Montecarlo-statistics on R x C matrices

Compares R x C table with randomized tables of same row and column totals.

Insert the R x C table here:
2 10 0
1 3 21
0 0 1

Methods, References and Acknowledgements

Method described in:
Nico Blüthgen, Florian Menzel and Nils Blüthgen: Measuring specialization in species interaction networks,BMC Ecology 2006, 6:9 (pdf)

RxC randomization algorithm based on:
Patefield, W.M. (1981) An efficient method of generating random RxC tables with given row and column totals. Applied Statistics 30: 91-97.
The StatLib of the Royal Statistical Society is acknowledged for permission.

Test statistic T=sum( f(r,c) * log f(r,c) ) or H=-sum( p(r,c) * log p(r,c) )

Minimum percentage of random matrices with T below or above observed matrix taken as signficance level of difference between given matrix and random matrices, see:
Manley, B.F.J. (1997) Randomization, bootstrap and Monte Carlo methods in biology. Chapman & Hall, London, 399 pp.


Please contact Nils Blüthgen (nils.bluethgen AT or Nico Blüthgen (bluethgen AT for further information.


26.09.10 Changed input form (number rows and cols automatically determined)

26.09.10 Fixed bugs in obtaining maximum H

12.12.05 Calculation of H, H', d and d' included

20.07.05 Calculation of Tmin, Tmax and S included

13.02.02 Changed input-buffer to allow bigger matrices

05.09.02 Output is now stddev and not variance