To specify what data source is to be used, and the dates to be included in the database, click on "SOURCE AND DATES":

 

 

 

To choose a data source, read about that in the topic, "Data Source".

 

So far, you have told TView the supplier of the data to be used and where the weekly DAT files are located. TView will tell you what weeks are available at the location you specified:

 

 

 

 

How Many Weeks of Data?

 

Now you need to specify which of these weeks are to be used for your new database. First, enter the number of weeks of data to be included in your new database:

 

 

Note: Be sure to read the next section for warnings on the use of periods longer than about 13 weeks.

 

 

 

Avoid Lengthy Spans of Weeks!

 

Nielsen Media Research advises its clients to not attempt reach analyses for periods longer than three months. We agree with Nielsen on this one. To remind you of this, TView will present a warning if you request a number of weeks greater than 13. There are some serious statistical issues that arise when one tries to analyze periods much longer than that.

 

The main purpose of television surveys is to prepare ratings reports for individual programs on individual days, and then to average this aggregate data for network and daypart summaries. But remember that the television data we use in planning is based on samples of real people, who are recruited to be in a study panel for some period. This makes it possible to do reach and frequency analysis as well as just ratings calculations. In the U.S., homes might be in the national Nielsen panel for two years.

 

But if you attempt to do calculations for a very long period, let's say 52 weeks, not all homes have supplied data for the entire period. Consequently, certain types of programming, especially anything that has a seasonal strength or character to it, may not be reflected adequately by a person panel that has changed so much.

 

Although homes are encouraged to stay in the panel for a long period, some homes may drop out much earlier than that. Sometimes this is the choice of the persons in the home, but it could be due to some kind of unusual technical problem. In any case, the presence of many homes that stay in the panel for a shorter period make long-term analysis even dicier.

 

Even more disconcerting, the kinds of homes that leave the panel have been shown to be systematically different than other homes, both in terms of demographics and viewing. Smaller families, single individuals, and older persons are more likely to stay in the panel, while bigger, younger families have more of a tendency to drop out. This means that older-skewing programs and networks might get an artificial "boost" from the use of longer periods, while family-oriented or younger-skewing programs may be disadvantaged by doing this.

 

There's more. Homes in which there is a lot of TV viewing, or where there are complicated viewing environments, are also more likely to drop out of the panel. This means that if you attempt a very long-term analysis, information will be based more and more on a panel of people that does less TV viewing than the universe at large. Remember that in collecting viewing data, Nielsen (and other firms in other countries) use a set of "edit rules" to make sure that only reasonable, valid information goes into their reports. When there is more viewing or more equipment, or households move, or there are more changes in the composition or the equipment in a household, there is more opportunity for things to go wrong in data collection. (As one Nielsen person says, the easiest way to be always in-tab is to never turn your set on.) This is not as harmful to ratings reports, because Nielsen (and the other companies) go to lengths to try to make the panel reflect the universe. But over a lengthy period, these changes in the nature of the panel can be significant.

 

OK, the bottom line: avoid long-term cumes! For the most accurate reflection of reach and frequency accumulation, limit your investigations to no more than 13 weeks.

 

We understand that from time to time planners like to show a single reach number as a boxcar result for a whole year, to be used in a client presentation. For that purpose, we can understand this desire, and recognize some possible value for this kind of use to give an overview of what a plan is supposed to accomplish. But avoid using such numbers for in-depth tactical media planning.

 

 

Include What Dates?

 

Next, specify the start and end dates of the weeks to be included:

 

 

As a default, the date range initially shown is the most recent four weeks of data available, once the source and path have been set. To change the dates, you can click on the individual components of the dates. You can also click on the to select dates through a calendar:

 

 

Note! The start date must always be a Monday and the end date must always be a Sunday!

 

If you change the number of weeks, the end date changes automatically (and vice versa).

 

If there is a problem with any of the date entries, TView will display a message.

 

If the path you gave does not include one or more of the DAT files needed for the dates you requested, TView will present a message like this:

 

 

If you get this message, either a) change the requested dates, b) install the required DATA files, or c) point to a different drive and directory that has the needed DAT files.

 

 

After the sources and dates are specified, you can define your dayparts and other components.

 

 

What's in the Data For Chosen Weeks?

 

TView also shows a list of broadcast and cable networks that are included in the data for selected weeks. That will provide you details on the weekly files that will be used, which is very handy to track down if any of the needed files have not been installed.  You'll also get a list of the networks in these files, and the dates for which their information is available:

 

 

A column labeled "Misc" identifies whether the weekly DAT file carries certain internal  features:

 

E: Extended station info (in MMRLD DATs since 1/29/2007, and AMRLD DATs since 5/6/2007)

L: enhanced telecast lengths

P: personal demos

C: cluster codes

v: Viewing data is compressed

M: file carries monthly weights

Q: file carries quarterly weights