Two underlying assumptions of the Cultural Indicators approach to television research were examined, using data on the television viewing habits of 76 second grade, 150 fifth grade, 509 eighth grade, and 350 eleventh grade students in Perth, Australia. The assumptions were that commercial television presented an organically composed total world of interrelated stories produced to the same set of market specifications, and that television audiences viewed largely nonselectively and by the clock rather than by the program. The evidence on selection and habit in television viewing countered a "purely ritual" overstatement of the Cultural Indicators assumption. Two of the three content types most related to social reality (crime adventure and cartoons) were quite predictable from viewing habits. But habitual television watching itself did not account for all content-specific relationships, because viewing comedy and news programs related to other viewing habits without being related to beliefs about social reality. This evidence pointed to content differences (selection) as a key factor in television viewing even though habit could be considered an important antecedent to much of the viewing that produced the cultivation of a biased conception of the real world. Greater than average viewing times for cartoons and game shows were associated with perceiving the world as relatively mean and violent, while the reverse was true for drama and news viewing. (RL)