Adjusted dip test quantile table, dip Handout II

          These quantiles are for the square root of n times the dip 
       statistic D.  Thus if the square root of n times D is larger 
       than the quantile for the given n and some q, then the p-value 
       is less or equal 1-q. Since the values of n with given quan-
       tiles are rather sparse, this table with values varying more 
       slowly with n is better suited to interpolation between values 
       of n than the table for the dip statistic itself.
   
       The quantiles in qDiptab are given to 8 decimal places. The 
       last several digits are unreliable due to sampling error. 
       Still, they were multiplied by square root of n before being 
       rounded to 4 places here.


          
          q   0.01   0.05    0.1    0.5    0.9   0.95   0.99  0.995  0.999
       n
       10   0.1929 0.2272 0.2466 0.3092 0.4126 0.4415 0.5050 0.5287 0.5711
       15   0.2116 0.2363 0.2489 0.3239 0.4264 0.4600 0.5267 0.5517 0.6021
       20   0.2121 0.2358 0.2539 0.3280 0.4348 0.4702 0.5394 0.5660 0.6197
       30   0.2168 0.2434 0.2596 0.3366 0.4463 0.4829 0.5558 0.5834 0.6417
       50   0.2223 0.2496 0.2668 0.3454 0.4590 0.4969 0.5742 0.6031 0.6653
       100  0.2283 0.2566 0.2740 0.3546 0.4715 0.5113 0.5902 0.6204 0.6845
       200  0.2334 0.2620 0.2799 0.3618 0.4808 0.5210 0.6037 0.6358 0.7031
       500  0.2379 0.2669 0.2849 0.3679 0.4892 0.5306 0.6138 0.6462 0.7159
       1000 0.2404 0.2695 0.2877 0.3714 0.4938 0.5355 0.6201 0.6530 0.7232
       2000 0.2422 0.2715 0.2896 0.3737 0.4965 0.5384 0.6246 0.6569 0.7276
       5000 0.2438 0.2732 0.2914 0.3756 0.4989 0.5410 0.6267 0.6606 0.7318

       

       In Handout III are regression formulas for q = 0.95, 0.99, and
       0.999, based on n = 500, 1000, 2000, and 5000, intended for
       n > 5000. For n < 5000 not given in the table, simple
       interpolation may well be more accurate than use of the
       regression formulas, although the formulas once programmed
       may be handier to use. One can also look at the residuals
       for each regression.