Read Viewing and Listening Rates

Datasheet by Jasper Keijzer, Utrecht University

In 2020 a collection of digitized Viewing Rates, Radio Rates and Commercial Rates was made available in the Media Suite. This information page provides additional information on the production of Viewing Rates specifically.

This document aims to provide insight into the process of data collection and processing that underlie the numerical rates reported. We believe that knowledge of the production of viewing rates can assist in understanding and interpreting the meaning of the measured values.

In this Fabeltjeskrant Tutorial (basic) and this Oil Crisis 1973 Tutorial (advanced) we provide steps to learn to work with the collection in the Media Suite’s.

This document will first describe the general methodological principles that underlie audience measurement. The second part of this document will identify and describe the historical periods of audience measurement. Finally this document will consider the accuracy of measurements in different historical periods from a Media Studies perspective.

The information in this document has three sources. First and foremost, this document serves as a summary of the methodological documentation by Stichting Kijkcijferonderzoek. Specifically the methodological description detailing both the historical periods and their differences has been a primary source. This document is publicly available and can be found here . Second, to clarify the practical procedures around the viewing rate diaries in the early period of audience measurement, beyond the scope of the methodological documentation, we conducted interviews with professionals tasked with producing viewing reports. Third, the reflections and considerations on the accuracy of rates and their relation to the activity of television viewing are connected to academic literature and will be referenced accordingly.

General methodology of viewing rates

Viewing rates are produced through a measurement of a TV Panel which serves as a representative sample of the population. Each year, 1200 households are selected to be part of the TV Panel. The viewing behaviour of the TV Panel is measured. Based on the demographics of the Dutch population, according to CBS, the data from this representative sample is then calculated to account for the entire population. Every viewing rate is therefore the product of a calculation based on the measurement of a representative sample.

Four historical periods can be distinguished:

• 1967-1974 Viewing rates diaries

• 1974-1987 Diaries combined with Intometer

• 1987-2002 Cable television and the 4900 peoplemeter

• 2002-present Time-shifted viewing and the TARis 5000

1967-1974 Viewing rates diaries The TV panel was the main source of information in the early days. There were 1200 respondents - not households, but persons. Measurement was carried out on a weekly basis, but the process of measurement itself took almost four weeks. 10 days before the broadcastweek the programming schedule was sent from the public broadcasters to Intomart GfK. These schedules were then manually typed into a computer and reformatted to viewing diaries. The respondents would fill in these diaries by checking a box if the program was watched more than halfway through, and by giving a rapport grade of 1-10.

Intometer The TV panel measurement was complemented with a measurement on the television set-level. A cassette recorder was placed behind the television that would collect a special audio signal that was transmitted through the broadcast at an interval of 5 minutes during programming and 30 seconds during commercial breaks. Between 1967 and 1974 these measurements were only executed on a separate sample of 200 televisions. Their results were used for the complementary data in the viewing reports and did not factor into the viewing rates themselves.

1974-1987 Diaries combined with Intometer From 1974 onward the TV panel and the television set-level measurements were conducted on the same sample group. This allowed for the entries in the viewing diaries to be verified by the set-level measurements. Between 1974 and 1987 the sample group which had Intometers connected to their televisions on top of their viewing diary duties gradually grew from 200 to 400 and to 600 by the mid 1980s.

1987-2002 Cable television and the 4900 peoplemeter The English Ratings company AGB developed a new device called the Peoplemeter 4900. This device would replace both the diaries and the Intometer, as it was capable of measuring viewing behaviour and television tuning at once. The Peoplemeter is a small box with its own remote control that was put near the television set. All persons in the household would register their viewing behaviour through the remote. The remote featured buttons that were connected to household members. The respondents were asked to register if they were in the room when the television set was activated. Every 15 minutes the device would check if the situation remained unchanged. The remote also features two guest viewing buttons, these were used when guests other than household members were also in the room when the television was active. Leaving the room meant logging out on the device.

An updated audio signal technology assured that the signal was inaudible to viewers with the latest televisions. The television set-level measurements and the viewer inputs would be sent electronically and automatically every night to an Icoder machine. By the next morning, the signal frequenties and the inputs were combined with the pre-loaded program schedule; producing viewing rates daily.

2002-present Time-shifted viewing and the TARis 5000 By December 2001 all of the peoplemeters were revoked and the new TARis 5000 meter devices were installed in the TV panels homes. Because of new systems in digital audio matching through VBI codes, advanced picture matching software and an ML code watermark for channels, it became possible to record time shifted viewing, DVD viewing and not much later also viewing through internet and second screen applications. The TARis 5000 meter is a more sophisticated version of the peoplemeter 4900, ensuring more accuracy through more advanced technology. But the input remains the same; the television tuning is measured through an audio signal sent through the broadcast and the viewing behaviour is recorded by the TV panel members themselves through a remote control.

Considering measurement accuracy

Since its inception, viewing rate measurement methodologies steadily improved but the underlying system was never altered. The system of “educated guessing” based on a representative sample and calculated results remains in place today. As well as the methodological combination of matching an audio signal to identify the television tuning and the manual input of TV panel members to register the viewing behavior of individuals. What has changed is the specific methods used and their necessary strengths and weaknesses.

The viewing diaries relied heavily on the involvement, honesty and memory of the respondents. This method has a number of weak spots related to human error. First, the respondents were instructed to enter their viewing behavior daily, although the diaries were delivered and collected on a weekly basis. Respondents could freely choose to fill in the entire diaries on Sunday night based on the memory of the week before, inviting the uncertainty of human memory into the fray. Second, respondents could alter their viewing behavior based on socially acceptable programming. The Dutch broadcast system was “pillared” through dimensions of religion. Catholics and Protestants were represented by different television production companies. Expectations about which program a household should watch, might have influenced which programs were reported as watched. The third weakness of the diary method was the inability to verify results.

The 4900 Peoplemeter relied less on human memory, by placing a focus on measurement frequency. The electronic meter allowed for the production of average viewing rates per minute. This meant that the respondents in the TV panel had a far more active role in the measurement process. Every time a family member entered or left the room, the Peoplemeter needed to be informed. In researching the peoplemeter in France, Bourdon and Meadal noticed that this meant a step away from human error and a step towards measurement fatigue. If the device received no entry for 15 minutes, a question would appear on the display: Who is watching? Respondents did not have the option to account for viewing retroactively. Instead, a period of no activity is understood as a gap. In the SKO methodological document for 2018 four types of gaps are identified and for each a calculation is given to account for the missed measurements.

As mentioned above, the TARis 5000 (peoplemeter) is primarily a more sophisticated version of the 4900 peoplemeter. Measurement occurred in a similar fashion, merely adding features and functions that would account for time-shifted viewing and external device use, such as dvd-players and gaming consoles.

What is television viewing?

Generally speaking, the history of television audience measurement consists of a diary phase and an electronic meter phase, each offering a different conceptualisation of television viewing in a more philosophical sense.

Before 1987 respondents were asked to note which programs were watched at least halfway through. This meant that a program was either watched or not watched. One could argue that this interpretation of television viewing entails the intentional watching of a specific program.

After 1987, with a better time based system, the figures in the viewing reports were based on a calculation of the average viewers per minute of the program. However, television viewing changed meaning. The threshold for measurement went from watching a program for more than half of the runtime to ‘being in the room with a television set’. This means that you could view television while reading a book, playing a board game and even without watching the screen at all. One could argue that this conceptualizes television viewing as being exposed to a television set. This raises questions about the meaning of television viewing from a media studies perspective. When did you watch a program more: if you watched 49% attentively or if you were in the room doing other activities during the entire runtime?

Legenda

Abbreviations - Meaning - Explanation

1. %KdH. - Viewer density (kijkdichtheid) Percentage of viewers within a certain category that viewed the program in relation to
   the total number of individuals in that category.

2. W or Wrd. - Appreciation (waardering) A grade 1-10 for the quality of the television program

3. S.D. or SD or Sdev - Standard Deviation A measurement of the diversion of figures in a sample.

4. 1834 - between 18 and 34 years Category based on the age of the respondent

5. 3545 - between 35 and 49 years Category based on the age of the respondent

6. +6 - above age of 6 Category based on the age of the respondent

7. N, O, Z, W. - North, East, South, West A division in viewing rates based on geography of the Netherlands