CIV Survey Technical Information

What was the survey response rate?
 
The overall survey response rate was 40.3% and is defined as completed interviews as a proportion of completed interviews plus refusals.  The response rate by LGA varied from a high of 51.5% to a low of 30.6% and the most commonly cited reason for refusing to participate was a lack of interest.
 
Were any people excluded from the survey?
 
Surveys of this nature do exclude some people.  Households without a 'landline' telephone connection are unable to be contacted as are people who live in institutions, refuges, hospitals and similar facilities.  People who are not capable of undertaking a telephone interview due to a health condition, temporary incapacity or language difficulties are also excluded.  The survey methodology attempted to overcome language difficulties through the use of interpreting services during the administration of the CIV Survey.  Interpreting services were provided for the eight main languages other than English spoken in Victoria which include Cantonese, Mandarin, Vietnamese, Italian, Greek, Macedonian, Arabic (including Lebanese) and Turkish.
 
How can the survey results provide an adequate representation of my area?
 
A weighting procedure was used to generate the survey estimates from the survey interviews.  ‘Weighting’ is the process whereby each respondent in the sample has his or her response inflated to represent the response from all similar people in the whole population. The weight of a respondent reflects the proportion of the population that he or she represents.
 
In most household surveys, the weights are adjusted so that they sum to an independent estimate of the population.  This process is referred to as ‘benchmarking’, and it creates more accurate estimates when the achieved sample does not accurately represent the population. This could occur due to the random selection of the sample or due to differing response rates for separate population groups.  In household telephone surveys such as the CIV Survey, there tends to be a bias in the responding sample towards females and older persons.  The weights of respondents in these categories are reduced, while those in under-represented groups are increased.
 
The CIV sample data were benchmarked according to preliminary released Estimated Resident Population from the Australian Bureau of Statistics (June 2006) for six groups in each LGA – males 18-34 years; males 35-54 years; males 55+ years; females 18-34 years; females 35-54 years; females 55+ years.
 
Only one respondent was interviewed in each selected household therefore the benchmarking process also needed to account for this chance factor.  In order to achieve this, each respondent was also given an initial weight to represent his or her chance of selection from within the household.