The purpose of this term paper is to show what different stages it takes to conduct a survey.
After defining the expression “survey” the uses and goals of survey are described. It will be explained how a questionnaire is constructed, in what form it has to be written down and which mistakes are often made and have to be avoided. In this context I will state different opinions of researchers and explain the dispute between two ways of conducting surveys, known as the qualitative and the quantitative researchers and will ponder which of these two contents should be preferred. Afterwards it will be defined how the data is collected before I explain how to analyse the collected data and expose the various different methods to make statements about the data. Here the most common computations are defined and exemplified with the help of some graphics. Afterwards these computations are all exemplified with the help of a sample-survey, which was prepared during the course at university. At last the most important ideas which I wanted to demonstrate through this term paper will be repeated very shortly as part of the conclusion.
2 Definition, uses and goals of surveys
Demoskopie, Teildisziplin der empirischen Sozialforschung, die durch Befragung eines repräsentativen Querschnitts der Bevölkerung nach einer statistisch verwertbaren Ermittlung von Meinungen, Befindlichkeiten, Stimmungen, Erwartungen und Bedürfnissen trachtet (auch Meinungsforschung oder Umfrageforschung genannt).
Standard ist die Anwendung des Repräsentativverfahrens, bei dem nach dem Zufallsprinzip per Interview zu befragende Personen aus Adressenkarteien der Einwohnermeldeämter bestimmter Regionen ausgewählt werden, wobei sich eine Stichprobengröße von 2 000 Auskunftspersonen als optimal erwiesen hat. Wahlprognosen beruhen in der Regel auf einem Quotierungsverfahren, bei dem den Interviewern bestimmte Quoten für die Anzahl der zu befragenden Männer und Frauen, Alter, Beruf und Herkunft vorgegeben werden. Daneben wurden noch spezielle Testverfahren entwickelt, mit denen ein- oder mehrdimensionale Skalen erarbeitet und Motivkonfigurationen sichtbar gemacht werden können. (…) (Microsoft ® Encarta ® Enzyklopädie 2005 ©).
2.2 Uses & goals of surveys
Today surveys play an important role in everyday life. If you read the newspaper you may find an article saying that “three in ten Americans keep guns in their homes” (Weisberg et al. 1989, p. 7) or another one with the lines “59 percent said police were doing a good job” (Weisberg et al. 1989, p. 7). Also the famous “Sonntagsfrage” (http://www.bpb.de/fsd/wahlometer/sonntagsfrage.html, 9.10.2006)gives us numbers and percentages to show how many people were willing to vote a specific party, if it was Election Day. All these numbers are the results of surveys.
The most common fields in which surveys are used are political polls, surveys in court, government surveys, consumer research, academic research and media polls (Weisberg et al. 1989).
The goal of a survey is to answer one or more specific questions. These questions can be divided into four broad classes of questions. The prevalence of attitudes, beliefs and behaviour, changes over time, differences between groups of people in their attitudes, beliefs and behaviour or causal propositions about these attitudes, beliefs and behaviour (Weisberg et al. 1989).
2.2.1 The prevalence of attitudes, beliefs and behaviour
Mostly surveys are used to analyze the frequency of specific attitudes, beliefs and behaviours.
We use a survey in order to see how many people have the opinion that the chancellor Angela Merkel has done a good job in her first hundred days in office (an attitude), what percentage of the people thinks that the FDP would be a better partner for the CDU (a belief) and to see how many unemployed people were looking for a job during the first quarter (a behaviour) of 2006.(Weisberg et al. 1989)
There are of course also other possible ways to make predictions in order to answer such questions, but most researchers believe that the best and easiest way to find out what people think is to ask them (Weisberg et al. 1989).
2.2.2 Changes over time
Taking in account the example above and knowing that fifty-three per cent consider the chancellor to have done a great job in the first hundred days this alone doesn’t tell us a lot. But if you ask the same question after the next hundred days of her work you can say whether her reputation is still the same or whether now more or less people consider her to be working well. It’s also possible to compare the fifty-three per cent with an older survey which was made after the first hundred days of chancellor Schröder’s time in office and so to be able to say which chancellor was considered to have done a better job (Braverman & Slater 1996).
2.2.3 Differences between groups of people in their attitudes, beliefs and behaviour
If the same survey is conducted twice and one time just women are asked and the other time only men it is possible to see how differently these two groups of society view Merkel’s work. Another possibility would be to compare unemployed and working people, people of different ages, different ethnic groups like Christians and Moslems or between people with a different colour of skin (Braverman & Slater 1996).
2.2.4 Causal propositions about these attitudes, beliefs and behaviour
Researchers are often interested in identifying the causes for different social behaviour, beliefs and attitudes. If we would consider an article which says that sixty-five per cent of the Germans think of the Government as having good ideas concerning the foreign policy with Iran and compare these with the Iranian people living in Germany, the author would probably guess that people from Iran living here would have more interest in and knowledge of this relationship between the two countries and it is more likely that the Iranian people in Germany would probably criticise the German foreign-minister more than native Germans would. If you now ask both groups the same questions and get the result that eighty per cent of theIranianpeople in Germany consider the German government to have good ideas, you can prove that the guess, the author had before, is wrong (Braverman & Slater 1996).
3 Questionnaire construction
The next step on the way to a good survey is the construction of the questionnaire. The question of how to create a good questionnaire is very important, because if the questions are not asked properly the results you get might be unusable.
The first important thing to look at is the form of the question. There are two forms of questions which are used in surveys: “closed-ended questions” and “open-ended questions”.
3.1 Closed- and open-ended questions
Closed-ended questions offer some given answers and the asked person then has to choose one or a specific amount of these. This is also known as the “multiple-choice question”.
Questions belonging to this group often appear as rating scales, where a question such as “What do you think of Angela Merkel’s work?” is asked, which can be answered with five different options, e.g. perfect, good, ok, bad or really bad (Fowler 1984 / Fowler 1993 / Weisberg et al. 1989).
Open-ended questions can be answered freely, so no answers to choose from are given. The asked person can answer in his / her own words and it is not important, whether these are one hundred words or just ten.
Closed-ended questions are easier to compare, because these questions directly categorize the answers. If open-ended questions are used the answers must be analyzed and categorized by co-workers of the enforcing company, which are then searching for similar answers. When open-ended questions are used it may occur , that one question is understood in a different way by the people answering the question than it was meant to be. With the help of closed-ended questions this problem occurs not as often as with open-ended ones, because with the given answers that can be chosen, it gets clearer what the given question aims at (Fowler 1984 / Fowler 1993 / Weisberg et al. 1989).
3.2.1 Incomplete and optional wording
The questions, which are asked, need to be asked in full sentences. If only “age?”(Fowler 1993, p. 79)is asked in stead of “What was your age on your last birthday?”(Fowler 1993, p. 79)it can lead to mistakes in the questionnaire. Interviewers or respondents might add words or change the questions, so that they are answerable for them, but when they do this in different ways the answers might not be comparable, because they were made for different questions.