Survey research is the bread and butter of mainstream social science. Surveys allow us to draw on representative samples to learn about people’s beliefs, behaviors, and experiences—at least this is the goal for scientific research. Unfortunately, as we have discussed previously, many people do not understand the scientific logic and process that underlies surveys as a research design. Since virtually everyone has taken a survey in some form or another, there’s a general understanding that any data gathered from a survey is useful. That is not always the case, however. Surveys cannot be separated from good instrument design (the survey itself is referred to as the “survey instrument”), nor can they be separated from our earlier discussion of censuses and samples. Just because you get information from a survey does not mean that information is useful—if the survey is poorly designed and/or if the survey does not rely on representative samples (or a significant response rate from a population census), then nothing “learned” from the survey can be generalized to a population.
Good survey design is complex and linked to a theoretical literature. When should surveys be the research design chosen? How do you write a good survey? How do you make sure you get the survey to a representative sample (or a census)? We spend this chapter looking into the complexities of what it takes to do good survey research.
The survey is a nonexperimental design that ...