Page 444 - Battleground The Media Volume 1 and 2

Rat ngs |

samples of only a few thousand voters, so too do ratings firms project the media
consumption habits of the population from representative samples of just a few
thousand media consumers.
A key question that frequently arises in the ratings world, however, is how
confident can we feel in the accuracy of these samples? We have certainly seen
political polls get election outcomes wrong in recent years, so we know that sam-
pling is an inexact science at best. Might there be reason to believe, for example,
that there are too many men in the sample, or not enough African Americans?
This kind of sampling error can arise for any number of reasons. Some cate-
gories of people may be less inclined to agree to participate in the measurement
process. Young people, for instance, often are less willing to have their radio lis-
tening measured than older people. Similarly, Hispanic households have proven
less likely to participate in the TV measurement process than white households.
The more people who refuse to take part in the measurement process, the less
likely that the sample that ultimately is created will accurately reflect the popu-
lation as a whole.
Some segments of the viewing or listening public might be neglected for a
wide range of logistical reasons. Until recently, for instance, Nielsen Media Re-
search did not include college dormitories in its sampling. Similarly Web ratings
services often have found it difficult to measure the behavior of workplace Web
users since most companies have tended to forbid the placement of the neces-
sary monitoring software on workplace computers. These shortcomings limit
the extent to which the sample can accurately reflect the population as a whole.
If the sample does not accurately reflect the population as a whole, then the
accuracy of the ratings produced from this sample becomes questionable.
There are other potential sources of inaccuracy beyond sampling. What hap-
pens, for instance, if large percentages of the people who receive radio diaries
don’t bother to put them in the mail at the end of the week? Or if some people
wait until the end of the week to fill their diary out but can’t remember accu-
rately what stations they listened to during the week and for how long? Or what
if somebody has a friend who is a DJ at a local radio station and to help that
friend out decides to write in his diary that she listened to her friend every day
for three hours when in fact she didn’t? These kinds of examples illustrate other
very important potential sources of inaccuracy in the ratings data that the media
and advertising industries rely upon every day.
What does it mean if the ratings are inaccurate? It is around this question
that some of the most intense controversies surrounding audience ratings have
arisen. When we consider that programmers and advertisers make their deci-
sions based on ratings data, if the ratings are inaccurate, these organizations will
make decisions that do not accurately reflect the tastes and interests of the view-
ing or listening public. Inaccurate ratings can, for instance, make a television
show appear to be doing worse than it actually is. For instance, for years, NBC
complained that the fact that Nielsen Media Research did not measure view-
ing in college dorms meant that the ratings for Late Night with Conan O’Brien
(which is very popular with college students) were being undercounted, and
that the show was actually much more popular than the ratings reports would