CODING THE DATA


The answers to some questions are numbers. If you ask how much people weigh, how many cigarettes they smoke daily, or how many brothers and sisters they have, the answers will be numbers , and you can simply leave space on the questionnaire form to write in each one. You should leave enough room for the biggest number possible, even if you do not expect to get it. If you ask enough people how many children they have, for example, you will certainly find somebody who has ten or more, so you should leave sufficient room for a two-digit number.

When the answer to a question is not a number, you should try to figure out in advance what answers are possible. Respondents would then select among the alternatives. If you do not think of the possible responses before the survey, you will be in serious trouble. For a question about how people view life in general, if you let people supply their own alternatives, you may end up with as many different answers as there were people. How are you going to analyze "Kinda OK," "Could be worse ," "Great, except for my job," "Today exciting, yesterday not"? You would spend hours deciding what to do with a few hundred of such answers. By forcing people to choose among specific alternatives (such as "Exciting," "Routine," "Dull"), you can get data that you can analyze.

What if you really are interested in the way people say things on their own? Sometimes you simply do not know in advance what people will say, and you want to allow them to say exactly what they please . You can certainly have interviewers write down exactly what the respondents say, word for word. Questions like these, which do not specify the possible responses, are known as open -ended questions. A computer will not be much help in analyzing the responses to open-ended questions directly, though, so you should either study the answers and assign codes to them before you enter your data or else ask the same questions again in a different way, with specified choices for responses.

For questions that require choosing among alternatives, think about all the possibilities. Make sure nothing falls through the cracks. Anticipate the unusual. For example, if you want to ask about housing, remember that not everyone lives in a house or an apartment. It is especially important to make provisions for responses such as "Don't know" or even "None of your business." Do not leave it up to an interviewer to decide what to do when somebody cannot or will not answer a difficult question. Anticipate these problems, and write clear instructions on the form in the places where such answers can occur. Whenever answers that do not fit into your coding scheme are possible, include an "Other" category, and leave space on the form for writing out the unusual answers. You may be able to do something with this written information later.

On any survey form, all acceptable answers to a question should be listed. Each answer has a code with it ” a number that represents that answer. For the exciting-routine-dull life question, a code of 1 is circled for the answer "Exciting," 2 for "Routine," and 3 for "Dull." The code number 8 is reserved for the answer "Don't know." These numerical values are the coding scheme for the variable. For each respondent, one of these numbers will go into the computer to represent the answer to this question.

Coding schemes are arbitrary. A code of 3 could just as well have been assigned to Exciting and a code of 1 to Dull. What is important is that each possible response has a code that is different from the others. For example, you would not code the states of the union by their first letters because the first letters for the names of many states are the same.

Coding is an excellent way to do analysis with variable data, whether those data come from engineering, marketing, personnel, or any other place to which the study is directed.




Six Sigma and Beyond. Statistics and Probability
Six Sigma and Beyond: Statistics and Probability, Volume III
ISBN: 1574443127
EAN: 2147483647
Year: 2003
Pages: 252

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net