In Focus: Crime and Justice Statistics methodology

1 This release presents results from the Australian Bureau of Statistics (ABS) 2012 Personal Safety Survey (PSS), conducted from February to December 2012.

2 The survey collected information about the nature and extent of violence experienced by men and women since the age of 15, including their experience of violence in the 12 months prior to the survey. It also collected detailed information about men's and women's experience of current and previous partner violence, lifetime experience of stalking, physical and sexual abuse before the age of 15 and general feelings of safety.

3 The statistics presented in this release, refer to the Data Cubes which can be accessed in the Data downloads section, are indicative of the extensive range of data available from the survey and demonstrate the analytical potential of the survey results. Additional information may be made available by request, on a fee for service basis, through the ABS Information Consultancy, or on the Confidentialised Unit Record File.

4 This release provides information to assist users in interpreting and using the results of the survey relating to stalking, including descriptions of the survey design and methodology, and explanatory notes on the quality of estimates.

5 The ABS would like to thank the people who completed the survey. Their participation has contributed to valuable information that will help to inform public debate about violence and will help further development of policies and programs aimed at reducing the prevalence of violence.

6 The PSS meets the need for updated information on the nature and extent of violence experienced by men and women in Australia and other related information regarding people's safety at home and in the community.

7 The need for data on the prevalence of violence and sexual assault is discussed in The National Plan to Reduce Violence against Women and their Children 2010 - 2022, and in the following ABS Information Papers:

• Defining the data challenge for Family, Domestic and Sexual Violence (cat. no. 4529.0);
• Directory of Family and Domestic Violence Statistics (cat. no. 4533.0); and
• Bridging the data gaps for family, domestic and sexual violence (cat. no. 4529.0.00.002).
   

8 ABS acknowledges the support and input of the Department of Social Services (DSS) which, under the auspices of the National Plan to Reduce Violence Against Women and their Children, provided funding for the 2012 Personal Safety Survey. A Survey Advisory Group, comprising experts in the field of crime and violence, provided the ABS with advice on the information to be collected and on some aspects of survey methodology. Members of this group included representatives from State and Commonwealth Government departments, crime research agencies, service providers and academics in the field.

9 The scope of the survey was persons aged 18 years and over in private dwellings across Australia. Interviews were conducted with one randomly selected person aged 18 years or over who was a usual resident of the selected household.

10 Both urban and rural areas in all States and Territories were included in the survey, except for very remote areas of Australia and Discrete Aboriginal and Torres Strait Islander Communities, which were excluded. The following groups were also excluded from the survey:

• Visitors at a dwelling whose usual place of residence is Australia (as they would have their chance of selection at their usual residence)
• Overseas visitors intending to stay in Australia for less than 12 months
• Non-Australian diplomats, non-Australian diplomatic staff and non-Australian members of their household
• Members of non-Australian defence forces stationed in Australia and their dependants
• People who usually reside in non-private dwellings
• Households where all residents are aged less than 18 years.
   

11 Personal face to face interviews were conducted with one randomly selected person aged 18 years or over who was a usual resident of the selected household. Interviews were conducted from February to December 2012.

12 While the survey was conducted under the authority of the Census and Statistics Act, 1905, participation in the survey was not compulsory.

13 Information was collected by specially trained ABS interviewers. Experienced ABS interviewers attended a comprehensive two day survey training program. In addition to the standard ABS training provided to ABS interviewers regarding the survey content and field procedures, interviewers also received tailored sensitivity and awareness training, designed to increase their knowledge and understanding of what happens when a person experiences violence. The ABS utilised external consultants, specialised in this field to provide this component of the interviewer training.

14 To help ensure respondent comfort and well-being, as well as encouraging participation, the ABS used female interviewers for the PSS. It was considered that men and women would be more likely to feel comfortable revealing sensitive information about their possible experiences of violence to a woman. This was based on collective advice from experts in the field during the survey development, was in line with the successful procedures followed for the 2005 PSS and was also supported by the 2012 PSS Survey Advisory Group. To cater for instances where this might not be the case, the ABS also trained a small number of male interviewers, in case a respondent preferred that their interview be conducted by a male. No requests for a male interviewer were made.

15 Due to the sensitive nature of the information being collected, special procedures were used to ensure the safety of those participating and the reliability of the data provided. A specific requirement of the survey was that all interviews were conducted alone in a private setting, ensuring that other members of the household were not aware of the survey content or the responses given. This ensured the complete confidentiality of any information collected and the security of both the respondent and the interviewer, where the respondent may have been living in the same household as a perpetrator. If preferred by the respondent, the option of conducting the interview at an alternate location or by telephone interview was also available.

16 Once the questions regarding a person's experience of violence were reached in the interview, respondents were informed of the sensitive nature of the upcoming questions and their permission to continue with the interview was sought.

17 In addition, no proxy interviews were conducted. Interpreters or other family members were not used: this was to ensure the safety of those participating (where the respondent may have been living in the same household as a perpetrator) and the reliability of the data provided (where the respondent may not have felt comfortable revealing sensitive information through an interpreter/other family member, who may not have been aware of the respondent's past or current experiences).

18 To cater for instances where a respondent did not speak English, a small number of interviewers with foreign language skills were trained to conduct PSS interviews. These interviews were mostly conducted over the phone. Where a respondent required the assistance of another person to communicate with the interviewer and an ABS interviewer who spoke their language was not available, interviews were not able to be conducted. Therefore it is possible that the PSS may under represent those from a non-English speaking background. Similarly, where a respondent required the assistance of another person to communicate with the interviewer, interviews were not able to be conducted. It is also likely that the PSS will under represent those with a profound or severe communication disability.

19 To further assist ensuring respondent and interviewer safety, persons in selected dwellings were not advised in advance of their selection in the survey, as would normally be the case for ABS household surveys. Instead, interviewers were cold calling. This was to help ensure maximum chance of participation, should a respondent have been living in the same household as a perpetrator. At this first approach, it was known whether a male or female was to be interviewed at each selected dwelling. This allowed interviewers to tailor their approach, depending on who answered the door and screening questions were asked to firstly determine whether the selected dwelling contained a person of the required gender.

20 The use of specially trained interviewers ensured that rapport could be established with respondents and that the relevant concepts and definitions could be explained as necessary

21 Interviews took, on average, around 30 minutes to complete.

22 The 2012 PSS was designed to provide reliable estimates, for selected key estimates of interest, at the national level for men and women and at the State and Territory level for women. While the survey was not designed to provide State/Territory level data for men, some data was able to be produced and has been included where possible in the survey outputs.

23 Dwellings included in the survey in each State and Territory were selected at random using a stratified, multistage area sample design. This sample included only private dwellings from the geographic areas covered by the survey. Dwellings were assigned as either male (where an interview with a male was required) or female (where an interview with a female was required). The sample for women was allocated roughly equally in each State and Territory in order to provide sufficiently reliable State and Territory and national level estimates for women. The sample for men was allocated to States and Territories roughly in proportion to their respective population size, in order to provide sufficiently reliable national level estimates for men.

24 There were 41,350 private dwellings approached for the survey, comprising 31,650 females and 9,700 males. The design catered for a higher than normal sample loss rate for instances where the household did not contain a resident of the assigned gender. Where the household did not contain an in scope resident of the assigned gender, no interview was required from that dwelling. For further information about how this procedure was implemented refer to Data Collection.

25 After removing households where residents were out of scope of the survey, where the household did not contain a resident of the assigned gender, and where dwellings proved to be vacant, under construction or derelict, a final sample of around 30,200 eligible dwellings were identified.

26 Given the voluntary nature of the survey a final response rate of 57% was achieved for the survey with 17,050 persons completing the survey questionnaire nationally. The response comprised 13,307 fully responding females and 3,743 fully responding males, achieving gendered response rates of 57% for females and 56% for males.

27 Weighting is the process of adjusting results from a sample survey to infer results for the total in-scope population. For the PSS the in-scope population was persons aged 18 years and over living in private dwellings across Australia (refer to Scope of the survey). To do this, a 'weight' is allocated to each sample unit, for example a household or person. The weight is a value which indicates how many population units are represented by the sample unit. For the 2012 PSS, person weights were used.

28 The first step in calculating weights for each person was to assign an initial weight, which was equal to the inverse of the probability of being selected in the survey. For example, if the probability of a person being selected in the survey was 1 in 600, then the person would have an initial weight of 600 (that is, they represent 600 persons). For the 2012 PSS, only one in-scope person was selected per household. Thus the initial person weight was derived from the initial household weight multiplied by the total number of in-scope males or females in the household depending on the assigned gender for the household.

29 The person weights are calibrated to align with independent estimates of the size of the population of interest, referred to as 'benchmarks'. Weights calibrated against population benchmarks compensate for over or under-enumeration of particular categories of persons in the sample. This ensures that the survey estimates conform to the independently estimated distribution of the population, with respect to the benchmark categories, rather than to the distribution within the sample itself.

30 The survey results were benchmarked to the estimated resident Australian population aged 18 years or more who were living in private dwellings, excluding very remote areas of Australia and those living in Discrete Aboriginal and Torres Strait Islander communities. This population was estimated to be 17,201,700 as at 30 June 2012. The benchmarks, and hence estimates from the survey, do not (and are not intended to) match estimates for the total Australian resident population (which include persons living in non-private dwellings, such as hotels and boarding houses, persons living in very remote parts of Australia and those living in Discrete Aboriginal and Torres Strait Islander communities) obtained from other sources.

31 Given the relatively low response rate for the 2012 PSS, extensive analysis was done to ascertain whether further benchmark variables, in addition to age, sex and area of usual residence, should be incorporated into the weighting methodology. Analysis showed that the standard weighting approach did not adequately compensate for differential under-coverage in the 2012 PSS sample for the variables of social marital status, country of birth and labour force status, when compared to other ABS surveys and the 2006 Census of Population and Housing. Additional benchmarks, for social marital status, country of birth and labour force status, were incorporated into the weighting methodology.

32 As a result, initial person weights were simultaneously calibrated to the following population benchmarks:

Number of persons by:

• State or territory by capital city/balance of state by age groups by sex
• State or territory by Social marital status (Married in registered or de facto marriage and Not married)
• State or territory by broad Country of birth (Australia, Main English Speaking categories and Other)
• State or territory by Labour force status (Full Time Employed, Part Time Employed, Unemployed, or Not In the Labour Force).
   

33 The additional benchmarks were obtained from other ABS survey data. These benchmarks are considered 'pseudo-benchmarks' as they are obtained from a sample survey and as such, have a non-negligible level of sample error associated with them. The monthly Labour Force Survey (from February to December 2012) provided the pseudo-benchmarks for labour force status, social marital status and country of birth. The pseudo-benchmarks were aligned to the resident population aged 18 years or more, who were living in private dwellings in each state and territory, excluding very remote areas of Australia and those living in Discrete Aboriginal and Torres Strait Islander communities, as at 30 June 2012. They were also made to represent the same population as the demographic benchmarks with respect to state/territory, part of state, age group and sex. The sample error associated with these pseudo-benchmarks was incorporated into the standard error estimation.

34 Due to the lower than expected response rate, the ABS undertook extensive non-response analyses as part of the validation and estimation process. The analysis included reviewing interviewer observations collected for all dwellings (both responding and non-responding) to determine whether these could be used to make an adjustment to initial selection weights as a means for correcting for non-response bias.

35 Investigations showed that there was a correlation between one of the interviewer observations and whether or not a fully responding interview was obtained. An explicit non-response adjustment based on this interviewer observation was therefore made to correct for some non-response bias.

36 Estimates of counts of persons are obtained by summing person weights of persons with the characteristic of interest. All the estimates contained in this publication are based on benchmarked person weights.

37 Further information on weighting, benchmarking and estimation is available in the Personal Safety Survey, Australia: Users' Guide.

38 All sample surveys are subject to error which can be broadly categorised as either sampling error or non-sampling error.

39 Sampling error is a measure of the difference between published estimates, derived from a sample of persons, and the value that would have been produced if the total population (as defined for the scope of the survey) had been included in the survey.

40 Indications of the level of sampling error in this survey are measured by Relative Standard Errors (RSEs). In this publication, estimates of counts and proportions/percentages with an RSE of 25% to 50% are preceded by a single asterisk (e.g. *3.4) to indicate the estimate or proportion should be used with caution. Estimates of counts and proportions/percentages with RSEs greater than 50%, annotated by a double asterisk (e.g. **0.6), are considered too unreliable for general use.

41 Non-sampling error may occur in any data collection, whether it is based on a sample or a full count such as a census. Non-sampling errors occur when survey processes work less effectively than intended. Sources of non-sampling error include non-response, errors in reporting by respondents or in recording by interviewers, and occasional errors in coding and processing data. Every effort is made to reduce non-sampling error by careful design and testing of questionnaires, training of interviewers, and extensive editing and quality control procedures at all stages of data processing.

42 Non-response occurs when people cannot or will not cooperate, or cannot be contacted. Non-response can affect the reliability of results and can introduce a bias. The magnitude of any bias depends on the rate of non-response and the extent of the difference between the characteristics of those people who responded to the survey and those who did not.

43 The following methods were adopted to reduce the level and impact of non-response:

• face-to-face interviews with respondents, conducted in a private setting
• the use of interviewers, where required, who could speak languages other than English (where the language spoken was able to be established)
• follow-up of respondents if there was initially no response
• weighting to population benchmarks to reduce non-response bias
• inclusion of an explicit non-response adjustment
   

44 Through careful design and testing of the questionnaire, training of interviewers, and extensive editing and quality control procedures at all stages of data collection and processing, other non-sampling error has been minimised. However, the information recorded in the survey is essentially 'as reported' by respondents, and hence may differ from information available from other sources, or collected using a different methodology.

45 Estimates with a relative standard error (RSE) of less than 25% are considered sufficiently reliable for most purposes, and only estimates of such precision are referred to in the text of this release (unless otherwise noted). Due to the relatively small numbers of persons experiencing certain types of violence, some of the estimates provided within the data cubes are subject to very high sampling errors.

46 All differences and changes mentioned have been tested for statistical significance with a 95% level of confidence that there is a real difference in the two populations being tested. To determine whether there is a statistical difference between any other two estimates, significance testing should be undertaken.

47 The 2012 PSS collected information about a person's experience of stalking. Stalking involves various behaviours, such as loitering and following, which the respondent believed were being undertaken with the intent to harm or frighten. To be classified as stalking more than one type of behaviour had to occur, or the same type of behaviour had to occur on more than one occasion.

48 The definition of stalking is based on State and Territory legislation. It is defined by a range of behaviours which the respondent believed were undertaken with the intent to harm or frighten. Behaviours include: loitering outside a person's home, workplace or place of leisure or social activities; following or watching a person; interfering with their property; giving or leaving offensive material; telephoning or sending mail or contacting a person electronically with the intent to harm or frighten them. In order to be classified as stalking more than one type of stalking behaviour had to occur, or the same type of behaviour had to occur on more than one occasion. Where a person had experienced more than one episode of stalking by a male/by a female, details were collected only about their most recent episode of stalking by a male and most recent episode of stalking by a female.

49 Where a person had experienced stalking by a male and stalking by a female, they are counted separately for stalking by a male and by a female but are only counted once in the aggregated total.

50 The characteristics of stalking by a male and stalking by a female are not able to be added to produce a total for characteristics of "stalking". Conceptually it is invalid to add together data about the characteristics for stalking by a male and stalking by a female, as actions a person may take could differ depending on the episode. For example, if a person had contacted the police about the stalking by a male but had not contacted police about stalking by a female, it is impossible to calculate an estimate of whether or not this person has contacted the police about "stalking" - they both have and haven't. To add together data about characteristics of stalking by a male and stalking by a female would also double count all persons who have experienced stalking by both a male and a female.

51 In addition, the previous partner who most recently stalked the respondent is not necessarily the same previous partner who was reported as the most recently violent previous partner, nor the most recently emotionally abusive previous partner. Information collected about previous partner stalking cannot be directly linked to previous partner violence, or previous partner emotional abuse data.

52 Further information on the interpretation of results is available in the Personal Safety Survey, Australia, Users’ Guide.

53 The scope, content and data collection for the 2012 survey was largely the same as the 2005 survey. However differences exist in the sample design and weighting procedures. Changes between the 2005 and 2012 surveys are noted below.

• Sample size – The sample size for the 2012 survey was significantly larger to accommodate the need to provide information for males at the national level, and for females at the national and state and territory level.
• Sample design – To facilitate changes in field procedures, selected dwelling were pre-assigned a gender during the sample selection process.
• Slight changes to question concepts and/or definitions – Some changes were made to definitions to assist with respondent understanding and these have been noted in the data items list available from the Data downloads section.
• Weighting – Additional benchmarks were used in the weighting process.
   

54 These changes have minimal impact on the comparability of data between the 2005 and 2012 surveys.

55 Further information regarding weighting is available from the Personal Safety Survey, Australia, Users’ Guide.

56 The ABS collects and publishes data relating to crime and safety from different sources. Comparisons of PSS data with data from other sources cannot be readily made because of differences in data collection methods and the concepts and definitions used to measure violence.

57 Other main differences which may affect the comparability of data presented in this publication are outlined in Appendix 1.

58 Further information on crime data measurement issues is available in the following: Information paper: Information Paper: Measuring Crime Victimisation, Australia: The Impact of Different Collection Methodologies, 2002 (cat. no. 4522.0.55.001).

59 Country of birth data were classified according to the Standard Australian Classification of Countries (SACC), Second Edition (cat. no. 1269.0).

60 Languages spoken at home were classified according to the Australian Standard Classification of Language (ASCL), 2005-06 (cat. no. 1267.0).

61 Area data (Capital city, Balance of state/territory; Remoteness areas) are classified according to the Australian Standard Geographical Classification (ASGC), 2005 (cat. no. 1216.0).

62 Main field of education is categorised according to the Australian Standard Classification of Education (ASCED), 2001 (cat. no. 1272.0).

63 The Census and Statistics Act, 1905 provides the authority for the ABS to collect statistical information, and requires that statistical output shall not be published or disseminated in a manner that is likely to enable the identification of a particular person or organisation. This requirement means that the ABS must take care and make assurances that any statistical information about individual respondents cannot be derived from published data.

64 Some techniques used to guard against identification or disclosures of confidential information in statistical tables are suppression of sensitive cells, random adjustments to cells with very small values, and aggregation of data. To protect confidentiality within this publication, some cell values may have been suppressed and are not available for publication but included in totals where applicable.

65 Estimates presented in this publication have been rounded. As a result, sums of components may not add exactly to totals.

66 Proportions presented in this publication are based on unrounded figures. Calculations using rounded figures may differ from those published.

67 ABS publications draw extensively on information provided freely by individuals, businesses, governments and other organisations. Their continued cooperation is very much appreciated; without it, the wide range of statistics published by the ABS would not be available. Information received by the ABS is treated in strict confidence as required by the Census and Statistics Act, 1905.

68 The ABS would like to thank the people who completed the survey. Their participation has contributed to valuable information that will help to inform public debate about violence and will help further development of policies and programs.

69 All tables, in Excel format, can be accessed from Data downloads section. The spreadsheets present tables of estimates and percents/prevalence rates, and their corresponding relative standard errors.

70 For users who wish to undertake more detailed analysis of the survey data, a Confidentialised Unit Record File is available. Further information about microdata, including conditions of use, is available via the CURF Microdata section on the ABS website.

71 The statistics presented in this release are indicative of the extensive range of data available from the survey and demonstrate the analytical potential of the survey results. Special tabulations with additional information, subject to confidentiality and sampling variability constraints, to meet individual requirements, may be made available by request, on a fee for service basis, through the ABS Information Consultancy.

72 More detailed information regarding survey content and methodology is also available from the Personal Safety Survey, Australia, Users’ Guide. The Users' Guide will assist with evaluation and interpretation of the survey results.

73 Current publications and other products released by the ABS are listed on the ABS website. The ABS also issues a daily Release Advice on the website which details products to be released in the week ahead.