EXECUTIVE SUMMARY:   
   - We look at the relationship between location of US nuclear plants   
     and UFO sightings of different types.   
   - We find there is a strong statistical link. States with more nuke   
     plants or more reactors seem to report more UFO sightings of certain types.   
   - But when we adjust all the data for state populations we find a   
     surprising result. States with more plants or reactors per capita   
     have FEWER sightings of certain types per capita.   
   - It seems some UFO types are "shy" of concentrations of reactors.   
   - We recall UFO's of certain types also seem to be shy to approach   
     concentrations of Air Force Bases (but seemingly not concerned about   
     Army bases).  It's unclear what about nuclear plants might explain   
     the pattern. Increased security?   
      
      
   I just dropped by the BV where they had some documents from the NRC.   
   It seems the Nuclear agency has a list of UFO sightings it's been   
   maintaining (more or less) from the 1950s. Some of them involve   
   sightings seen at nuclear plants by staff, and some are related to   
   letters they receive from the public about UFO activity in the   
   neighbourhood -- something the NRC tells writers it has no remit   
   to iinvestigate but thankyou for your letter.   
      
   And of course the Big Report released recently did mention at least in   
   passing that some nuclear weapons facilities (and maybe some aircraft   
   weapons systems) have apparently been "interfered with" at the same   
   time as one or other UFO was in the vicinity. The has elsewhere been   
   taken by some commentators as a message "get rid of your nukes; they   
   are dangerous".  But some of the more jaded of us also spot a possible   
   continuation of a pattern of low-level intimidation and the message   
   might really be "surrender now; your best weapons are useless against   
   us".  of course the first part might also be "leave us alone; we don't   
   want to talk with you".   
      
   So, naturally, we have to take a look at whether nuclear plants   
   actually seem to "attract" UFO activity or not. Or something else. :)   
      
   From the NUFORC we can break down sightings by US state.  And various   
   lists are available counting commercial nuclear plants and total   
   number of reactors for each state. The situation is complicated a bit   
   because reactors and plants as a whole can come and go as the   
   economics and other factors dictate. So a count of reactors is   
   generally only applicable to some specific year.   
      
   But we can run the numbers through some robust regressions and see   
   what we get. As usual I'll only concentrate on those regressions that   
   pass 2 statistical tests -- a T-test on the \bate and a rank test on   
   the comparative ordering of the X and Y data -- at 90% confidence. And   
   then we only will look at the top10 results by R2 -- the so called   
   "explanation power" of the statistical model.  The higher the R2 the   
   larger the percentage of state-by-state variation in UFO activity is   
   matched by similar state-by-state variation in the nuclear plant or   
   reactor numbers.   
      
   We'll also break the UFO numbers down in terms of "type" (mostly   
   determined by visible shape as determined by the folks at NUFORC) and   
   also pre 2006 and post 2006 when NUFORC started using a web report   
   form that greatly changed the character of reports they received.   
      
   OK. Got all that under your belt? Then let's look at the results and   
   prepare for a shock.   
      
      
   Nuke type       UFO type                R2              \beta +- 90% CI   
                   Years   Shape   
   plants          <2006   Triangle         0.46505011     5.19158 +- 1.34795   
   plants          <2006   Circle           0.43177366     3.73168 +- 1.04776   
   recators        >2006   Egg              0.42952165     2.16787 +- 0.604824   
   recators        >2006   Cigar            0.42177169     5.56501 +- 1.59478   
   recators        >2006   Light            0.42115611     56.481 +- 16.2064   
   plants          >2006   Triangle         0.41961490     42.8254 +- 12.1928   
   plants          >2006   Egg              0.41869157     3.78347 +- 1.07923   
   recators        <2006   Triangle         0.41813878     2.78489 +- 0.795291   
   plants          >2006   Cigar            0.41392097     10.9026 +- 3.14065   
   plants          >2006   Fireball         0.41316531     40.22 +- 11.7318   
      
   Which all seems to indicate for those states with more nuke plants   
   (or indiv reactors) there seems to be more Triangle and Circle UFO   
   activity, at least as reported prior to NUFORC before 2006.   
      
   The \beta for the "best model" (first line) says for each nuclear   
   plant in a state there is an overall pattern of about 5+-1 Triangle   
   UFO reports prior to 2006. The 2 stat tests say the association is   
   <10% likely to be some noise in the data or a chance link.   
      
   It seems open and shut, more or less. UFO's of certain types seem to   
   be "attracted" to nuclear plants. Or maybe that is people that see   
   UFO's of certain types are likely to live in states with more nuclear plants.   
      
   And this is almost universally what UFO researchers and some chunk of   
   the general public believe. "They" are hanging around nuclear sites   
   "for some reason".   
      
   But -- the surprise bit is now coming -- we would expect larger states   
   to have a larger number of reactors and probably plants. And we would   
   expect larger states to also report more UFO activity. Maybe the link   
   is due entirely to this underlying "common factor".   
      
   So let's re-do the whole exercise by this time based on plants/mn   
   cap and UFO sightings/mn cap instead of raw counts.   
      
   The top 10 largest R2 models then become:   
      
   Nuke type       UFO type                R2              \beta +- 90% CI   
                   Years   Shape   
   plantspc        >2006   Diskpc           0.20324200     -11.2117 +- 5.43318   
   plantspc        >2006   Otherpc          0.17687100     -17.9411 +- 9.57942   
   recatorspc      <2006   Conepc           0.14538338     -0.569959 +- 0.338218   
   plantspc        <2006   Flashpc          0.13766746     -0.93742 +- 0.594288   
   plantspc        >2006   Cylinderpc       0.13480402     -4.18458 +- 2.59467   
   recatorspc      <2006   Eggpc            0.13343271     -0.62297 +- 0.397457   
   recatorspc      <2006   Unknownpc        0.13294601     -1.589 +- 1.00437   
   recatorspc      >2006   Conepc           0.12749719     -1.38299 +- 0.895438   
   recatorspc      <2006   Crosspc          0.12639032     -0.433907 +- 0.282346   
   recatorspc      >2006   Changingpc       0.12563661     -4.16071 +- 2.71668   
      
   SURPRISE!   
      
   No only are the R2's reduced a lot from the above table of raw counts,   
   but all the \beta's are now negative. These are the top10 results but   
   no combination of plant/reactor and year/type showed a positive \beta   
   that was statistically significant.   
      
   It seems UFO's are actually "scared" of nuclear plants.   
      
   Taking the best model, above, we find for each nuclear plant per   
   mn cap there are 11 FEWER Disk sightings per mn cap in that state   
   in the data prior to 2006.   
      
   If you're the curious type we can look at the first model in detail:   
      
   MODEL (plants/mn cap vs pre 2006 Disk sightings/mn cap):   
      
   y = -11.2117*x + 13.2465   
   beta in -11.2117 +- 5.43318  90% CI   
   alpha in 13.2465 +- 1.31111   
   T-test: P(beta<0.000000) = 0.999425   
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)