yuenContrast: Generalized Yuen's T-tests

In multicon: Multivariate Constructs

Description

Computes a t-test for multiple groups using a generalization of Yuen's (1974) method for trimmed t-tests. In the case of K > 2 groups, a t-contrast use calculated based on the given contrast weights.

Usage

yuenContrast(IV, ...)

## Default Method
## Default S3 method:
yuenContrast(IV, DV, wgt = c(1, -1), tr = .2, 
	alpha = .05, EQVAR = FALSE, alternative = "unequal", ...)

## Method for class 'formula'
## S3 method for class 'formula'
yuenContrast(formula, data = NULL, wgt = c(1, -1), tr = .2, 
	alpha = .05, EQVAR = FALSE, alternative = "unequal", ...)

Arguments

IV	A factor of the same length as DV containing the independent variable codes.
DV	A numeric vector of the same length as IV containing the measured values.
formula	A formula of the form lhs ~ rhs where lhs is a numeric vector containing the data values and rhs is a variable containing the corresponding groups.
data	An optional data frame containing the variables in the formula.
wgt	A numeric vector containing the contrast weights corresponding to each successive level of the IV. Defaults to c(1, -1), implying that the first group is expected to have a higher mean than the second.
tr	A numeric element between .00 and < .50 specifying the amount of trimming to be used.
alpha	A numeric element > .00 and < 1.00 specifying the Type I error rate.
EQVAR	A logical indicating whether equal variances amongst the groups should be assumed. Defaults to FALSE (Yuen's Method).
alternative	A character vector specifying the alternative hypothesis. Must be one of "unequal", "greater", or "less".
...	Further arguments to be passed to or from methods.

Details

This function computes a t-value based on Yuen's (1974) method for calculating T for trimmed means if tr is greater than 0 and Welch's method for tr=0 and EQVAR=FALSE. The wgt option allows one to specify contrast weights to test hypotheses with more than 2 levels of an IV. By default it tests the hypothesis that two means are unequal. If a directional hypothesis is known ahead of time, use "greater" to predict that higher contrast weights have higher means and "less" to predict the opposite. A robust measure of the mean differences or the contrast is obtained by using some level of trimming. By setting the EQVAR option to TRUE degrees of freedom are consistent with Student's method. If EQVAR is FALSE (default) then degrees of freedom are calculated using the Welch-Sattertwaite approximation. The entire family of possible T-test equations can be found here:

http://rynesherman.com/T-Family.doc

Value

A list containing...

Ms	A data.frame with the sample size, mean, and weight for each group.
test	A data.frame with the test statistic (stat), the degrees of freedom (df), the critical value for the test statistic (crit), and the p-value

Author(s)

Ryne A. Sherman

References

Yuen, K. K. (1974). The two-sample trimmed t for unequal population variances. Biometrika, 61, 165-170. Student (1908). The probable error of a mean. Bimetrika, 6, 1-25.

See Also

winvar t.test tContrast

Examples

dv <- c(rnorm(30, mean=1, sd=2), rnorm(20))
iv <- c(rep(1,30),rep(2,20))

  # Student's t-test (assuming equal variances)
t.test(dv ~ iv, var.equal=TRUE)
  # Welch's t-test (not assuming equal variance)
t.test(dv ~ iv, var.equal=FALSE)
  # Yuen's t-test with 20% trimming assuming equal variances
yuenContrast(iv, dv, EQVAR=TRUE)
  # Yuen's t-test with 20% trimming not assuming equal variances
yuenContrast(iv, dv, EQVAR=FALSE)
  # Same as Student's t-test using yuenContrast
yuenContrast(iv, dv, EQVAR=TRUE, tr=0)
  # Same as Welch's t-test using yuenContrast
yuenContrast(iv, dv, EQVAR=FALSE, tr=0)

  # Contrast with 3 Groups
dv <- c(rnorm(30), rnorm(20, mean=-.5), rnorm(10, mean=-1))
iv <- c(rep("c",30), rep("b", 20), rep("a", 10))
    # With no trimming (t-contrast with Welch-Sattertwaite DFs)
yuenContrast(iv, dv, wgt=c(1, 0, -1), tr=0)
    # With 20% trimming
yuenContrast(iv, dv, wgt=c(1, 0, -1), tr=.2)
    # With the formula method
yuenContrast(dv ~ iv, wgt = c(1, 0, -1))

multicon documentation built on May 2, 2019, 3:18 a.m.