Adjust x and y variables for SITAR random effects
xyadj Adjusts x and y values for subject-specific
random effects from a SITAR model.
xyadj(object, x, y = NULL, id, abc = NULL, tomean = TRUE)
object |
a SITAR model. |
x |
a vector of x coordinates. If missing, |
y |
a vector of y coordinates (default NULL). |
id |
a factor denoting the subject levels corresponding to |
abc |
a data frame containing random effects for a, b, c and d (default
|
tomean |
a logical defining the direction of adjustment. TRUE (default) indicates that individual curves are translated and rotated to match the mean curve, while FALSE indicates the reverse, the mean curve being translated and rotated to match individual curves. |
When tomean = TRUE the x and y values are adjusted to
(x - xoffset - b<fixed> - b<random>) * exp(c<random>) + xoffset + b<fixed>
y - a<random> - d<random> * x
When tomean = FALSE they are adjusted to
(x - xoffset - b<fixed>) / exp(c<random>) + xoffset + b<fixed> + b<random>
y + a<random> + d<random> * x
In each case missing values of the fixed or random effects are set to zero.
The list of adjusted values:
x |
numeric vector. |
y |
numeric vector the same length as x, or NULL. |
Tim Cole tim.cole@ucl.ac.uk
data(heights) ## fit sitar model for height m1 <- sitar(x=age, y=height, id=id, data=heights, df=5) ## plot unadjusted data as growth curves plot(m1, opt='u') ## overplot with adjusted data as points with(heights, points(xyadj(m1), col='red', pch=19))
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.