincy Adams in the contest for the presidency. Jackson won the popular vote, not only in Pennsylvania but also in the whole country (which wasn’t big in those days) so straw polls gradually became more popular, but generally on citywide levels.
However, it is a little known fact that the first poll was really one conducted by a group of French university students around 1776. The students followed Paul Revere on his famous ride warning colonists about the British coming and asked one question relating to his warning–did they believe that the British were coming? Recorded responses were—
- 33 percent believed it was the British
- 22 percent didn’t hear what Paul was yelling because the silver rattling in his saddlebags made too much noise
- 21 percent were mad because it wasn’t Italians
- 18 percent didn’t care and wanted to go back to sleep
- 6 percent said they could see Russia from their front porch
This poll has an error rate of +-5 percent because a number of colonists were in a meeting trying to figure out why some people smiled at the mere mention of the word Caribou.
An interesting footnote to this poll was a speech in one small village by a man simply known as Willard, aka “Mitt the Twit”, who said that 47 percent of the colonists just took corn and never added to the corn stores. These people, Willard said, were the takers.
Back to Scott’s use of tracking polls for a minute. Here, I’ll let Wikipedia describe it, “A tracking poll is a poll repeated at intervals generally averaged over a trailing window. For example, a weekly tracking poll uses the data from the past week and discards older data. A caution is that estimating the trend is more difficult and error-prone than estimating the level – intuitively, if one estimates the change, the difference between two numbers X and Y, then one has to contend with the error in both X and Y – it is not enough to simply take the difference, as the change may be random noise.” Is then, Scott’s tracking poll, random noise on a daily basis?
The next logical step in my pursuit of all things polling was to conduct a public opinion poll myself. Being one who likes perfection (well, laziness), I decided to keep the population small—you know around 100. Then I found out I’d have to call all 100 and that would never do. So, I went for the big time, you know, extrapolating the results over the entire adult population of the United States. I got my calculator out, punched in the population of 150 million and got a manageable sample size—13. That was more my speed because it meant I would only have to call the two people who like me and relatives who don’t. Then my wife informed me that I had to do a random sample…it just couldn’t be people who liked me and relatives who didn’t. Further research and brain farts led to my random sampling.
The questions posed another quandary. I sat at the kitchen table and after aggravating my wife with, and for, ideas, I finally arrived at three questions I wanted to ask along with one asking how they voted in 2012 presidential election. Questions and responses as duly recorded in Patrick’s Big Book of Polling Research follow.
- Republican legislators believe they are doing a good job and should be considered heroes. Do you agree or disagree?
- 76 percent answered “No” and called me bad names for interrupting their supper even though I called about 9:30 a.m.
- 22 percent were laughing too hard to answer
- 2 percent didn’t know what that had to do with Caribou husbandry.
2. Do you believe the daily Rasmussen polls are accurate?
- 76 percent thought the Russian Mafia had killed him over 100 years ago for diddling with the Czarina
- 14 percent were still laughing too hard when they thought of question one
- 8 percent were giggling about question one and kept asking me to repeat the question
- 2 percent wanted a question about Caribou husbandry
- Do you believe McConnell can be reelected to his U.S. Senate seat in Kentucky?
- 75 percent were laughing too hard to answer
- 13 percent said it didn’t matter, Kentucky wasn’t part of the United States
- 10 percent wanted to know if one could see Russia from Kentucky
- 2 percent were getting mad because there were no questions about Caribou husbandry
Responses to the question of how they voted in the 2012 presidential election are—
- 51 percent voted democratic
- 47 percent voted republican
- 2 percent forgot to vote because they were watching the Honey BooBoo Marathon
The poll has an error rate of +-5 percent which depends on the Caribou mating season.
Are opinion polls worth their salt? Well, the short answer is that “it depends”. Are the pollsters impartial? We know, or can assume, that Scott and Fox News want responses that favor republicans or their conservative positions, so unless you’re conservative these polls can be thrown out with the trash. To determine if we want to ”believe” the poll, we need to know who conducted the poll and the specific questions asked. Then we can decide.
Do you think the 13 people Scott knows get tired of him calling them at suppertime?