No I think the so called paradox was made up, To verify:

  http://www.nhtsa.gov/FARS
  Select tab "Run a Query Using the FARS Web-Based Encyclopaedia"
  "Query Fars DATA" which takes you to 
  http://www-fars.nhtsa.dot.gov//QueryTool/QuerySection/SelectYear.aspx
  For each year 1998,1999,2000
    Select Query Year, e.g. 2000, then Submit
    In the top section Tick 
     “Roadway Function Class” 
     and “Month”, then Submit
    Select State: Montanta, and RFC Roadway Function 
    Class “Rural-Principal Arterial-Interstate”
    Select Univariate Tabulation
    Select Variable - Crash Month and “Show Zeros” Yes

Output

  1998-4,0,2,4,5,1,5,4,0,0,3,2,30
  1999-1,1,3,2,1,0,2,9,4,2,1,4,30
  2000-2,1,3,5,1,5,3,3,3,1,4,4,35

Then lose your doubts.