I brought up this topic like almost a year ago, upon the occassion of taking statistics. Now, after a year of studying it plus reading some books about it and I found it extremely interesting and can't help but to share it to you. (Foreword: I'm not really an expert, so if there is anything wrong, please correct me)
It is a new field (relatively), one concentration of it is the study of number. Basically what they do to collect data, then people analyze that data to give an answer to something. it sounds alright because there is a saying that: " people may lie but number doesn't." However when it comes across politics (especially) or some other fields, people could make number lie, through statistics.
Polls, the most used statistics in politics. CNN, FOX conduct polls, political institutions conduct polls. What's the matter of these polls?
For one thing, to make number does not lie, we must collect data impartially. And to collect data impartially it requires certain conditions such as Simple random Sample or unambiguous and unmisleading questions etc
For example, CNN conduct a polls about president Bush with a question: "Do you agree with the direction president Bush leads this country to?" by emailing them.
What is wrong with this poll?
The first thing wrong about this poll is that the question is misleading. Why it is misleading? Because it is ambiguous, "the direction president Bush leads this country to?" in immigration policy, or national defense, or economy, or enforcing the law?
The question would most likely to "make" a person answer "no" if he/she disagrees with president Bush in immigration issues, while agreeing in economy (psychologically, people think of negative thing before positive thing when giving answers to polls)
The second thing wrong about this poll is the way the data is collected. How? Because it is a voluntary poll. The first rule of conducting a poll is involuntarily selecting based on equal proportion of the population. In simple word, who do you think would waste time to email to CNN, answering the question? Most likely angry people, strong anti-Bush people because psychologically people think of negative thing before positive thing. And more importantly, neutral people, slightly pro-Bush, or even strong Bush are less likely to email answering the question.
The third thing wrong is also about the way the data is collected. Emailing? That subtracts a lot of people who don't have internet access or don't know how to use internet access.
Furthermore, only people that watch CNN know about the poll. Let's give it top 60% of the population watching it (a highly unrealistic rating), there is 40% of the population didn't watch it and they wouldn't likely to answer the poll.
Moreover, CNN gives news in English, so non-english speakers are not included into the polls at all.
My conclusion is that the poll was totally rigged to serve its political agenda, many more polls are just like that if you pay enough attention. However, from this, I want to assert that statistics is a useful course, take it if you are able to (even though I don't like that class, but maybe because of the teacher). You'd learn many interesting things. And be careful of the "polls" you read every day.
( but unfornutately easy to understand , to cary out and sometimes useful to write an article 