TripleR (version 1.5.4)

parCor: partial correlation

Description

Performs partial correlations between x and y, controlled for z.

Usage

parCor(x,y,z)

Arguments

x

First variable

y

Second variable

z

Control variable. This variable is coerced into a factor; in the TripleR context z usually is the group id.

Value

par.cor

partial correlation

df

degrees of freedom for the t test

t.value

t value

p

p value

Details

Performs partial correlations between x and y, controlled for z. The control variable is coerced into a factor; in the TripleR context z usually is the group id. Do not use this function with a continuous control variable - results will be wrong! Degrees of freedom for the t test are reduced by g - 1 (g is the number of groups).

See Also

RR, getEffects

Examples

# NOT RUN {
data(multiGroup)
data(multiNarc)

# the function 'head' shows the first few lines of a data structure:
head(multiNarc)

# calculate SRA effects for extraversion ratings
RR.style("p")
RR1 <- RR(ex ~ perceiver.id * target.id | group.id, multiGroup, na.rm=TRUE)

# merge variables to one data set
dat <- merge(RR1$effects, multiNarc, by="id")

# We now have a combined data set with SRA effects and external self ratings:
head(dat)

# function parCor(x, y, z) computes partial correlation between x and y, 
# controlled for group membership z
d1 <- parCor(dat$ex.t, dat$narc, dat$group.id)
d1


# disattenuate for target effect reliability
parCor2 <- d1$par.cor * (1/sqrt(attr(RR1$effects$ex.t, "reliability")))
parCor2
# }