The study Absolute vs. Relative Direct Pen Input by the Mitsubishi Electric Research Labs and the University of Toronto can give you some actual data on the subject:
We present the findings from two experiments designed to explore the
effect of absolute vs. relative direct pen interaction on both small
and large scale displays where the input and display spaces are
co-incident.
An absolute mapping – where there is a one-to-one
correspondence between the pen and cursor positions – was found to be
superior to a relative mapping – where the pen and cursor positions
can be offset with a variable mapping – for all distances on the small
screen of a Tablet PC; however, on a large wall-sized display, the
relative mapping outperformed the absolute mapping in situations
requiring cursor movement over large distances. Our findings can
inform the design of pen interfaces, in particular for large scale
displays.
And if you jump to the conclusion:
While absolute input was superior in terms of selection time for all
target distances on a Tablet PC, relative input overtook absolute
input for distant target selection on a wall-sized display.
The crossover point at which relative input performed better than
absolute input occurred at a distance of around 2200 pixels (about
200cm) indicating that relative input is preferable for displays
whiteboard sized or larger.
