multicon (version 1.6)

rand.test: Randomization Test

Description

Computes a randomization test for the number of significant correlations and the average absolute r between set1 and set2.

Usage

rand.test(set1, set2, sims = 1000, crit = 0.95, graph = TRUE, seed = 2)

Arguments

set1

A data.frame containing the variable(s) to be correlated with set2. Can be a single vector, but must be converted to a data.frame.

set2

A matrix or data.frame containing the variables to be correlated with set1.

sims

A numeric indicating the number of randomizations to be conducted.

crit

A numeric between 0.0 and 1.0 indicating the critical value at which you will reject the null hypothesis of no relation between set1 and set2.

graph

A logical indicating whether graphical output should be returned.

seed

A numeric specifying the random seed to be used. If set to FALSE, no seed is used.

Value

A list containing...

AbsR

A vector containing the results for the average absolute r between set1 and set2. Includes the N (for complete cases), the observed average absolute r, the expected average absolute r under a null hypothesis, the standard error of the average absolute r, the p-value of the observed average absolute r, the 99.9 percent upper and lower bound confidence intervals for the p-value, and the critical value for the test to be statistically significant.

Sig

A vector containing the results for the number of significant correlations between set and the set2. Includes the N (for complete cases), the observed number significant, the expected number significant under a null hypothesis, the standard error of the number significant, the p-value of the observed number significant, the 99.9 percent upper and lower bound confidence intervals for the p-value,and the critical value for the test to be statisttically significant.

Details

When correlating a single variable of interest or a set of variables with another set of other variables, one practical consideration is the number of correlations one would expect to find by chance and/or the average absolute r between the two sets of variables. Following Sherman and Funder (2009), this function empirically estimates the sampling distribution for the number of statistically significant correlations and the average absolute r.

References

Sherman, R. A., & Funder, D. C. (2009). Evaluating correlations in studies of personality and behavior: Beyond the number of significant findings to be expected by chance. Journal of Research in Personality, 43, 1053-1063.

See Also

q.cor, ~~~

Examples

data(caq)
data(beh.comp)
head(caq)
head(beh.comp)
	#Note: In practice 'sims'=1000 is a better baseline
rand.test(caq,beh.comp,sims=100)