MSbetAvg: Calculate Mean-square-between

In shuffle: The Shuffle Estimator for Explainable Variance

Description

MSbetAvg calculates the mean-square-between contrast according to the design vector. Responses for each condition are averaged, and the sample variance is calculated for these averages.

Usage

MSbetAvg(dat, avgmat)

Arguments

dat	The vector of measurements on which the constrast is computed.
avgmat	The design parameters, as extracted by getAveraging().

Value

The value of the quadratic contrast computed on the data vector.

Author(s)

Yuval Benjamini

Examples

data(fMRI_responses,design_vec)

msbet = MSbetAvg(fMRI_responses[1,], getAveraging(design_vec))

# Compute same value using "aov" when design is balanced ... 
ns =tapply(design_vec,design_vec, length)
# (check that design is balanced)
stopifnot(length(unique(ns))==1)

m = length(unique(design_vec))

aov_sum = summary(aov(fMRI_responses[1,] ~ factor(design_vec)))
ss_bet = aov_sum[[1]][1,2]
# In unbalanced designs, each example should require more care...
msbet_aov = (ss_bet / ns[1] )/(m-1) 

cat(msbet, msbet_aov)

shuffle documentation built on May 2, 2019, 1:45 p.m.