Survey Research

I

f we look at the assignment/survey, a proper analysis requires us to asses all the choices made by the researchers who designed it. What are the necessary questions we need to ask ourselves? How does the setting or scene of a �firm’ instead of an �individual’ as a unit of analysis, influences our research design? How to deal with the content of the survey and minimize the survey errors? These are all questions that need to be addressed in order to conduct a methodological correct research that provides valid measurements and reliable results.

Our starting point is that we want to geographically map the existing business networks in different regions. Let us start with the research question as it is a part of the design effects. In general, the research question should not be too narrow and not be too broad. One should bear in mind that all concepts need to be defined clearly in advance (Babbie, the practice of social research, p. 122). Since the research question is not stated in the assignment it is not possible for us to asses the quality of the question.

Presentation is another important part of the design: make sure that a clear description is given about the research and ensure the respondents anonymity.

The research method is also part of the design effects. We want to geographically map existing business networks: meaning that we want to collect information about different regions i.e. groups at a certain moment in time. Therefore the research is a survey research for descriptive purposes using cross-sectional data. The easiest and most cost-efficient way to conduct this research is via a web-based survey. This is in line with the assignment-survey. We have to keep in mind that such a survey has disadvantages like self-selection and non-response. The problem of non-response is clearly evident in this survey, with a response rate of 9.1. %. Babbie refers to a minimum response rate of 50% as being adequate for analysis, while Malhorta and Birks (2000) are referring to a 15% response rate. Even when we choose to use the rule of thumb of Malhorta and Birks, the response rate of this survey falls short. In addition this survey, encounters the problem of self-selection. It is more likely that small firms, where each employee has a good overview of their activities and figures (easy to answer questions), and where they can not afford to buy second data would be more willing to participate in such a research.

The sampling provides a challenge. If we categorize firms into regions we are faced with the problem of non-exclusivity of the categories. A firm can have offices simultaneously actively participating in business networks in different regions. Stratified sampling has the criteria of exclusive groups; therefore this type of sampling is not valid for the research at hand. Cluster sampling provides a solution. First we sample regions and then we can sample firms by head office locations for example.

Having established the research design and taking into account the design effects we can now move on to the questionnaire. The questionnaire effects can be summarized in three categories; content of questions, the order of questions and the number/content of reply categories .

This research is directed to firms, involves facts and the questions concern the firms directly. Since the questions will relate to their business activities we can expect some questions to be perceived as threatening. Another factor that we have to take into account is time for answering since it time is scarce resource for firms. Our questions therefore should be short, easy, memory independent, unambiguous, about facts and taking into account the sensitivity of the information asked. The assignment-survey questionnaire has several questions that show errors in this respect . Babbie (“The practice of social research, p. 245-246) distinguishes the following;

Reply Categories

• Questions should be short, clear and unambiguous.

• We should avoid double-barreled questions (or more) .

• The measurements used should be precise and accurate (e.g. exact nr. of employees).

• Open questions are more flexible, but they require coding which is time