The real indication of whether the BARB ratings are reasonably accurate is whether the advertisers are happy with them, and as far as one can tell they are.
I would assume they are 'happy' with them because it it the only information they have. It does not indicate they are accurate.
1 in 2000 is only 0.05% of viewers so it can only be 0.05% accurate.
I can't see how such a low sample (BARB ratings) can be in any way accurate.
If we're talking EE/Corrie Top Gear etc then I can accept that.
If Virgin1 is aimed at men, many men live alone. How many men living alone would have a "ratings box" in their home? Perhaps the actual ratings for channels like Virgin1 are far higher.
I'm sure this topic has been done to death elsewhere
I can't see how such a low sample (BARB ratings) can be in any way accurate.
If we're talking EE/Corrie Top Gear etc then I can accept that.
If Virgin1 is aimed at men, many men live alone. How many men living alone would have a "ratings box" in their home? Perhaps the actual ratings for channels like Virgin1 are far higher.
I'm sure this topic has been done to death elsewhere
They choose their sample carefully for this exact reason.
They don't just ask people at random.
It's very accurate. There is a lot of complicated statistical maths involved.
A very large sample size would be required in order to find even a single viewer for most of the digital channels.
Extrapolating from viewing figures amongst the sample of zero or an extremely small number cannot give any meaningful results, regardless of how they are processed mathematically.
I thought the essense of accurate polling WAS randomness?
I'm not convinced.
Lies, dammed lies and statistics.
It's not polling and even then you can't have randomness like that.
If I were to stand in Windsor town center and ask people how they were going to vote and do the same in Lewisham - I think you know what the majority of each will say.
I thought the essense of accurate polling WAS randomness?
No absolutely not. You pick a sample that is representative of the general population. Otherwise you only get peoplev with time on their hands who like taking surveys.
I would assume they are 'happy' with them because it it the only information they have. It does not indicate they are accurate.
1 in 2000 is only 0.05% of viewers so it can only be 0.05% accurate.
As for the first point - they would be willing to pay if better data could reasonably be attained. Several billion pounds per year is allocated based on the BARB data.
As to the second point, I think you need to do a bit of research on statistics.
No absolutely not. You pick a sample that is representative of the general population. Otherwise you only get peoplev with time on their hands who like taking surveys.
"Pick a sample", and then use torture techniques to get answers?
Your answer was self-contradictory.
The choosing bit must be random to be accurate, some will refuse to participate. That refusal would have to be taken into account.
What I am saying is that certain groups are far less likely to participate in ANY poll. Young men for example.
Comments
I would assume they are 'happy' with them because it it the only information they have. It does not indicate they are accurate.
1 in 2000 is only 0.05% of viewers so it can only be 0.05% accurate.
If we're talking EE/Corrie Top Gear etc then I can accept that.
If Virgin1 is aimed at men, many men live alone. How many men living alone would have a "ratings box" in their home? Perhaps the actual ratings for channels like Virgin1 are far higher.
I'm sure this topic has been done to death elsewhere
They choose their sample carefully for this exact reason.
They don't just ask people at random.
It's very accurate. There is a lot of complicated statistical maths involved.
I thought the essense of accurate polling WAS randomness?
I'm not convinced.
Lies, dammed lies and statistics.
Extrapolating from viewing figures amongst the sample of zero or an extremely small number cannot give any meaningful results, regardless of how they are processed mathematically.
John
It's not polling and even then you can't have randomness like that.
If I were to stand in Windsor town center and ask people how they were going to vote and do the same in Lewisham - I think you know what the majority of each will say.
Trust me - it's all carefully worked out.
No absolutely not. You pick a sample that is representative of the general population. Otherwise you only get peoplev with time on their hands who like taking surveys.
As for the first point - they would be willing to pay if better data could reasonably be attained. Several billion pounds per year is allocated based on the BARB data.
As to the second point, I think you need to do a bit of research on statistics.
"Pick a sample", and then use torture techniques to get answers?
Your answer was self-contradictory.
The choosing bit must be random to be accurate, some will refuse to participate. That refusal would have to be taken into account.
What I am saying is that certain groups are far less likely to participate in ANY poll. Young men for example.