One way they use is to have a cross-section of the viewing audience's tvs connected to a box which stores the information of the channel and times they were watching. Once a day this information is transmitted by phoneline to the reasearch company for analysis. As I have only come across a handful of these in a number of years of tv servicing, it seems a pretty small sample which suggests some degree of inaccuracy, and doesn't take into account 'passive' watching where the tv is just left on all evening.
Or another way is they send out these viewing pamphlets every week and you just fill em in on what youve watched and what radio youve been listening to I did it last year for about 6 months.
Alot of them base it on, say, 10,000 of the most random people. Anything from a web designer from manchester ;-) to a business chairman from london then use clever mathematics and statistics to calculate the gross watchedness (no, its not a word)
i heard that they calculate it using power usage. for example if you were watching corrie at 7.30 and at the break you put the kettle on then they'd make assumptions. this just proves that its a load of nonsense! there is no way they could know what any of us are watching! (hopefully!)
Smorodina is right. We have one of these boxes and you key in when you are watching the TV or not. I think there are 5000 homes on the pannel (thats about 15-20000 people). The people who run it can be found here http://www.atruk.co.uk/
Hamish
thats no many to get an accurate rating really. no wonder Tel and gabs left our screens. Five must be mad thinking they'd get more watching with brainteaser!