A Declining Television Audience?

At the MRG Evening Meeting on Monday 25 April, the subject under discussion was BARB, and the apparent decline in television viewing. First to speak was Sue Read of the Time Exchange. Her paper was called “The Meaning of the Universe”.

Sue talked about the ways universes are calculated, and the effect of the smoothing methodology on universes, and therefore the seeming changes in impacts.

In August 1991 the new Audience Measurement System was introduced. From August 1991 to March 1993, the straight method of calculation was used to obtain universes; between April and July 1993 a two year rolling average was taken, and between August and September 1993 the smoothing model was introduced. The smoothing model was adopted because the two year rolling average method gives out of date figures.

The smoothing model, as well as being more up to date also uses information going back as far as August 1990. However, altering the calculation methodology (even if it is improving) means that it affects the universes; for example, AB homes have dropped by 5%, and can even cause up to 20% difference on some regional audiences.

To sum up, Sue said that the new smoothing method:

• caused a change in data
• is more responsive, but also unstable
• produces regional fluctuations

A decision on universe calculation methodology has to be made by June. Sue’s final plea was “Let’s not seek perfection but practical compromise.”

Next to speak was Peter Bowman of WCRS; his paper was entitled “Is TV Viewing Falling?” His initial response to his own question was yes, audiences are in decline; last year the average person watched one hour less television per week than in 1992. However, Peter pointed out, by adjusting the viewing figures for VCR and satellite viewing, the average hours viewed per week since ’85 is 25.9 hrs. 1993 data is less than 1% below that (this average excludes 1992, when viewing was 3% above this figure).

This means, Peter believes, that there is in fact no crisis among viewing merely that 1992 is the high watermark of TV viewing in this century. Those people who expected a boom in viewing with the introduction of satellite were wrong. However, those who are pessimistic and predict that the audiences are going downmarket are also wrong; in fact among ABC1 viewers there has been an increase in viewing. Another important point to remember is that peak viewing has been unscathed; these losses have come from off-peak viewing times. Peter summarised, “Some individual people, rather than households are watching a little less”.

The reasons for the small decline are several. Peter believes that these may be attributable to the BARB methodology; one reason may be the time limit imposed on including VCR playback – the time limit of 163 hours should be taken into account. Also there is no method of measuring out of home viewing, i.e. MTV or sport seen on satellite in pubs, although this out of home participation is included on the NRS. Also, the alteration in universes means that the decreases seem worse.

The conclusions were that a slight decline in audiences suggests that TV audiences may be static in the future, rather than the boom sometimes suggested.

Peter Doe of RSMB’s paper was entitled “Does the BARB panel represent Satellite Viewing?” He began with the simple equation,

Accurate Universes + Balanced Panel = Representative Data

To calculate the Satellite Universe Model, SMATV, Cable and Dish are all measured. The sampling error on the monthly estimates is plus or minus 150,000; when this is compared to the monthly net growth of 40,000 then it is obvious that accurate growth data cannot be assured.

The panel over the network is balanced and reliable; however, by individual ITV areas it is not so reliable. For main audiences the profiles are not controlled, which means that specific targeted channels suffer more; it is judged a “reasonable measure” for large audiences.

For sub categories the universes are not controlled and the samples are too small; this is judged “not a good idea”.

Bill Meredith of BARB gave details of the Coincidental Study carried out in November and December of 1993. The objectives of this particular study were to measure the effectiveness of BARB’s own metering, more especially if the panel are pressing the buttons properly. The study also set out to check their accuracy in measuring commercial break viewing. The method used for the study was selecting 40% of the panel and conducting telephone interviews, and then checking the answers by what the electronic data reported.

The results showed that 92% of respondents used the buttons correctly, and that there was little difference between the programmes and the commercials. The reporting of non-viewers (i.e. people who have left the room) was not so accurate – 88% mistakes made both ways mean that the reported viewing index is 99.

When this data is broken down by audiences, the result across most was accurate, the lowest being among C2s, Bill suggesting that they perhaps tend to forget to press the buttons. The findings showed that commercial breaks do not reveal any deterioration in button pushing.

Questions from the floor were led by Mike Goodman, who wanted to know if all regions were as accurate as each other in button pushing – Bill responded that in this particular survey the sample sizes were too small to see.

Michael Stewart pointed out that television viewing is very dependent on the weather, and asked if there was any point in comparing year on year data considering the weather changes likely. Sue Read’s response was that they do in fact monitor the weather, but it has to be very extreme weather to cause any change.

Hugh Johnson of Channel 4 asked if universes should be kept constant over the year – Sue Read pointed out that the changing of data constantly is not healthy or helpful. The final question from Hugh Johnson was what the future consequences would be for TV. Peter Bowman did not think that viewing is falling enough to affect expenditure, because the reach is still there.

There will be an MRG Champagne Tasting Evening on May 23, and in June an NRS Evening.