Meelis.test: Meelis and James tests and R statistic

In rich-d-wilkinson/precision: Statistical methods for detecting non-binomial sex allocation when developmental mortality operates

Description

Calculates the R statistic for count data, which is a ratio of the empirical variance to the expected binomial variance. Can take a mix of data of different sizes (i.e., different n in a Bin(n,p) distribution). Also runs the Meelis and James tests, testing for under or over dispersion.

Usage

Meelis.test(experiment, TwoSided = FALSE)
James.test(experiment, TwoSided = FALSE)

Arguments

experiment	The dataset. Should be a matrix with 2 columns. The first column contains the counts of the clutch sizes, and the second column contains the counts of the number of males in each clutch.
TwoSided	logical; if TRUE, a two-sided test is performed. If we only want to test for under-disperion, a one-sided test could be used.

Value

vals	Intermediate values used in the calculation of the other higher-level statistics
R.av	The descriptive R ratio which is the ratio of empirical and expected variance
s2	McCullagh's $s^2$ statistic
U.av	The U test statistic as described in Wilkinson et al. 2013.
p.av	The p-value corresponding to the U statistic, calculated as part of either a one or two-sided calculation depending on the option specified.
exp.table	The experimental data organised into table format.

Author(s)

Richard Wilkinson

References

Krackow et al. 2002, in Hardy 2002, Sex Ratios: Concepts and Research Methods, CUP.

Examples

data(GlegneriSecondary)
Meelis.test(GlegneriSecondary, TwoSided=TRUE)
James.test(GlegneriSecondary, TwoSided=TRUE)

rich-d-wilkinson/precision documentation built on May 27, 2019, 7:41 a.m.