Methodology (How do we know what we know?).

Here's some words from the text to help you focus your study:

Political culture
Political socialization
Public opinion
Political opinion
Scientific polling
Sampling
Exit poll

YOU should go to graduate school - at least keep the possibility open - if you can do undergrad all the way to a BA - you can go a couple extra years and get an MA, MBA, MS etc. The life long payoff is greater for people with masters degrees and above.

In graduate school you really don't argue about conclusions - you argue about how they got there. All those types of issues are called: "methodology." Methodology is  a fancy word for the methods used to generate the data, view the data, and analyze the data. Remember, Political Science wants to be scientific, that means we have to focus on verifiable reality.  That really means "quantifiable reality" or things that can be measured and expressed in numbers.

You are who you are by age five (or so we are told).

Most of our socialization takes place in the first five years of our lives and we remain who we are throughout our lives. Political socialization is one minor aspect of overall socialization and it occurs through out our life.  The three most important variables in your life - where and when you were born and to whom you were born.  Can you over come your bad luck - of course you can - do most people? Sadly, no.  These three variables have the largest of impact on your life and they are all outside of your control.  Such is life.

Political Science deals with demographic data.

We have a tremendous amount of data on our population . For over fifty years we have been collecting data on people's opinions, as well as a host of demographic data ("Demographic data" puts people into categories.) Political scientists know a great deal about the American public. But so do advertising firms and corporations.

We can tell a great deal about people by what demographic category they fit into.  Not just gender, race, age, income but rather specific questions, taken together can paint a demographic picture of your and me.  Do you rent or own?  How old is your car and is it paid off?  How much money do you (and your family) spend on food OUTSIDE of the house compared to food prepared and eaten inside the home?  How often do you go the the movies, rent videos, go on vacation, fly, rent a car, etc.?

The data collected from these and many other issues can be used to test theories or explanations for practically everything.  They can be used to construct models of understanding for everything from marketing plans to political policies.

Your book is correct to point out that "random sampling" is the key to scientific polling.  Also important is the sample size and that it is representative of the population you are trying to understand.  If I wanted to find out the opinions of college students. I could randomly select students on a corner on campus.  That's not random.  How about a phone book?  Does everyone have a number and only one number? Truly random sampling is harder to get than you might think and involves computers.

And your sample has to look like the population you are studying.  I could ask all the 18 year olds on campus but that would not be representative of the whole student population, is it?  This is random sampling with control(s).

Everyone can use polls and polling data to support or refute their arguments.  But statistics do not prove anything.  Statistics simply demonstrate correlations or show possible connections.  They cannot prove anything.

So here are a few tests you can apply to any poll you read about.

(1) Did they share the question with you? If not, there may be problems in the wording of the question. Asking: "Do you support Clinton?" isn't exactly the same as asking: "Do you support the President of the United States of America?" Leading questions have no place in social science research. Not because they are "manipulative or "bad," but because they muck up the validity of the results.

(2) Did they provide information on sample size, margin of error or any other details? If not, it's probably because they have not followed a standard methodology. (e.g.; By convention (that means general agreement)) most statisticians say 800-1000 responses are needed for accurate calculations. If the responses are split 50/50, maybe the question is ambiguous. If more than 25% respond "don't know," or "undecided," maybe the question is vague.)

(3) Who sponsored the report? While you can't always figure this out and not all science is "for sale," one can question the methodology of a research effort funded by the Tobacco companies that "discovers" healthy benefits of smoking. Or if Vons keeps "coming up the lowest" it's no surprise the methodology has been compromised.