How much observational data is enough? An empirical test using marital interaction coding.

Using three different samples of couples (clinic, nondistressed community, and engaged), we found that 15 minutes was sufficient to witness enough behavior to make reliable (i.e., internally consistent) estimations of most Rapid Marital Interaction Coding System (Heyman & Vivian, 1993) code frequencies. Ten minutes is sufficient for many codes of interest. The ease in which "how much time is necessary" calculations can be made should entice behavioral investigators from a variety of content areas to publish such figures. By empirically investigating a factor that in most fields becomes reified through convention, investigators can conduct observational research that is both maximally efficient and maximally scientifically defensible.