Originally Posted by
themaster24
20,000 is roughly .00006% of the US population, and even at the supposed estimate that around 1/3 of the population are TV watchers, that's still .00018%. Purely on that fact alone, for something like TV, that is not a very good sample size. Just because 20,000 is a good sample size for some statistical studies, doesn't mean it is for all. TV is far more complex from this standpoint than a yes or no poll, whether or not a certain medicine has severe reactions, or who will you vote for. There are very limited data points the be looked for in these kinds of studies, and thus, smaller sample sizes are viable.
TV, on the other hand, has a vast swath of data points, far greater than 20,000 can properly handle. A single person can only watch one show live (generally speaking), but that doesn't properly indicate engagement for the whole spectrum of shows. Look at TV shows that get cancelled for "poor" ratings, and then millions upon millions of fans appear, and then suddenly another network picks up the show.
Nielsen is not only woefully outdated, they are also a monopoly in this business. No one else exist in the viewership game, therefore allowing Nielsen to do whatever it wants without reproach. That's what makes the figures Nielsen puts out the word of god in the business world, even when many executives in the TV industry have openly admitted the system is inaccurate and antiquated.