V3rmillion how to do a survey

Conduct a survey | Step-by-step instructions

Conducting your own survey as an empirical study is the supreme discipline forscientific work. In both quantitative and qualitative social research, the survey is still the most common form of survey (cf. Baur / Blasius 2014, p. 54). A certain number of people are asked about one or more topics. When well planned and carefully carried out, a survey produces reliable scientific results.

Conducting a survey: preparation

Before you really start som, you should make a (time) plan, what everything has to be done, in which order it has to happen and how everything can be organized. The following points must be observed right from the start:

Define subject

First of all, it is important to clarify on which topic the survey should be carried out. For scientific work - e.g. B. for a bachelor thesis - there is a Research question in the room from the Hypotheses that are to be checked. These serve as a guideline for the orientation of the survey questions.

Literature research

Is this theme of course, you should research which publications are already available. The Literature research is an important part of the preparation to get an overview of the chosen research field. In addition, any research gaps can be discovered, which can be closed with a survey.

Stake out the theory framework

The theoretical framework for the research should also be clearly defined. One or more theoretical approaches can be selected for this. Depending on which one you choose, they lay another cornerstone for the orientation of the questions.

Define a sample

Often it is not possible due to time and financial resources to survey the entire population, for example all eligible voters in Germany, on a topic. In order to avoid a census, but still achieve a representative result, there is the sampling. So you only question a part of the population, but this should represent it as well as possible so that conclusions can be drawn about all of them. There are different variants for drawing the sample.

On the one hand, there is the random sample, in which the desired sample size is drawn purely randomly from the population and every potential respondent has the same probability of getting into the sample.
On the other hand, non-random sampling can be used. It is used when the research field is difficult to access and the population is not exactly known. An arbitrary selection can be made here, for example using the snowball principle. You give the questionnaires to someone in the field and ask them to pass them on to others in the group to be examined that they know.

The much better variant of the selection process is the random sample (see a Overview of the selection process at the University of Cologne). The results are much more accurate than those of a non-random selection.

Survey design

Next you have to clarify how you want to interview the survey participants. In the face-to-face questionnaire, as the name suggests, questions are asked face to face, while the participants are contacted by telephone in the case of a telephone survey. There is also the possibility of a written-post survey as well as an online survey, in which the participants receive the questionnaire by post or email and should send it back. Often there are factors, such as the size of the sample, that suggest a design. A “mixed mode design” from several of the aforementioned types of survey is also possible, for example to prevent a high refusal rate (cf. Blasius / Reuband 1996). Sometimes the universities also offer assistance (cf. ITS manual of the University of Kassel).

Conduct a questionnaire for the survey

Once the framework and the design questions have been clarified, the questionnaire can be created. There are different types of questions that can be used and that will be used depending on the focus of your research. It is also important to check whether the questionnaire worked out works for the desired data in the field.

Open or closed questions

While the qualitative interview There are only open-ended questions with no answer specifications that should encourage the respondent to tell their own story, quantitative interviews are much more standardized and use closed questions, for which the respondents have given answers to tick (cf. Raithel 2008, p. 69). These should be mutually exclusive (be exclusive) and contain all answers (be exhaustive) so that the respondent can find “his” answer.
Due to their standardization, closed questions are much easier to evaluate than open questions, for which the content has to be evaluated using qualitative analyzes.

Questions of opinion or fact

The type of questions asked depends on the research interest. With opinion questions, the attitudes on certain topics are queried. Fact questions relate to the behavior and characteristics of the participants. In many opinion-dominated questionnaires, there are also factual questions in the form of socio-demographic questions at the beginning or end of the questionnaire.

The pretest

Before going into the field, the completed questionnaire should be tested. The most suitable test subjects are people who are as similar as possible to the actual respondents. The conditions should also be the same as for the planned survey. A so-called pretest can uncover construction errors. For example, whether categories overlap or not enough processing time is available, but also whether there are problems understanding questions or answers. These can then be adapted and improved for the "real" survey.

Conduct field phase in survey

Once the preparations have been completed and the questionnaire has been tested, the time has come: the actual survey begins. In doing so, it is important to adhere precisely to the previously defined implementation parameters. In this way you can guarantee the scientific work in the surveys and generate data that then lead to high-quality results in the analysis.

Conduct survey: data acquisition, data preparation, data analysis

The completed questionnaires are available, but a few steps are still missing in order to be able to present the results. Back at the desk, the first thing to do is to systematically record the data. There is also a wide range of software, especially for quantitative data, such as STATA or SPSS, into which the data can be entered. The programs also offer the option of preparing the data, for example to sort out inconsistent answers or to clean up invalid information. For the subsequent analysis, you can combine the available results with a few clicks and put them in relation. Most of the tools also offer clear forms of representation, such as Graphs and tables or charts, for the presentation of the results.

The survey is the most popular form of survey in social research. Good preparation lays the foundation for creating a survey. This includes finding a topic, a thorough literature research, setting the theoretical framework, the sample, the selection process and the survey design.
When creating the questionnaire, there are different ways of asking the participants. Depending on the research interest, these can be used. After creation, a pretest should be carried out to test whether the concept worked out works or whether it may still have to be adapted.

In order to generate the most reliable scientific data possible, it is important to adhere to the previously defined guidelines in the field phase and to carry out the survey professionally. For the subsequent data acquisition and data processing, there are various programs that make the work easier and often also offer solutions for the analysis and presentation of the results.


Baur, Nina / Blasius, Jörg (2014): Handbook Methods of Empirical Social Research, Wiesbaden.

Blasius, Jörg / Reuband, Karl-Heinz (1996): Face-to-face, telephone and postal surveys: response rates and response patterns in a city study, in: Kölner Zeitschrift für Soziologie und Sozialpsychologie, pp. 296-318.

Raithel, Jürgen (2008): Quantitative Research: A Practical Course, Wiesbaden.