We all love to take polls and read surveys that come our way, but how much is really involved that can change the outcome of a survey? Are there any things that can skew the results?
On ESPN.com they always have great polls for an avid sports fan. I always love to read them, and I never really paid attention to the wording of these polls. I always just read them and answered what I thought. But today, while skimming through the polls on ESPN.com/sportsnation, I found one that had certain wording that could change ones answer if they were to word it differently. The question read: "Is Alex Smith an ELITE quarterback?" The reason why I capitalized the word elite is because that is a strong word, and it can easily make someone change their thoughts. For example, I think Alex Smith of the San Fransisco 49ers is a very good quarterback, but do I think he is elite, probably not, so I would answer no even though I do think he is very good.
Wording isn't the only thing that can skew the results of a poll. The population sampled is also very important. When we did the M & M's sampling in class we found that the more M & M's we sampled the closer we were to the actual percentages released by the company. It is the same with surveys. The more people you have (to a certain extent) the better chance your survey will be accurate and represent the entire population of the world. Also, you need different people with different backgrounds. If we are going to do a survey on block scheduling and we wanted to get the views of the whole school, but didn't want to sample the whole school and we only polled the people in our class, the results would not represent the thoughts of the whole school.
There also has to be random sampling, which is that every person has an equal chance of being chosen. So back to our class. If we just sampled our class we would not have random sampling because not everyone in our school would have an equal chance of being chosen.
Responder bias can skew the results of a survey and wording can have something to do with it. It is when the responder to the survey answers according to what he think the poller wants to hear. The way that wording can have something to do with it is because if the survey is asked in a way that leads the responder in one way or another than the responder will answer accordingly.
Clearly, wording has a huge impact on the results of a poll, and it is imperative that the wording is proper so the results are not skewed in any way.
Bentzi is absolutely right. I love reading TIME Magazine and the Wall Street Journal. I find the polls that they include within their articles to be especially fascinating, in part because they are generally so inclusive an example of the country. Yet I always find myself getting caught up wondering what type language was used and did all the people polled have an equal environment when answering the questions. This class has only made me question the polls even more now :P
ReplyDelete