The Eternal Breakfast Debate in a New Place

[Eine Alpensinfonie]

Yearly averages Breakfast, 6.30am-9am:

2012/13 - 675,000 (quarters were 723,00, 665,000, 606,000, 704,000)
2013/14 - 590,800
2014/15 - 572,800

Albert Einstein is reputed to have said: "The definition of insanity is doing something over and over again and expecting a different result."

[french frank]

Albert Einstein is reputed to have said: "The definition of insanity is doing something over and over again and expecting a different result."

You mean I shouldn't keep quoting the statistics and expect the message to get through after the nth attempt?

[french frank]

Holy mackerel! I don't think I've come across such a snide and condescending post in a long time. What's with this "old" business? What's your definition?

And there's never an attempt to define what's meant by 'accessible' - dumbing it down? 'Keep it pleasurable'? They've made it unlistenable - at least for one segment of the audience: that can't be denied.

[Eine Alpensinfonie]

You mean I shouldn't keep quoting the statistics and expect the message to get through after the nth attempt?

[Eine Alpensinfonie]

'Keep it pleasurable'?

That assumes that it is pleasurable. If only. . .

[ahinton]

That assumes that it is pleasurable. If only. . .

Quite - but then FF did next write that "they've made it unlistenable - at least for one segment of the audience: that can't be denied" - in which case any sense of pleasurability for that segment at least (and doubtless also others) must already have gone out of the window, n'est-ce pas?...

[Eine Alpensinfonie]

(In the last quarter, Rajar reported a year on year 20% increase in Breakfast weekly reach, and over 20% increase in total weekly hours of listening. Changes of this magnitude are definitely statistically significant and show recovvery from a temporarily poor situation. CFM was 5% down in both categories.)[/I]

Wasn't someone on maternity leave at the time of the alleged recovery?

[mercia]

that's a heck of a long poem they are asking to be set for their carol competition, otherwise I might have been tempted

[french frank]

Wasn't someone on maternity leave at the time of the alleged recovery?

What is on record is that someone took over from someone else, mid quarter, and when a big leap was recorded at the end of that quarter, there was a suggestion (not, I think, by her) that this was the result of the change in presenter ("Figures for the Radio 3 breakfast show have however begun to rise since Clemency Burton-Hill took over a weekday role in December"). Unfortunately, the following quarter - which was entirely during said presenter's watch, reach hit that lowest ever, down from 673,000 to 526,000 - a drop of 22%. We are now celebrating the fact that they have risen to 629,000, down 6.5% from the initial good figure, but very much better than rock bottom.

It has to be remembered, however, that the number of Radio 3 listeners found in the Rajar sample is quite small, so the number of Breakfast listeners is even smaller (usually only about 30 people). With such a small sample, quarterly figures can fluctuate. Long term trends are statistically significant, quarterly comparisons - like Zucchini's "20%" - aren't.

[Eine Alpensinfonie]

It has to be remembered, however, that the number of Radio 3 listeners found in the Rajar sample is quite small, so the number of Breakfast listeners is even smaller (usually only about 30 people).

Then the figures are too insignificant to be considered reliable. Are they hand-picked to give the answers they want to hear?

[antongould]

that's a heck of a long poem they are asking to be set for their carol competition, otherwise I might have been tempted

You can do it mercs .......

[french frank]

Then the figures are too insignificant to be considered reliable. Are they hand-picked to give the answers they want to hear?

They are too insignificant to be reliable on a quarter-on-quarter basis. But 30 a quarter (Breakfast only) is 120 a year and 600 over five years 1,200 over 10 years. If they display a smooth trend the argument that results are 'random' doesn't stand up.

And no, Rajar selects people on post codes and as 'representative' in terms of demographic. They aren't selected for being Radio 3 listeners or BBC radio listeners. Given that Rajar is jointly owned by the BBC and RadioCentre (for all the commercial broadcasters), there are no answers that "it" 'wants to hear': where there is direct competition (eg for share of the 100% listening), what's good for the BBC is bad for Rajar and vice versa.

Sorry, carry on Carolling

[Eine Alpensinfonie]

Er. . . OK. I think I get it.

[Zucchini]

It has to be remembered, however, that the number of Radio 3 listeners found in the Rajar sample is quite small, so the number of Breakfast listeners is even smaller (usually only about 30 people). With such a small sample, quarterly figures can fluctuate. Long term trends are statistically significant, quarterly comparisons - like Zucchini's "20%" - aren't.

Your're wrong by a factor of ten - 300 or so respondents are Breakfast listeners.

Year on year quarterly comparisons are certainly statistically significant (i.e. the change is real within a + or - %age) when the number of respondents is of the magnitude 900+ and the change is of the magnitude of 20% quarter on quarter.

[french frank]

Your're wrong by a factor of ten - 300 or so respondents are Breakfast listeners.

300 per week? The figures are published quarterly, but they report average weekly listening, and according to our calculations that means on average each week the sample turns up 30 or fewer Breakfast listeners. Cumulatively a weekly sample of 25 over 12 weeks would total 300, but the published figures are taking the 12 or 13 week total and averaging them to find the weekly figures. Is that not correct?

If the quarter on quarter figures were that 'significant', how to explain the last 4 years second quarter figures: 723k, 615k, 673k, 629k, other than the vagaries of the samples?

Actually, the fluctuations are even worse - I was looking at the wrong quarter: 673k should be 526k.