- From: Ronan Heffernan <ronansan@gmail.com>
- Date: Mon, 25 Mar 2013 19:02:11 -0400
- To: Dan Auerbach <dan@eff.org>
- Cc: public-tracking@w3.org
Received on Monday, 25 March 2013 23:03:00 UTC
Does it take evidence to know that if you take a carefully modeled panel, with known mixes of geography, demographics, etc., and allow some unknown number (let's say 40% at the outside) of the panel to opt-out of being counted (despite their signed contracts) via an in-band exception mechanism, that all of the tuning and extrapolation that was done based on that population has to go out the window? Thanks to the fact that we could not collect identifiable information during the in-band exception process, we couldn't even estimate the damage or tell which demographic segments have been the most heavily skewed. --ronan On Mon, Mar 25, 2013 at 1:49 PM, Dan Auerbach <dan@eff.org> wrote: > On 03/23/2013 07:45 AM, Ronan Heffernan wrote: > > using the in-band exception mechanism would skew research horribly, > > and the balanced and tuned panels constructed by our Measurement > > Science department would be replaced by biased and un-measurable crowds. > Can you provide evidence for this assertion? > > -- >
Received on Monday, 25 March 2013 23:03:00 UTC