A survey to weigh the pax are both costly and impossible
Some of the posters are missing the important bit ..
(a) if you are going to use standard weights, the weights used MUST be representative of the population form which you are drawing your passenger loads
(b) standard weights usually are drawn from population studies (eg John K used NHMRC Australia wide population study data, as I recall). One could weigh all passengers for a period but that would be both a pain and take a moderate time before the sample set was sensible
(c) the 707 incident simply highlights a deficiency in the squadron's weight control procedures
(d) the fall back always remains .. if the commander considers that the passenger mix doesn't reflect the population studies then weigh them on the day
(e) there is NO reason why appropriately derived and applied standard weights will not give a pretty accurate figure, especially as the size of the aircraft increases .. other than operator/crew use of inappropriate protocols.