So the social workers consulted an eminent statistician, Professor Bowley. He advised them that not less than 408 men and 408 women would prove to be a fair sample. According to mathematical calculation this number would not show a greater deviation from the average than 1 in 22. [Footnote: Op. cit., p. 65.] They had, therefore, to question at least 816 people before they could pretend to talk about the average workingman. But which 816 people should they approach? "We might have gathered particulars concerning workers to whom one or another of us had a pre-inquiry access; we might have worked through philanthropic gentlemen and ladies who were in contact with certain sections of workers at a club, a mission, an infirmary, a place of worship, a settlement. But such a method of selection would produce entirely worthless results. The workers thus selected would not be in any sense representative of what is popularly called 'the average run of workers;' they would represent nothing but the little coteries to which they belonged.
"The right way of securing 'victims,' to which at immense cost of time and labour we rigidly adhered, is to get hold of your workers by some 'neutral' or 'accidental' or 'random' method