[R-sig-eco] Obtaining P-values from rankings

[Alexandre F. Souza]

Dear friends,

I have fun a series of null models of tree species traits (e.g., height) in
permanent forest plots using the code found in Kraft's web page (
http://life.umd.edu/biology/kraftlab/Code_files/trait_tests.R). Null
distributions of trait metrics were generated by creating 999 null
communities of equal richness to the sample quadrat by drawing species at
random from our entire trait database, weighted by plot-wide occurrence
(the number of quadrats in which each species is found).

For each of 245 plots, the output consists in the following output:

test.richness 11 reps 999 mean_rank 972 obs_mean 0.850482 null_mean_mean
0.816427 null_mean_sd 0.019774
Where mean_rank means the ranking of the observed mean.

Now I need to obtain a P-value from the ranking information. I understand
that a one-tailed P-value would be = observed rank/(number of runs + 1),
and two-tailed P values should be twice this value, as found in the manual
of Phylocom. Having 999 runs this gives almost equal Ps as the ranks of the
means, with P = 0.972 for a one-tailed test in this case. However, a
two-tailed test would yield a P-value > 1.0 in this case.

How do I use the rank of the mean with the number of runs to obtain a
one-tailed and two-tailed value?

Sorry for the introductory level of the question, but it is for a good
cause.

Best,

Alexandre

-- 
Dr. Alexandre F. Souza
Professor Ajunto III
Universidade Federal do Rio Grande do Norte
CB, Departamento de Ecologia
Campus Universitário - Lagoa Nova
59072-970 - Natal, RN - Brasil
lattes: lattes.cnpq.br/7844758818522706
http://www.docente.ufrn.br/alexsouza

	[[alternative HTML version deleted]]