parCor {TripleR} | R Documentation |
Performs partial correlations between x and y, controlled for z.
parCor(x,y,z)
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. |
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).
par.cor |
partial correlation |
df |
degrees of freedom for the t test |
t.value |
t value |
p |
p value |
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