I was thinking about whether the high rate of contiguousness of the 28 states (25 of the 28 are contiguous) was probabilistically significant or not, so was going to run a quick probability analysis.  But, as I thought about it for a moment, the probability analysis is a little more complicated than I'd my inital thought.  

 

The complication arises because any one state in the contiguous 48 has anywhere  from 1 to 3 contiguous neighbors so that the model required to do a probability analysis reqiures a 2 dimensional matrix which mimics the contiguousness of the individual states with one another.  Given such a matrix, then each of the 48 units of the matrix can be assigned a random binary value, with which the count of contiguous units in the matrix with the same binary value can be found (not quite sure yet how to do this algorithmically or programatically, but that's just another less complicated detail.  The probability analysis would consist of several thousand runs of random binary value assignments each of which found the contiguousness of each of the binary value assignments.

 

I haven't done the analysis yet, but am eager to see (challenged) if I can figure out the matrix model for physical contiguousness... which is the 1st step.  I'm thinking of a 2 dimensional matrix where some cells of the matrix are given a physical property attribute assignment that represent a state relative to its physically neighboring states... this would also give some of the matrix cells a value of "null".  The random binary value assignments for each of the 48 physical property attribute matrix cells would be assigned to each state's physical attribute property values (therefore as many of the matrix cells that represent a single state would all recieve one of the 48 random binary value assignments) and "null" cells would be skipped in the assignments...  It's a little more complicated than that but that's the line of thinking I'm using.

 

Just thought you might have a better or another idea on the means by which the random probability of contiguousness of a given attribute condition could be set up for modeling / analysis. 

 

By the way, considering that the 28 represent 27 of the 48 contiguous states, and that this is only 3 more than the 50% probability of a random binary assigment to each of the 48 contiguous total states, it's probably well within the 1 sigma probabilty distribution of a random binary probability.  If that's true, then whether there were 27 or 26 or 24 or 22 states bringing the court case is not significant from a statistical probability point of view if there's a 50:50 probability of any one state having a GOP or Dem Governor.  

 

Anyway, an interesting thought challenge to occupy my spare brain use time (as opposed to that proportion of it I use when doing my physical activities).... then implementing it in a non-labor intensive manner to do the anaylsis.  But... one step at a time.