Poisson cumulative probabilities of systematic errors in single-subject and multiple-subject time sampling

Previous studies have shown that both accurate frequency and unbiased duration with very small random error can be extracted from the three modes of time sampling under the following conditions: when observational interval length is (a) shorter than both the shortest behavior bout and the shortest interresponse time for momentary time sampling, (b) shorter than the shortest behavior bout and shorter than one-half the shortest interresponse time for partial-interval sampling, and (c) shorter than one-half the shortest behavior bout and the shortest interresponse time for whole-interval sampling. This paper suggests a method to use prior information to assess the a priori probability that these conditions are met with a chosen interval length.