MAIN POINTS

Data in the social sciences are obtained in either formal or informal settings and involve either verbal (oral or written) or nonverbal acts or responses. A variety of combinations of these two settings for data collection and the two types of acts led to the development of the four major forms of data collection just alluded to. To a certain degree, research findings are affected by the nature of the data collection method used. Findings that are very strongly affected by the method used could be artifacts rather than objective facts. In order to minimize the risk of erroneous conclusions, two or more methods of data collection can be used. This is the essence of triangulation: When possible, hypotheses should be tested and variables should be measured with two or more data collection methods.

Modern social science is rooted in observation, and this technique of studying behavior has a number of distinguishing characteristics and/or advantages: 1) it is direct; 2) it takes place in natural settings; 3) it allows the study of people who cannot or will not report their own behavior; 4) it invites comparison of verbal reports and actual behavior; and 5) it facilitates analysis of the context in which behavior occurs. Whatever the purpose of the study and the observational procedure utilized, three major considerations are to be dealt with if the obtained data are to be systematic and meaningful: what to observe, when to observe and how to record, and how much inference is required.

The first significant consideration concerns a decision on what should be observed. Observation is particularly well suited for studying nonverbal, spatial, extralinguistic, and linguistic behavior.

A second consideration in observational studies concerns the timing and recording of observations. Researchers using observation must develop techniques to sample time periods. Observation requires the development of systems to categorize behavior; these systems may be inductive, deductive, or both. They must comprise a set of explicit, exhaustive, and mutually exclusive categories.

A third consideration in structured observational studies relates to the degree of inference required by the observer. The greater the inference, the greater the reliability problem, and the greater the need to train the observers in the use of observation systems.

The extent to which decisions regarding the types of behavior, timing and recording, and degree of inference are systematically and rigorously implemented is a criterion by which we can distinguish between controlled and noncontrolled observational systems. Controlled observational systems are typified by clear and explicit decisions on what, how, and when to observe; a noncontrolled system is considerably less systematic and allows great flexibility. Controlled observations are carried out either in the laboratory or in the field.

The most controlled method of data collection in the social sciences is laboratory experimentation; it involves the introduction of conditions in a controlled environment (laboratory) that simulates certain features of a natural environment. There is a distinction to be made between two senses in which any given experiment can be said to be realistic. In one sense, an experiment is realistic if the situation is realistic to the research participants: if it involves and has an impact on them. This kind of realism is commonly termed experimental realism. The second sense of realism refers to the extent to which events occurring in a laboratory setting are likely to occur in the "real world," and this type is known as mundane realism.

There are three sources of bias in laboratory experiments: 1) bias due to the demand characteristics of the experimental situation itself (when respondents are aware that they are being observed and that certain responses are expected from them); 2) bias due to the unintentional influence of the experimenters (when experimenters know what effects they desire from individuals, they may unintentionally communicate their expectations in various ways); and 3) measurement artifacts (when the process of measuring sensitizes participants or when measurements are imprecise or made at the wrong time).

A field experiment is a research study in a natural situation in which one or more independent variables are manipulated by investigators under as carefully controlled conditions as the situation permits. The main weakness of field experiments is limited control of intrinsic and especially of extrinsic factors. This problem is somewhat offset by field experimentation's advantages with respect to experimental and mundane realism.

A natural experiment is exactly as it sounds, namely an experiment carried out in an environment that has not been manipulated by the experimenter. Of course, this is not to say that the environment has not been manipulated at all. Relative to laboratory experimentation and field experiments, the key difference is that the manipulation of the environment is naturally occurring.