1. THE PURPOSE OF SURVEY RESEARCH METHODS

The aim of survey research is to measure certain attitudes and/or behaviors of a population or a sample. The attitudes might be opinions about a political candidate or feelings about certain issues or practices.

2. FOCUS

Survey research focuses on naturally occurring phenomena. Rather than manipulating phenomena, survey research attempts to influence the attitudes and behaviors it measures as little as possible. Most often, respondents are asked for information.

3. TYPES OF DATA

Survey research is primarily quantitative, but qualitative methods are sometimes used, too.

4. SAMPLING

Once in a while, a researcher may be able to gather data from all members of a population. For example, if you want to know what a neighborhood thinks about a local land use issue, you may be able to measure all residents of the neighborhood if it is not too big. However, most of the time, the population is so large that researchers must sample only a part of the population and make conclusions about the population based on the sample. Because of this, gaining a representative sample is crucial in survey research.

Some common sampling strategies:

1. Simple random sampling. Members of the population are drawn at random to be in the sample. Each member of the population has an equal chance of being in the sample. Think of putting the names of all the possible survey respondents into a hat and drawing them out one by one to build your sample.
2. Stratified random sampling. Strata (sub-groups) are identified and respondents selected at random from within the relevant strata. For example, if I want to know the extent of certain health behaviors among the students at my college, Metropolitan State College of Denver, I would identify the relevant dimensions. These might be day or night students (since these are two fairly distinct sub-populations at MSCD) and major (since Letters, Arts, and Sciences majors tend to be different from Business majors). Thus, I would have 4 sub-groups: day students in Business, day students in Letters, Arts and Sciences, night students in Business, and night students in Letters, Arts and Sciences. (It turns out that MSCD's Office of Institutional Research has compared such a sampling strategy to the population characteristics and found it works.) Then, I would randomly choose respondents rom within each of these four groups. The step of stratifying gives me a more targeted sampling strategy.
3. Proportionate samping. This imposes the constraint that the sample must reflect the same proportions of sub-groups as is found in the population. For example, I could insist that my samples have the same proportion of traditional-age students (18-22) and non-traditional students as the population of MSCD students has.
4. Non-probability sampling. This is any procedure in which the sampling strategy does not give a representative sample. Examples include convenience sampling,where the sample is made up of those whom it is most convenient to survey, say one's friends or people who pass by a certain street corner; self-selected sampling, in which the respondents get to choose whether to be included in the survey, such as leaving questionaires at a table in a public place; and snowball sampling, in which previous respondents recruit subsequent respondents.
   
   Note that although these non-probability sampling strategies do not yield representative samples, they may still be useful to researchers in gaining a preliminary picture or as a pilot project.
   
   

5. POSSIBLE SOURCES OF BIAS IN SURVEY RESEARCH

1. Demand characteristics. Respondents tend to say wat they think the researcher wants to hear.
2. Acquiesence. Respondents tend to say "yes" more easily than "no."
3. Reactivity. Thinking about the questions tends to change respondents' opinions. For example, you may not have thought much about environmental damage until a survey asks for your opinions on rainforest depletion.
4. Response Bias. Some people tend to answer more positively or in more extreme terms. If there is a consistent tendency for one group to give more extreme responses and a consistent tendency for another group to give more middle-of-the-road responses, we might mistakenly conclude they have different opinions. In fact, we may only be observing a bias in their response tendencies.