== 1. controls on raw counts: the day / mean(same weekday -7 and +7 days)
31 Oct: mean -11.77 %, median -12.42 %, range -15.81 to -2.08 %, years below zero 15 of 15
   by year: 2000:-11.9 2001:-13.0 2002:-14.3 2003:-12.3 2004:-2.1 2005:-15.6 2006:-13.6 2007:-15.0 2008:-15.5 2009:-6.6 2010:-4.9 2011:-15.8 2012:-12.4 2013:-11.8 2014:-11.7
   weekday: 2000:Tue 2001:Wed 2002:Thu 2003:Fri 2004:Sun 2005:Mon 2006:Tue 2007:Wed 2008:Fri 2009:Sat 2010:Sun 2011:Mon 2012:Wed 2013:Thu 2014:Fri
   weekend years mean -4.51 % (n 3), weekday years mean -13.59 % (n 12)
14 Feb: mean +4.51 %, median +4.12 %, range +0.12 to +9.28 %, years below zero 0 of 15
   by year: 2000:+9.3 2001:+6.2 2002:+5.3 2003:+4.1 2004:+5.1 2005:+7.9 2006:+3.9 2007:+1.3 2008:+4.7 2009:+0.1 2010:+1.6 2011:+7.0 2012:+4.0 2013:+3.8 2014:+3.1
   weekday: 2000:Mon 2001:Wed 2002:Thu 2003:Fri 2004:Sat 2005:Mon 2006:Tue 2007:Wed 2008:Thu 2009:Sat 2010:Sun 2011:Mon 2012:Tue 2013:Thu 2014:Fri
   weekend years mean +2.30 % (n 3), weekday years mean +5.07 % (n 12)

== 2. a known excess injected on window days of the real SSA series
+0.3 % injected: estimate +0.264 % (-0.040 to +0.568), verdict inconclusive
+0.5 % injected: estimate +0.462 % (+0.157 to +0.767), verdict survived

== 3. SSA against NCHS on the days both cover, 2000-2003
days 1461; SSA / NCHS mean 1.0199, sd 0.0040, min 1.0073, max 1.0339; correlation 0.9998