Sampling television programs for content analysis of sex on TV: how many episodes are enough?

Content analysis is widely used to analyze sexual themes and messages on television. Although researchers generally use a sampling procedure to identify programs for analysis, there is currently no gold standard for the number of episodes needed to establish a valid, representative sample. This study sought to explore how many episodes of weekly television programs should be examined for an accurate assessment of sexual content present throughout the season. Various sampling strategies were used to select episodes from programs that ran in the 1998-1999 season, and results were compared with data collected from all episodes of these programs to test representativeness. Findings suggest that to provide a representative assessment of sexual content on television programs for behavior-level variables, a random sample of five episodes of weekly television programs is sufficient, while three episodes may be acceptable if variation is minimal. A random sample of seven episodes is needed for character-based analyses.