1) as before, the electoral roll is somewhat comprehensive and useful for random samples, but it excludes non citizens, people who do not update their addresses when they move house, and those not interested in voting.

2) Unemployment registers are useful in identifying the needy unemployed, as government benefits will not be delivered unless person officially registered.

3) Workplace lists: some occupational groups, e.g. plumbers, engineers, etc. have information on age, gender, place of residence. Usually restricted access but summaries may be obtained for research purposes. Workplace survey data usually made available to health authorities or authorised researchers.

4) Population-based surveys: Usually collected via telephone or face-to face interview, and it covers people over 18 years of age who are at home at the time of the survey; this also applies to women.

5) Inpatient data: as before. Major problem with this collection is that number of separations refers to episodes of care, and not necessarily number of patients; I.e. some patients nay be re-admitted and counted more than twice.

6) Volunteer members of the public: Method used often by marketing research, but has the obvious self-selection bias (healthy subjects, or particular at-risk groups depending on the topic of survey). Only useful for hypothesis generation, as data not generalisable. This also applies to women.