Sampling Techniques and Sampling Error in Naturalistic Observation: An Empirical Evaluation With Implications for Cross-Cultural Research

Abstract

Naturalistic observation plays an important role in cross-cultural research, and in virtually all cases observations are made of popu lation samples, not whole populations. A variety of sampling strat egies have been devised for naturalistic fieldwork, but deviations from actual target populations differ widely depending on the strategy used. This study uses two sets of continuous observations of child touching behavior to empirically evaluate these various sampling strategies. We find that random samples are the only ones that reliably represent the target populations. This finding has a number of implications for cross-cultural research, primarily that (a) data collected using differing sampling techniques are likely to vary in their representativeness, making samples gathered with different techniques difficult to compare; and (b) because the repre sentativeness of any sample is partly a function of the nature of the target population, careful attention should be paid to the definition of target populations.