Sampling methods in behavior research.
Animals perform a continuous stream of behavior throughout their lives. Because their behavior is not random, appropriate sampling methods can be used to obtain data that accurately reflect the actual behavior and are valid for answering research questions. Answering questions related to several variables assists in narrowing the choices of sampling methods. First, a determination must be made of what behaviors to measure. If the behaviors are few and easily measured, then All Occurrences Sampling is the method of choice because it generates accurate frequency and duration data through continuous recording. Sequence and Sociometric Matrix Sampling are specialized types of All Occurrences Sampling that are restricted to sampling intra- or interindividual sequences and social interactions (e.g., agonistic), respectively. Second, if who (e.g., specific individual, sex, or genotype) performs the behavior is a major component of the research question, then consideration should be given to Focal Animal (Pair, Group) Sampling. Third, if when or where the behavior is performed is of interest (e.g., activity budget), then Instantaneous or Scan Sampling can often be effective. Ad libitum Sampling does not produce valid data for analyses, but it is useful when formulating and fine-tuning research questions. One-Zero Sampling is not recommended except when the research question relates to the presence or absence of behaviors only. Other factors to consider in selecting a sampling method are duration of the behavior (event or state), desired scale of measurement (nominal, ordinal, interval, or ratio), and logistics (e.g., time, and equipment and facilities available).