[BC] Fractionalized Audiences with IBUZ and Internet Radio

[Broadcast List]

Let's back up and think about streaming audio for a second:

All Unicast streaming servers KNOW for a FACT how many connections  
there are at any one time; exactly when they connect and exactly when  
they disconnect.  This is better than the original Nelson People  
Meters used for TV before the addition of the IR sensors.  The only  
thing you don't know is how many people are listening to that  
computer, Roku box, Extreamer box or other appliance.  This is  
CERTAINLY MUCH BETTER than either Arbitron diaries or Personal People  
Meters.  By the way, the number is 1.33.  Now imagine that, you scale  
streaming listeners by multiplying by 1.33.  Some demographics in  
some markets are scaled by factors of over 10,000 and more because of  
low diary return rates!  Which do you think are more accurate?

Furthermore, if you want to dismiss the scaling on the Internet  
streams, you would only be wrong by a quarter.

Through a simple diary sample, one could easily determine how many  
people, on average, are listening to each stream, making the  
knowledge of the number of people listening to the stream pretty  
darned accurate.  It already has been sampled as I stated before, and  
the rule of thumb is 1.33 people per stream.

ALL of these, including the Nelson People Meters, are just taking a  
SAMPLE.  They are only giving meters or diaries to a VERY SMALL  
sample of the audience, and multiplying the sample size up to  
estimate the audience is indeed difficult, akin to determining the  
value of Pi by using an actual pie and a straight edged ruler to  
figure it out.

I know EXACTLY how many connections there are to out Internet  
streams, but I only know what Arbitron ESTIMATES are listening to our  
stations.  I dare to say I have much more confidence in the former  
than the later.

--chip

On Oct 22, 2007, at 12:36 PM, broadcast-request at radiolists.net wrote:

> Message: 16
> Date: Mon, 22 Oct 2007 11:21:00 -0400
> From: Rich Wood <richwood at pobox.com>
> Subject: Re: [BC] Fractionalized Audiences with IBUZ and Internet
> 	Radio
> To: "Broadcasters' Mailing List" <broadcast at radiolists.net>
> Message-ID: <20071022152123.5E5D58B001 at sceptre.sasl.smtp.pobox.com>
> Content-Type: text/plain; charset=us-ascii; format=flowed;
> 	x-avg-checked=avg-ok-25FD583A
>
> ------ At 09:13 AM 10/22/2007, Mike McCarthy wrote: -------
>
>> [...]
>
>> In so far as the internet, unless an audio source is being encoded
>> or tracked by software, there is no possible way to model the
>> listening of streams...anywhere.
> [...]
> Assuming there's a revolution in listening caused by wireless
> broadband I wonder which of the tens of thousands of streams will get
> free encoders. I wonder if each of the SIRIUS and XM channels will be
> encoded to, finally, get an idea of which channels actually have  
> audiences.
>
> I have to assume Arbitron has considered this scenario and has enough
> encoders for everyone who wants one. With terrestrial radio it was
> about 14,000. Now we add Internet and satellite. Manufacturing will
> have to go into overtime.
>
> Rich