Research Methods and Statistics

CHAPTER PREVIEW

Asking questions in survey research is an important aspect of research methodology. Surveys have become a fixture in modern life, with professional pollsters examining details of our lives and social scientists uncovering trends in attitudes and patterns of behavior. Most surveys rely on samples rather than on an exhaustive questioning of the entire population, which is a census.

Surveyors can choose from among different question types and content. The construction of the questions is probably the hardest part of a project because the form of a question influences responses, depending on the wording. So researchers need to take great care in creating questions in their survey research. In addition, respondents don't want to be seen in unfavorable light, so they may alter responses to make themselves look good and may tailor their answers to meet what they think are the expectations of the researcher. Fortunately, there are ways to avoid pitfalls in asking questions.

Finally, survey researchers prefer probability samples, and are wary of self-selected, convenience samples that do not represent the entire population. Researchers continue to develop new sampling strategies to overcome potential problems in current strategies.

Surveys: Answering Diverse Questions

We tend to take questionnaires for granted today, but they are really a relatively recent form of data collection. Late in the nineteenth century, G. Stanley Hall and psychologists in his laboratory made extensive use of questionnaires to study children's thought processes.

The first two listings of attitude research in the PsycINFO database were published in The American Journal of Psychology and Proceedings of the Society for Psychical Research, which is now considered a nonscientific journal. Both deal with belief in life after death.

Beginning a little over half a century ago, investigators began developing the theory and techniques of survey research. Surveys took place that increased our information base on various topics, but by today's standards they were methodologically suspect. For instance, in 1936, the now defunct magazine Literary Digest confidently, but inaccurately, predicted that Franklin Roosevelt would lose the U.S. presidential election in a landslide. This may be the most famous error in the history of research. The problem was that the editors of that magazine used a sampling method that led to an unrepresentative picture of the voting population.