Statistics in road transport: sample surveys
Page 70
If you've noticed an error in this article please click here to report it so we can fix it.
IN COMPILING many statistics it is impracticable to survey every item or person. It would be a colossal task, for example, to find out from every person above schoolleaving age their travel habits in regard to the use of private or public transport_ It would be an enormously lengthy process if a whole town was surveyed or even a single ward within a town, to Say nothing of the expense involvtd. A cheap and speedy way of collecting information is to survey only a representative sample of the activity about which information is required. Clearly it is a great attraction to interview one person in 10,000 or 1,000 instead of every person included in the survey. The difficulty is that the sample chosen will not be typical of the whole body from which it is picked.
Probably the best known instances of the use of sample surveys are public opinion polls which try to discover, from a representative sample, the party political attachments of the whole nation. Other common examples include those concerning the nation's television viewing-the figure that so many million watched a favourite programme on a particular evening is based on multiplying a representative sample. The sample employed may be completely random or it may be "stratified". In a completely random sample every member of the population surveyed has an equal chance of being interviewed. The London Traffic Survey employed a purely random sample in its investigation of travel habits. In a stratified example, on the other hand, residents in the Greater London area would be divided into particular sections, such as schoolchildren, those not gainfully employed and those in full employment. The proportion of these groups to each within the same sample survey would be identical to the actual proportions in the total population being considered.
Normally a stratified sample eliminates the risk of the sample being over-biased towards any particular group and provides a useful breakdown of the statistics into sensible groups. The average length of daily journey for Londoners as a whole (as in the London Traffic Survey) is not really of much relevance. A very important figure, however, for transport operators is the average length of journey by the commuter.
In all samples, errors are inevitable. The seriousness of the error depends on the degree of variation between the sample and the whole population under investigation. Errors also creep in because of vagueness and ambiguity in the questions asked. And, there is the added hazard of people not answering questions truthfully. I am told that misleading answers are given in transport surveys as there is always a belief that, although the interviews are being carried out by a transport agency, it is the income tax inspector in novel disguise. Or, perhaps a spy employed by the local rating assessment officer.
Thus sample surveys have many weaknesses. But at least they provide a quick method of producing definite trends and often the only reasonable way of gaining
information. The real danger lies in that figures produced tend to become accepted as positive facts, as if the whole population had been surveyed.
Having briefly indicated some of the commoner methods of presenting and compiling statistics, it would be useful next week to look at a few current vital statistics involving road passenger and haulage. And, in studying these figures, the limitations inherent in statistical presentation should always be borne in mind.