Latest RAJARs

[handsomefortune]

never mind the latest rajar stats, trouble at mill .....http://www.guardian.co.uk/media/2013...xtended-strike

[Resurrection Man]

In the strict context of Rajar, yes. If you were a Rajar diarist, you would be expected to tick a (time) box on any listen lasting more than 5 minutes. This is perhaps the greatest danger of drawing too many conclusions from the overall station Rajar figures, or indeed of taking them too seriously. .......
Russ

Indeed so. Because of this relatively coarse granularity of sampling, Rajar figures could be out by a factor of 3! Reason being that you mark the box if you listen for more than 5 minutes...but that equates in Rajar-speak to 15 minutes.

So being Devil's Advocate....dipping in to Breakfast at 8.00 am for five minutes - then hurriedly changing stations - coming back to see if it got any better at 8.15 for five minutes - driven away again - back at...you get my drift.....those four 5 minute slots is then interpreted by Rajar that here we have a listener who has spent an hour with Radio 3.

[french frank]

Indeed so. Because of this relatively coarse granularity of sampling, Rajar figures could be out by a factor of 3! Reason being that you mark the box if you listen for more than 5 minutes...but that equates in Rajar-speak to 15 minutes.

So being Devil's Advocate....dipping in to Breakfast at 8.00 am for five minutes - then hurriedly changing stations - coming back to see if it got any better at 8.15 for five minutes - driven away again - back at...you get my drift.....those four 5 minute slots is then interpreted by Rajar that here we have a listener who has spent an hour with Radio 3.

Sort of. But you'd have to posit that that was a usual pattern of listening for a significant number of people for it to make a vast difference over each of the 12/13 succeeding weeks, hence to the published weekly averages.

People are asked to say when they accept the diaries which stations they're likely to be going to be listening to during the week. And the survey has a different set of listeners each week. Most people who opt for the Radio 3 sticker (shall we say) will be those who, with a certain regularity, do listen to the station.

[An_Inspector_Calls]

Indeed so. Because of this relatively coarse granularity of sampling, Rajar figures could be out by a factor of 3! Reason being that you mark the box if you listen for more than 5 minutes...but that equates in Rajar-speak to 15 minutes.

So being Devil's Advocate....dipping in to Breakfast at 8.00 am for five minutes - then hurriedly changing stations - coming back to see if it got any better at 8.15 for five minutes - driven away again - back at...you get my drift.....those four 5 minute slots is then interpreted by Rajar that here we have a listener who has spent an hour with Radio 3.

That might introduce a systematic error that affected all the results, but being systematic it's unlikely to influence the relative listening levels of the various stations nor, (for purposes elsewhere in these threads) the relative use of the various platforms such as FM, DAB, internet, etc

[french frank]

That might introduce a systematic error that affected all the results, but being systematic it's unlikely to influence the relative listening levels of the various stations nor, (for purposes elsewhere in these threads) the relative use of the various platforms such as FM, DAB, internet, etc

Agreed. If a significant number were doing that (tuning in and out for 5 minutes) it would show up on the average weekly hours total. And on individual programmes, but we're not allowed to see those figures 'for editorial reasons'.

[Russ]

It's that time again folks, and 2013's Q1 results are rather good for R3. At 2163k, reach is up nearly 5% over last quarter, and nearly 14% up on last year's Q1. Hours listened to, at 13883k, are up 4.3% over the last quarter, and are the highest they've been for several years. Average weekly hours per listener remains healthy at 6.4. And, shock horror, breakfast listening has bounced back to 704k, not brilliant perhaps, but a reversal of the recent downward slide.

It'll be trebles all round in the Controller's office tomorrow morning.

Trends? Hmmm, things tend to move slowly in radio listening land, but 5% of all radio is now listened to online. And R2 continues to dominate everything, its reach topping a monster 15m.

Russ

[french frank]

And, of course, the downside

That 14% 'increase' reflects the fact that this time last year the reach was particularly poor at only 1.902m (which might itself have been a sampling blip), and the listening hours were similarly depressed. So rather than unalloyed triumph, it's Russ's 'nearly 5%' (up on last month's 2.061m) which is nearer the mark. This isn't any sort of 'record high' however, although, I agree with Russ, the result will stave off the critics for a while.

Breakfast figures: yes, as Russ says, back up again, if not brilliant - and still posing the same questions about how many new listeners it brings in, as against existing ones lost. Listening hours - yes, they'll be 'over the moon' about those: share up roughly 0.1% on the average/high. Factually, though, there is a tendency (only) for listening hours to be up, not suprisingly, when reach is up (for example, in the last September quarter reach was 2.150m, listening hours 13,515m, both exceeded a little by this quarter's figures, but very comparable.

Classic FM has also done a little better than it has been doing - a number of station figures are quite buoyant this month. Some quarters appear to have keen radio samples (everyone up) and other samples where people are less so. This one seems quite a good one for many stations (Radio 2 reaching over 15m this quarter).

[DracoM]

The worse the winter maybe the more people listen indoors?

[aka Calum Da Jazbo]

it should be noted that there has been a noticeable improvement in the jazz programming; this will have added at least two listeners ......

[french frank]

The worse the winter maybe the more people listen indoors?

Certainly outside 'oddities' like that can affect listening.

I've just been reading the BBC news story about the latest Rajars. You can understand why people who read a version that has come directly or indirectly (via a press release) from the BBC itself get the wrong end of the stick, with only half a story. This is the only mention of Radio 3:

"Radio 3 had a reach of 2.16 million, 14% more than the equivalent period last year." Yippee!

Except, that was 13.7% up on March 2012 (the very low quarter). It's 4.2% down on March 2011 and 5.5% down on March 2004. So there is a difference between saying, factually, that it has increased by 14% since this time last year, and concluding from that that the reach is, in some general way, 'going up'. Schedule changes are a major factor in the increase/decrease in reach.

[Sir Velo]

Certainly outside 'oddities' like that can affect listening.

I've just been reading the BBC news story about the latest Rajars. You can understand why people who read a version that has come directly or indirectly (via a press release) from the BBC itself get the wrong end of the stick, with only half a story. This is the only mention of Radio 3:

"Radio 3 had a reach of 2.16 million, 14% more than the equivalent period last year." Yippee!

Except, that was 13.7% up on March 2012 (the very low quarter). It's 4.2% down on March 2011 and 5.5% down on March 2004.

The irony of it is that, as a news organisation, the BBC would be all over any corporation which sought to manipulate its figures so egregiously.

[Russ]

The worse the winter maybe the more people listen indoors?

The Winter quarters (Q4, Q1) always tend to be strong for R3. The slump comes in Q2!

It was remiss of me not to qualify the 'year-on-year' change, and FF is quite right to point to 2012 Q1's poor figures. As for Sir Velo's point on manipulation of the figures, the BBC is a past master at selecting those aspects that sound good, although station controllers tend to be more careful nowadays on how they phrase their puff pieces on blogs.

Here's a good RAJAR PR bullsheet guide.

Russ

[french frank]

Here's a good RAJAR PR bullsheet guide.

Russ

I'd never seen that - so it isn't just me trying to downplay R3 success! And lazy journalists just rewrite the press release because they know no better.

Another point. This is work in progress, so there may be a terrible flaw in my reasoning -

Last quarter the sample was 24,698 over 13 weeks. This suggests that they aim to distribute maybe 2,000 diaries per week but they don't all come back with usable data. So this quarter there was on average a sample of 1,900 per week (and it's the weekly averages which are reported). That 1,900 represents the population. If they found 78.5 people who listened to R3, that's 4.132% of 1,900, and 4.132% of the 'real' population (52.325m) is 2.161m listening to R3.

So they would have found 77-78 people in their weekly sample who listened to Radio 3 for at least 5 minutes in any 15 minutes during that week. That would give the total reach which was actually 2.163m.

Now look at the Breakfast figure - 704,000. That's 32.547% of 2.163m. So the sample turned up 32.547% of 77-78 people. Or 25.5 people out of the total 1,900 sample. Together they 'represent' 704,000 which means that each person they find represents 704k/ 25.5 or 28,000 people. So imagine what a difference it makes if one or two more Breakfast listeners turn up in your 1,900 sample: they could be all that's needed to turn a poorish result into a 'record high'. And that's not statistics - it's luck!

[Figures not exact because of rounding up]

[Russ]

Quite so, ff. There's lies, damn lies, and RAJAR stats...

[Serial_Apologist]

I'd never seen that - so it isn't just me trying to downplay R3 success! And lazy journalists just rewrite the press release because they know no better.

Another point. This is work in progress, so there may be a terrible flaw in my reasoning -

Last quarter the sample was 24,698 over 13 weeks. This suggests that they aim to distribute maybe 2,000 diaries per week but they don't all come back with usable data. So this quarter there was on average a sample of 1,900 per week (and it's the weekly averages which are reported). That 1,900 represents the population. If they found 78.5 people who listened to R3, that's 4.132% of 1,900, and 4.132% of the 'real' population (52.325m) is 2.161m listening to R3.

So they would have found 77-78 people in their weekly sample who listened to Radio 3 for at least 5 minutes in any 15 minutes during that week. That would give the total reach which was actually 2.163m.

Now look at the Breakfast figure - 704,000. That's 32.547% of 2.163m. So the sample turned up 32.547% of 77-78 people. Or 25.5 people out of the total 1,900 sample. Together they 'represent' 704,000 which means that each person they find represents 704k/ 25.5 or 28,000 people. So imagine what a difference it makes if one or two more Breakfast listeners turn up in your 1,900 sample: they could be all that's needed to turn a poorish result into a 'record high'. And that's not statistics - it's luck!

[Figures not exact because of rounding up]

Ah - I smell block vote syndrome operating at the BBC: in this case the "vote" implicated by our license fee payment.