ab2a1bb1   
   XPost: misc.phone.mobile.iphone, comp.mobile.android   
   From: ithinkiam@gmail.com   
      
   On 10/12/2015 09:51, Paul M. Cook wrote:   
   > On Thu, 10 Dec 2015 09:30:24 +0000, chris wrote:   
   >   
   >> None is needed. You are making an argument based on a hypothesis and   
   >> data. All one needs to do to refute it is to pick holes in it. The data   
   >> cannot prove your hypothesis therefore your argument is flawed. End of   
   >> story.   
   >>   
   >> One *could* have an alternative hypothesis which contradicts yours, but   
   >> no one here is holding this position other than you. And your   
   >> alternative argument is equally flawed. So you're arguing with yourself   
   >> and both arguments are unprovable with the accident data. Now *that* is   
   >> extraordinary.   
   >   
   > The fundamental problem, as I see it, is that our "intuition" (yes, yours,   
   > mine, and everyone else's intuition) clearly would "think" that the user   
   > of a cellphone while driving would be distracted to the point of having   
   > more accidents than a user who wasn't distracted by the use of the cellphone.   
      
   Intuition has nothing to do with it. The data *cannot* prove either of   
   your theories.   
      
   > However, I can't *find* those accidents.   
   > Neither can you.   
      
   I'm not looking for them. I'm criticising your methodology.   
      
   > Neither can anyone else!   
   >   
   >    *Where are the accidents?*   
   >   
   > More specifically, why don't the accidents show up in reliable data   
   > (that was not designed specifically to grab headlines)?   
      
   The data are inappropriate. Just having big numbers often isn't enough.   
      
   > Why do the accidents *only* show up in the so-called "studies" that   
   > attempt to *prove* [sic] that cellphone use is inherently dangerous?   
      
   It's called a carefully designed experiment. They have designed an   
   experiment to test a specific hypothesis. The data they generate are   
   explicitly there to answer a single question. Why they don't match up   
   with the global statistics is a valid question, but the two datasets are   
   not intrinsically incompatible.   
      
   What I'm saying is that the lack of increase in accident rate /and/ the   
   increased danger in driving due to cellphone use can *both* be correct.   
      
   > I repeat the fundamental question:   
   >    * Where are the accidents predicted by the model? *   
      
   "All models are wrong, some may be useful" ;)   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)