Chosing a value for the prior Q is necessary at some point in the
process of evaluating the identifications. Even using equation refeq:lhr,
the value of 
is required to estimate the reliability 
.
In this section we present some
methods used by the authors who chose to include Q in the expression
for the likelyhood ratio.
Benn [3] uses an iterative method to obtain an estimate for Q, based on the fact that
such that
Having calculated the left--hand side of the above from
equations 
and , Benn iterates over non--zero
values of Q until both sides converge.
Estimating Q through iteration is one method we may investigate in the
future.
Wolstencroft et al. [5] have put forward a method for estimating
Q which is quite different.
They take 
, where 
is the
overall probability that an identification exists for the ensemble of all
sources in the sample, and 
is defined to account for the fact that a
bright candidate is more likely to be the true identification than a fainter
one:
where 
is the magnitude of the 
candidate, 
is the number
of candidates with magnitude in the range M to M+dM, and 
is the
plate limit of the optical sample.
We have preferred to include information on the magnitude of the candidate
only in the value taken for 
, the background density, as in
Rutledge et al. 2000 [9].
