It is not possible to create a high quality web questionnaire without considering the general principles of web performance and survey methodology:
- Not considering the general principles of web performance such as rudeness, inappropriate language, language not being adapted to respondents, inappropriate text on the introduction (when creating short surveys text can be redundant if it is on its own page) or closing page, bad grammar, technical deficiencies, weak flow, insufficient structuring of content, unfitting design, unsuitable use of multimedia, inappropriate promotion and publishing of the survey, unsuitable electronic survey invitations, violation of data privacy and secrecy principles, violation of the basic rules of web etiquette, etc.
- Without considering the general principles of survey methodology it is not possible to develop an expert approach that is distinguished by exhaustiveness and precision when forming theoretical background, understanding research process stages, understanding ethical standards, as well as the correct assessment of the human resources and time required for the creation of the questionnaire. In this sense it is especially critical – and exceedingly typical – to abandon or underestimate the role of questionnaire testing.
In this sense the background process of creating a questionnaire is especially important. In completely simplified form we can present it on the next case:
When creating a web survey the first step - as with any social science research - is previous conceptual work:
- In the beginning it is necessary to formalize and write down goals of the research. This is also reasonable in the case of the simplest researches. If we are interested in the satisfaction with studying on a certain faculty, we write down adequate goals such as researching satisfaction for the purpose of a student council or for needs of faculty or for the needs of an evaluation body (eg. NAKVIS). When conducting more sophisticated researches we help ourselves by defining goals with qualitative research such as expert opinion or in depth talks with client.
- On the basis of defined goals basic content sections are then structured. If we are interested into evaluating study for the needs of a faculty, content blocks can be: (a) general satisfaction with conditions for studying, (b) satisfaction with individual subjects and (c) satisfaction with lecturers.
- Each content block is then divided into components. Section of satisfaction with general conditions can for example be divided into (a1) satisfaction with classrooms, (a2) satisfaction with technical equipment, (a3) satisfaction with personnel and services in various departments such as (a3i) library, (a3ii) student office, (a3iii) dean’s office, (a3iv) computer center etc., (a4) satisfaction with study system and conditions, (a5) satisfaction with administrative work, (a6) satisfaction with IT support etc.
- Individual components are then transformed into questions. Before that we can also transform them into subcomponents (eg. different and more in depth aspects of library functioning). When researching opinions we demand more indicators, very often at least two, for each individual concept (when we are interested into the view to a certain service, we can ask a question about agreement with certain sentence such as "I like the functioning of the library" and "I would recommend services of the library to friends"). In the case of asking about facts one question is often enough (eg. "Did you visit the library in the last month")
- Designing the final question is often a complex process where it is worth following the established principles such as use of common language (without jargon), simple sentences, avoidance of negations (especially double negations), transition from general to specific, treatment of just one dimension in the question etc.
- Design of a whole questionnaire is also a complex process. We usually begin with interesting and shock questions. The rule of thumb is that a certain flow from general to specific is established. Sensitive questions are usually near the end and demographic questions are in the end. One of main questions if the length of a questionnaire where judgement about what we need is essential.
- Questionnaire must then be tested. At first author/researcher can put him/herself into the role of a respondent and test if all questions are understandable. Then questions can be tested in the inner circle of friends and co-workers. More complex surveys also require formal testing. A lot of different methods can be implemented during formal testing.