If we accept that the PNLF method yields accurate distances, it seems
odd, at first, that the errors from population differences and
extinction aren't larger. A potentially
dominant population effect is age, as discussed already. The key point
is that intermediate age populations all produce nearly identical
central star masses. This follows from the initial-to-final mass relation
(Weidemann (1987)).
That is, for progenitor masses between
1 and 2 M
corresponding to ages of about 1 to 10 Gyr, the central star mass will
be in the narrow range of ~ 0.58 ± 0.02
M. This narrow
range is close to that observed for white dwarfs
(McMahon (1989)).
Another important effect arises in young (< 0.5 Gyr) populations to
inhibit [OIII] luminous PN from forming.
Kaler & Jacoby (1991)
showed, and
Dopita et al. (1997)
confirmed, that for young progenitors producing central stars more
massive than 0.65
M, the surface
abundances are strongly altered such that nitrogen is greatly enhanced.
The added nitrogen competes with oxygen in cooling the nebula, to the
detriment of the [OIII] luminosity. Thus, PN deriving from young,
massive progenitors fail to populate
the high luminosity end of the PNLF and the effect of a young population
on the PNLF is lost.
Similarly, metallicity seems like it ought to play a large role. A
competition exists,
though, between the efficiency of the nebula to radiate in [OIII] and
the luminosity input
from the central star. Higher metallicity values enhance the nebula's
ability to radiate
at 
5007. The central star,
however, is predicted to have a lower mass and luminosity as
a consequence of experiencing higher mass loss prior to leaving the
AGB. The reduced heating compensates to first order for the enhanced
radiative efficiency as metallicity increases
(Dopita et al. (1992)).