Routine air quality monitoring produces filter samples that, when analyzed,
yield the total amount of the aerosol present in the volume of air drawn b
y the pump in the monitoring device during the given sampling period. From
this we obtain an average concentration of the aerosol for the given durati
on. The samples are therefore really aggregate samples. A natural question
then is "what is the effect of the duration of aggregation on the accuracy
and precision of the estimate of the quantity of interest?" The answer depe
nds on a number of factors, such as the quantity that is being estimated: a
mean, or an extreme value, or some other quantity; the nature of the measu
rement error-additive versus multiplicative; the costs of laboratory analys
es, and so on. In this paper, we investigate these issues when the interest
is in estimating the mean concentration of a specified aerosol species ove
r a fixed time period. In particular, we propose a method for determining a
sampling duration that will yield the "best estimate" of the mean concentr
ation for a given cost whenever appropriate statistical assumptions hold.