[Foam-rangers] Dixie Cup Cheating -- (Once again, with pictures!)

[philip verdieck]

There are three types of lies.   Lies,  damn lies and statistics.
Benjamin Disraeli,  British PM.
On Oct 22, 2013 10:05 AM, "Cliff Peery" <cpeery2 at earthlink.net> wrote:

> This is great!!
> For more there is a book out there, "How to Lie With Statistics" by
> Darrell Huff, Copyright 1954.
> This is very educational for those of use who are not operational research
> types.
> cliff
>
> On MondayOct 21, 2013, at 10:23 PM, Scott DeWalt <scott at dewalts.net>
> wrote:
>
>
>
>
> I've discovered some shenanigans.  And I have the proof!
>
> Look at this graph:
>
> <first.png>
>
>
> This is a graph shows, along the bottom, the top 20 clubs by total entry
> count.  The blue bar represents the visual percentage of how many of their
> beers went to the medal table for consideration.   Note that this does NOT
> indicate how many actually won medals nor can it be viewed as an indication
> of the quality of entries as a whole.
>
> Now this can be problematic.  Imagine I'm club SCOTT IS AWESOME and I sent
> in 1 beer.  If that beer made it to the medal round, I'd show as 100%.  Now
> imagine I'm in the same incredibly named club and sent in 100 beers but
> only 1 made it to the medal round.   I'd show up above as 1%.
>
> This is the problem with averages so don't get all excited thinking one
> club is better because their graph is better.  What you're interested in is
> the distribution of the numbers which falls within a deviation.  (Trust me,
> I'm a poorfessional.)
>
> Drat!  No cheating here!  Maybe we should dig more!
>
> Now look at this graph!  Ah ha!  The Smoking Gun!
>
> <second.png>
>
> This is a graph, by club, of the number of entries that won a medal
> against the total entries for the club.  You can see nefarious stuff here!
> On second thought, maybe not.
>
> Again, the distribution looks predictable, in fact, you could make a
> function that would fairly predict a likely result for the SCOTT IS AWESOME
> CLUB.  Back to the drawing board.
>
> But, here's the proof of cheaters!
>
> BEHOLD!!!
>
> <third.png>
>
> Now this graph says it all!  This is the percentage of medals awarded, by
> club, to beers that made the medal round.  You can CLEARLY see the trend is
> completely shattered by a whopping 100% rate for the Texas Carboys.  This
> means that EVERY SINGLE ONE OF THEIR BEERS THAT MADE THE MEDAL TABLE TOOK
> HOME A MEDAL!
>
> Now that's clearly impossible without cheating.  They must have a double
> agent!
>
> So, there you have it.  NTHB is correct, there is cheating at the Dixie
> Cub -- by the Texas Carboys!  Off with their foam heads!
>
> Scott "Numbers Lie, So Do I" DeWalt
>
>
>
>
>
>
> P.S.  Brilliant bloody-good show, Texas Carboys!  You didn't enter in big
> numbers but all four beers pushed to the medal table took home bling.
>
> P.P.S. And Mashtronauts, great job at getting nearly 50% of all your
> entries onto a medal table!  Outstanding!
>
> P.P.P.S.  Foam Rangers, we're keeping the Dixie Cup.  After all, in the
> end it is about numbers and the graph of the final tally would make
> everyone else cry.  Excellent job!
>
> _______________________________________________
> Foam-rangers mailing list
> Foam-rangers at foamrangers.com
> http://foamrangers.com/mailman/listinfo/foam-rangers_foamrangers.com
>
>
>
> _______________________________________________
> Foam-rangers mailing list
> Foam-rangers at foamrangers.com
> http://foamrangers.com/mailman/listinfo/foam-rangers_foamrangers.com
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://foamrangers.com/pipermail/foam-rangers_foamrangers.com/attachments/20131022/43e65f63/attachment.html>