S4C Viewing figures.

It is a sad state of affairs that the Department of Culture, Media and Sport appear to have based their decision on cutting S4C funding partly if not fully on viewing figures of the channel. I decided to try and get a better understanding of how this works having read Brokentv's excellent blog where it describes S4C's viwing figures as:

But what of S4C? After all (scoff! Guffaw! Etc!), doesn’t that have No Viewers At All? Sadly for fans of lazy topical comedy (and hey, you’re reading this blog), it doesn’t do too badly on our grand list, placed in 69th position – that’s above Comedy Central Extra, ESPN (which has exclusive live Premier League football) and UKTV’s Home. Not bad for a channel targeting the 1% of the UK population able to speak Welsh fluently (and no, there’s no analogue TV in Wales any more, so none of that figure is boosted by English-language Channel 4 programming, as used to be the case. Everyone in Wales can view ‘proper’ Channel 4 if they want to nowadays).

The definitive guide to understanding how viewing data is collected is the Broadcaster Audience Research Board (BARB) website
and this is how the process works:

BARB (Broadcasters’ Audience Research Board) is the organisation responsible for providing the official measurement of UK television audiences.

BARB commissions specialist companies to provide the television audience measurement service on its behalf – RSMB, Ipsos MORI and TNS.

BARB is responsible for providing estimates of the number of people watching television. This includes which channels and programmes are being watched, when they are watched and the type of people who are viewing at any one time. Viewing data is collected second-by-second and delivered on a minute-by-minute basis for channels received within the UK. The channel viewed the longest in a clock minute is attributed the viewing of that minute.

Viewing estimates are obtained from a panel of television owning private homes representing the viewing behaviour of the 26 million TV households within the UK. The panel is selected to be representative of each ITV and BBC region, with pre-determined sample sizes. Each home represents, on average, about 5,000 of the UK population.

Panel homes are selected via a multistage, stratified and unclustered sample design so that the panel is representative of all television households across the UK. A range of individual and household characteristics (panel controls) are deployed to ensure that the panel is representative. The prime control is a 20 cell matrix, made up of means of TV reception, life stage (pre-family, young family, older family, post family and retired), and social grade. As estimates for the large majority of panel controls are not available from Census data it is necessary to conduct an Establishment Survey to obtain this information.

RSMB is responsible for producing the sample design, including maintaining the panel controls to ensure their relevance to the panel. RSMB also conducts continuous quality control checks on BARB’s behalf.

The BARB Establishment Survey has the purpose of measuring the characteristics of UK households (demographics, viewing equipment, etc), producing universes for panel control and weighting purposes and providing addresses from which to recruit the panel.

The survey is carried out on a continuous basis and involves some 53,000 interviews per annum. It is a random probability survey, which means that every private residential household within the UK has a chance of being selected for interview. Interviews are conducted face-to-face using Computer Assisted Personal Interviewing (CAPI) and take, on average, 20 minutes to complete. The survey ensures that any changes taking place in the population can be identified so that the panel can be updated and adjusted to ensure that it continues to reflect the television owning population. In addition to being the prime source of television population information, the Establishment Survey also generates the supply of addresses from which the panel is recruited.

The BARB Establishment Survey is conducted by Ipsos MORI, which has responsibility for contacting each household selected for the survey and conducting interviews.

When a household agrees to join the panel their home has all their television sets, PVRs, DVDRs, VCRs etc. electronically monitored by a meter. Each piece of equipment that is connected to each TV set in the home is also connected to the BARB meter, which electronically monitors the equipment and determines which one is feeding the TV screen at any point and what it is doing. Over 30,000 devices, including equipment that is identified but not measured by BARB (games consoles, for example) are connected into BARB meters. Each TV in a home is connected to its own meter which holds an electronic record for the set. The meter is a small box which is put close to each television set and connected to it. The meter automatically identifies the channel that the panel member is viewing.

The company responsible for recruiting panel homes and installing the metering equipment is TNS.

Going by the figures quoted above; that one household with a BARB meter represents 5000 households then we can presume that there approx 5,200 households in the UK with BARB meters fitted to measure viewing, which includes S4C. In Wales there would be approximately 256 households but I gather there are an additional 200 households selected in Wales specifically to measure S4C viewing figures... which is all very well up to now.

The problem that now arises is how well do RSMB/Ipsos MORI and TNS allocate the BARB meters particularly to Welsh speaking households? Well according to an industry insider who has an intimate knowledge of how viewers are selected which includes "possible" S4C viewers; describes the system as 'unreliable' and 'underestimates' S4C viwers.

Basically what happens is that Ipsos Mori interview potential panelists and for the S4C panelists they are asked whether there is someone in the household who speaks or is learning Welsh, now bearing in mind that every child in Wales is a learner it is likely that all 200 households surveyed would answer yes to that question. Panellists are also incentivised to claim they speak Welsh by the offer of monthly High Street vouchers which add up to the cost of the BBC licence. Even if the children who were learning Welsh watched children's programmes on S4C it is unlikely they would be counted in the viewing figures. Apparently technicians who fitted the BARB meters were regularly reporting to their TNS managers that households who had S4C meters fitted did not speak Welsh but they ignored the problem because of the expense incurred in recruiting new panel members.

So here we have it; government decisions affecting the Welsh language made on wholly inaccurate evidence. Hopefully Alun Ffred and the Assembly's Culture Committee would seek to investigate these allegations further.

NB - As a matter of interest how many Welsh speaking household that we know have a BARB meter? I know of none.