In the "information era," we like facts. We like to know (quantitatively, if possible) the opinions of people around us are. This data is frequently found by newspapers or organizations conducting surveys or polls, which the community generally accept as correct. From newspapers and internet blogs one can easily discover that the public's opinion on political ideas or occurrences is one of the most polled topics. The problem when searching for the truth, which I previously stated is known to be of great importance in this age, is that different polls often contradict one another-sometimes fairly significantly.
For instance, early in 2010 there was a senate race in Massachusetts to replace Senator Ted Kennedy. Republican Scott Brown was running against Martha Coakley, a democrat. On January 9, 2010, Public Policy Polling released survey information concerning this race. (1) In this statement, they said, " The race to replace Ted Kennedy in the US Senate is looking like a toss up, with Republican Scott Brown up 48-47 on Martha Coakley." As you can see, this poll clearly states that the race is a fairly even match between the two. In the poll it states that Scott Brown may have a slight hand up on Martha Coakley, but frankly, the difference is fairly negligible.
The next day, on the tenth of January, 2010, The Boston Globe magazine released a poll of its own. (2)In this poll, made public one day later, stated "Democrat Martha Coakley, buoyed by her durable statewide popularity, enjoys a solid, 15-percentage-point lead over Republican rival Scott Brown as the race for the US Senate enters the homestretch, according to a Boston Globe poll of likely voters."
Why is there such a difference between these polls? How could one clearly state that the race between these two politicians "is looking like a toss up," and another, which came out one day later show that one politician "enjoys a solid, 15-percentage-point lead" over the other? Is it possible that over one day the standings changed this much? That's probably not the case, simply because these polls were most likely administered at about the same time (since they were published one day apart); even if the view of the general public did change during the evening of January ninth, the chances are slim that the findings of a legitimate and evenly sampled poll (involving either telephones using random-digit-dialing, or the internet) could be published by the next morning.
Well, after we conclude that the possibility of having a serious overnight change in public opinion is highly unlikely, we can move on to the question of sampling error. One doesn't have to be a statistician to know that a coin flipped fifty times won't yield twenty-five heads and twenty-five tails without question. Because you can't flip the coin an infinite amount of times, there's going to be a sampling error. The same concept is true in polling: because you can't survey every single person, you're not going to get the exact truth. Having said all this, could a fifteen percent discrepancy be a sampling error? No. A sampling error is generally accepted as plus or minus three percent-not fifteen percent.
While the Public Policy Polling poll did disclose information including how many people they polled ("744 likely voters"), and the questions they asked, they did not include the demographics of the poll. In the poll by The Boston Globe, no information was included concerning the poll except for the actual "15-percentage-point" statement. There is a possibility that one or both of these polls were administered to areas that lean heavily one way or the other as far as politics go.
After seeing each of these polls, how would the politicians involved react to the information? It obviously matters which poll they see. For instance, if Scott Brown saw the poll that showed him as being far behind, he may re-evaluate his own campaign and send a few polls of his own out that make him look a little better. He also may (depending on the background of the poll) be able to discover what groups of people he needs to convince. Polls support the democratic process by helping politicians know what the public thinks and help them know what they need to do to please the public. This is a basic principle of democracy.
Frankly, there are many polls that take place all the time, and unless you have a lot of time to research the background of each poll, you're probably going to just have to figure out which source you find reliable and trust it.
(1) Click here to see the first poll.
(2) Click here to see the second poll. (out of commission, sorry)
No comments:
Post a Comment