Sky News

To be honest I doubt the general public will have noticed any differance. My mum didn't notice a thing when News24 had their relaunch.

That was on a much smaller scale though - Sky's revamp is a much bigger affair.

I for one like it - but I've never been a fan of Sky News. Being new to digital TV, the presenters were all unknowns to me, and the graphics were just hideous while the studio never had much of an impact.

The new look has a much nicer graphics package, is introducing presenters to the viewers through it's "appointment to view" schedule and has a fantastic set. Personally I find it very reminisent of the early days of Five News with the walking around the newsroom.

Give it a few weeks and they'll stand still a bit longer - just let them play with their new toys for now.

Talking of the set - the main newswall, podiums and desks are great, but I don't reckon much to that wire globe thing or the set used for Sky Report.

You're probably right about most of the public. I haven't committed the last couple of overnights I saw to memory, but there is a theme developing (and MASSIVE CAVEAT - BARB point out that any numbers below 100,000 are very inaccurate - based on 10-20 viewers across the country. Almost everything I'm about to say is on audiences below 100,000). The rolling news bits - the bits like the "old" Sky News - do quite well, the new strands don't do so well. Sky News has picked up viewers as the week went on - and on Thursday, 0600 - 0200, for the first time since relaunch they had more than News24 (that's with the usual health warning about the breakfast audiences). News24 still won in peak time.

So... what can we draw from Week 1? Not much. It's been a quiet news week, Sky suffered at first, their old style programming did OK, the new stuff started badly but got better, News 24 had the kind of overall lead it's had for most of this year. People have said before me that it'll take months to judge the success of this relaunch and they're right - let's have a serious look in January.

One little snippet that might be interesting - there were two breaking stories this week when News24 slaughtered Sky on air - the Liverpool train derailment and the Israel suicide bomb. On the Liverpool story Sky's audience collapsed and News24 hit 166,000 (that's pretty huge in news channel terms). On the Israel suicide bombing, Sky and News24 stayed roughly level-pegging on audiences - when News24 were practically rolling on it and Sky struggled to cover it at all (I suspect the spirit was willing, but the technology was weak). People out there give a sh!t about a train in Liverpool, but apparently couldn't give a toss about more murder in Israel. As a London radio station used to say, that's the way it is......

With the ratings so relatively small to get a clear picture of how it's performing they'll have to be considered over a significant number of weeks.

Josh 205 Posted Here

An insignificant sample doesn't become significant just because it's all you've got

Actually it becomes more accurate - the margin of error declines with an increase in the number of measurements.

What increase of measurements? The rate of sampling is exactly the same for News24, sky and BBC1

However while BBC1's rating will be based on the viewing habits of 500 people,, Sky's and News24 will be based on those of about 5. If one of them goes to Auntie Violet's 80th Birthday party one Thursday night, it knocks 20,000 off the figures

yes but we'd be measuring their old look. We're talking about seeing how the new look affects the channel and it's only been on telly for a week.

The increase in the number of INSTANCES of measurements then. The margin of error is only going to be due to people tuning in on a whim. If the number holds firm for months then it's more likely that you can extrapolate from that proportion of people watching it.

It's not perfect, yes. But it gets more reliable with a longer time range.

Statistics, baby.

I see what you mean. Yes you can establish trends from the statistics but the day to day figures are almost meaningless.

Not that that stops the channels quoting them when it thinks they are winning.

Being unable to resist nit-picking... (Statistician hat on)

Whilst increasing the number of measurements using the same system may make the results more reliable , it certainly won't make them more accurate . You can show that you get similar figures all the time, but you'll be none the wiser as to whether or not that figure is the correct one.

I could use a faulty ruler to measure something - the more times I measure it, the more of my experimental error is eliminated, and so the more reliable the results become. But the results won't become any more accurate , because the ruler will still be faulty. Or similarly, a fixed gun pointing at a target - the more times you fire it, the more sure you are that it will hit a particular point, because other factors like the wind will slowly become less significant and average out. The gun will become more reliable , but it won't suddenly start hitting the middle of the target if it's been way off.

Or in this case, over time we'll get a better idea of what the twenty-odd people in the sample usually do, but this tells us absolutely nothing about the accuracy of the extrapolation from that. No matter how many times you ask 20 people, there is no significant increase in the logic of extrapolating that to represent the habits of millions. (Though in reality, of course, you're not just considering 20 people, because there's a whole lot more in your sample who aren't watching, and you're actually considering the habits of the group as a whole. But let's not even go there.)

(Statistician hat off)

As you were...