In reply to a post by RobertoS:

The trouble with publishing selective analyses of statistical data, such as "2014 Ofcom data from across providers shows a mean of 91GB usage per month for FTTC users and rising, and about a quarter of users exceed the mean" is that most people associate "the mean" with the mid-point of a Normal distribution.

Of course, silly me! I should have known most people on reading "a quarter of users exceed the mean" would understand this to mean half of users exceed the mean! And in a normal distribution, what else lol

In reply to a post by Chrysalis:

The likes of BT can sell unlimited at their price point because they have millions of granny type users who use about 100kb a month.

Yep, all those grannies who flocked to sign up to unlimited FTTC right ? If AAISP lifted the lid I'd think the usage profile and mean or total usage would be broadly similar to BT FTTC.

In reply to a post by vimto_girl:

Of course, silly me! I should have known most people on reading "a quarter of users exceed the mean" would understand this to mean half of users exceed the mean! And in a normal distribution, what else lol

Heh .

Try again. The mean is nothing to do with the number of users or values each side of it. So exactly the point I'm making. It isn't a normal distribution.

A simple numerical example. 10 users at 90GB and 30 users at 10GB.

What are the axes? Presumably the usage is the "y". What is the "x"?

Or is the number of users the "y"? Then the "x" becomes the usage?

Edited by RobertoS (Wed 26-Aug-15 23:48:16)

It’s straightforward and clear: the mean usage is 91GB and around a quarter of users exceed the mean.

It obviously isn’t a normal distribution with 91GB at the midpoint if a quarter of users exceed 91GB ! So the point is: why on earth do you believe most people would think otherwise? It doesn’t make any sense, people aren’t that stupid.

(Far from hiding something by only specifying the mean, there is insight into the distribution that usefully answers the Q: ‘yes, but how many people actually used >91GB ?’)

I think you either forget or didn't reach the final paragraph of the original post that upset you. Thus losing the whole point.

I maintain that the statement you made, (or possibly quoted), is misleading to most people and a view of a graph would be more understandable.

The vast majority of the populace haven't a clue what we are arguing about. They don't know anything at all about statistics. The statement in your post is only of interest to those that do, and to them isn't much help either due to the number of ways the condition could arise.

The only possible reaction is "So? Please can we have more meaningful information". Wildly different usage patterns of statistical significance could be causing the "25% exceed the mean" result.

and you may be right, but that will be down to the packages AAISP sell.

if they sold unlimited packages at anything near the BT consumer price point, they would probably attract a unhealthy balance of heavy users which would crash it down.

They don't know anything at all about statistics.

There are three kinds of lies: lies, damned lies, and statistics.

Any news yet? I have been really impressed with your service over the past 1.5 years but I have once again gone over the 200GB limit this month and feel like the price difference is quite considerable compared to the likes of Zen Internet etc.

200GB can get used up quite easily by a family of 3-4 doing a mix of gaming and video streaming.

In reply to a post by Pipexer:

Any news yet? I have been really impressed with your service over the past 1.5 years but I have once again gone over the 200GB limit this month and feel like the price difference is quite considerable compared to the likes of Zen Internet etc.

200GB can get used up quite easily by a family of 3-4 doing a mix of gaming and video streaming.

It is - Zen is £46,99 for the 80/20 service less for 40/10.

I hope they bring out new packages too - their TS is what I would like if they had a decent package

Take a look at the comments from Mark. J on these two threads(nothing definitive, but strong hints)-

http://www.ispreview.co.uk/talk/showthread.php/33995...

http://www.ispreview.co.uk/talk/showthread.php/33909...