In a size-biased or weighted sample, the probability that a unit is sa
mpled depends on a property of the sampling unit. Cars sampled by an o
bserver traveling at a fixed velocity, v0, can be modeled as a size-bi
ased sample because a car with a large absolute difference between its
velocity and v0 is more likely to be sampled. A consistent estimator
of the population mean velocity is found that depends solely on counts
of the number of cars the observer passes and the number of cars that
pass the observer. This estimator has the unusual property that when
counts of passed and passing cars are equal, the sample median velocit
y and the estimated population mean velocity are both the observer's v
elocity, v0.