Also, different researchers have used different definitions of some basic concepts such as TV violence which has caused some confusion when widely different results have been attained.
The basic unit is called the unit of analysis. Defining this is important and difficult.
Once the unit is defined, it is important to set up adequate categories for analysis and you should begin with too many rather than too few.
The coders must be trained and then the results tested for inter-coder reliability.
The validity of content analysis studies causes the most problems. Face validity is the most we strive for in many cases.
Inferential statistics can be useful to help generalize from our sample to the "universe" of shows, characters, acts, etc. They are usually not helpful if the total universe is analyzed.