sig.con: Calculate significance level for Monte Carlo simulation
In freesortphi: Cramer's Phi Analysis of Freesort Data

Description Usage Arguments Details Value Note Author(s) See Also

Calculates probability an observed value comes from a distribution estimated by Monte Carlo methods.

1	sig.con(x, val)

`x`	numeric; observed value
`val`	vector; values from a Monte Carlo simulation

The function estimates the percentile occupied by value x within the distribution of numbers of val. One of the following estimates is returned:

n.s., <.1, <.05, <.01

where 'n.s.' >= .1.

The significance value returned is 1-tailed, with the direction of the one-tailed test assumed from the relationship between val and the mean of x. For example, if val = 0.8 and mean(x) = 0.7, then it is assumed the test concerns val > mean.

The primary use of sig.con is to interpret the results of a Monte Carlo simulation conducted with phi.intra.chance or phi.inter.rnd. Examples of usage of sign.con are provided in the help file for those functions. The use of a one-tailed test is justified in these cases because the hypotheses are fundamentally directional (e.g. it makes little sense to ask whether phi-intra is below chance).

A character string indicating significance level (e.g. "< .05")

The quality of the estimate provided by this function depends on the quality of the estimated distribution provided in val. It is recommended that val contains the results of at least 1e5 Monte Carlo iterations.

Andy J. Wills (andy@willslab.co.uk)

phi.intra.chance, phi.inter.rnd

freesortphi documentation built on May 2, 2019, 5:54 p.m.