iii
required to be selected to represent each industry/establishment
unit size category was then simply calculated by dividing the
number of sample members allocated to that category by the
maximum number of employees per firm allowed for that unit size.
The establishment units were then randomly selected from the
total listing of the various registers and when each industry/
size category was full, further selection of units falling within
it was ignored. Once all the individual establishments were
selected, the management of each was asked for a list of all
its white-collar and blue-collar employees. The requisite
number of individual employees were then randomly selected from
this list stratifying by the proportion of white to blue-collar
pertaining in that particular establishment-unit.
Procedures followed for the selection both of establishment
units and of individual workers in transport and communications
and the services industries were similar although the Census and
Statistics Department did not hold complete registers for all
private concerns here, important omissions being taxi, hire-car
and public light bus companies together with land freighting in
the former and an incomplete coverage of personal and household
services in the latter. The most important omission however, was
that of the employment of merchant seamen, but data supplied by
(5) the Marine Department enabled this gap to be closed. Complete
coverage was available on the other hand for Government establish-
ment units and this was obtained chiefly from the Government
Annual Report on Expenditure. In the case of the services
industry, the public/private status of the establishment unit
was stratified for in addition to the size factor.
The Census and Statistics Department kept no register of
construction companies and this industrial sector was represen-
ted in the sample by a random selection of work sites currently in
operation from the Department's copies of planning permission
records. Each selected site was visited immediately prior to the
day fixed for the interview and a random selection was made of all
workmen present at that time.
The total number of employment establishments finally selec-
ted and actually included as the primary sampling units of the
survey was 120. (Due to the rapidly shifting nature of much of