Microdata and TableBuilder: Personal Safety, Australia

The Household level contains compositional and geographic information about the person’s household and household income at the time of the survey.

This level contains one record for the household of each selected person and has a weight attached (the same weight as applied on the Person level). As such, it can be used as per the Person level (see below).

Additional notes:

• Estimates should not be produced for males or all persons for Tasmania, the Northern Territory, and the Australian Capital Territory, as the weights have not been benchmarked to these states/territories. For further information on sample design and weighting, refer to the Methodology and the Data Quality and Technical Notes sections in the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).
• Household weights have not been produced for the PSS. As such, Person weights have been applied to all household data items, and all estimates produced using Household level variables represent the person population and not the household population.
• Caution should be used when making inferences regarding household characteristics and experiences of violence, as household characteristics at the time of survey may not be the same as they were at the time of a given experience. For example, a person may reside in a different state at the time of survey to the one they experienced violence in, or the composition of the household may have changed since experiencing violence. Therefore, it is recommended that household characteristic data items only be used in conjunction with data items about experiences restricted to the last 12 months.

For more information on the data available on this level, including interpretation points, refer to the Household and Demographic Characteristics chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) as well as the Data Item List available in the Data downloads section of this product.

The Person level is used to produce socio-demographic data. Each person is represented by a single record, with a single value for each data item. Each record has a weight attached, signifying the number of population units that the respondent represents, and needs to be used in order to produce estimates for the entire in-scope population. This weight is applied automatically by TableBuilder, but needs to be manually applied when using the Detailed Microdata (more details regarding applying weights can be found in the Using Weights and Producing RSEs and MoEs in Detailed Microdata collapsible of this product).

In addition to containing socio-demographic information about the respondent, the Person level also contains information about a person’s current partner who they are living with at the time of survey (if applicable). For a full list of the data items available refer to the Data Item List which can be accessed from the Data downloads section.

The Person level also contains broad level prevalence data for all experiences collected in the survey, which can be found in the aggregate data items. These aggregate data items can be used to produce prevalence rates for various experiences. For further information, refer to the Using Aggregate Data Items section of this product.

This level also contains detailed information about the:

• most recent episode of stalking by a man and by a woman
• the first incident of physical and sexual abuse experienced.

Person level data items can be used in conjunction with data items on the other levels. For example, when looking at experiences of emotional abuse by a partner, data on the Emotional abuse by a partner level (such as the type of emotional abuse behaviours experienced) can be cross tabulated with data on the Person level (such as the sex of the respondent). This is done by simply cross tabulating the data items in TableBuilder, or by merging the Person level file with the Emotional abuse by a partner level file in the Detailed Microdata (for details on merging, refer to the Identifiers and Copying Data Across Levels in Detailed Microdata section of this product).

Additional notes:

• The aggregate data items on the Person level should not be used in conjunction with data from the lower level from which they are derived (e.g. using ‘Whether experienced violence’ data on the Person level with data from the Violence prevalence level). For more information, refer to the Using Aggregate Data Items section of this product.
• Caution should be used when making inferences about the relationship between current socio-demographic characteristics at the time of survey and any experiences that occurred in the past. It would be misleading to assume that a person's current socio-demographic characteristics were present at the time of the experience. Their current characteristics such as disability or employment status may not reflect their characteristics at the time of the violence. Accordingly, comparisons should be limited to experiences that occurred more recently, e.g. in the last 12 months.
• Some socio-demographic data items will not change over a person’s lifetime (for example country of birth and first language spoken as a child) and can be analysed in conjunction with experiences which occurred at any time in the past.

For more information on the data available on this level, including interpretation points, refer to the relevant Survey Content chapters in the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) as well as the Data Item List available in the Data downloads section of this product.

This level contains detailed information about characteristics of the most recent incident (MRI) of violence experienced by men and women aged 18 years and over. Information about the most recent incident is only collected if the incident occurred in the 10 years prior to survey (and since the age of 15). Detailed information is available about the MRI for each of the following eight types of violence:

1. Sexual Assault by a male (perpetrator)
2. Sexual Assault by a female
3. Sexual Threat by a male
4. Sexual Threat by a female (available on Detailed Microdata only)
5. Physical Assault by a male
6. Physical Assault by a female
7. Physical Threat by a male
8. Physical Threat by a female

Note: Due to the low prevalence of males experiencing sexual threat by a male, and males and females experiencing sexual threat by a female, less data is available for these types of violence compared to the others. As such, the populations for each data item (as specified in the Data Item List available in the Data downloads section of this product) should be carefully referenced when producing and analysing data.

The Violence most recent incident level contains one to eight (or seven for TableBuilder) records per respondent. The number of records is determined by how many of the types of violence the respondent had experienced (e.g. if they experienced sexual assault by a male, physical threat by a female and physical assault by a male they will have three records). People who have not experienced violence, or who have reported that their most recent incident of any type of violence was not in the last 10 years, will have one record with ‘not applicable’ responses for each data item, including the Index item.

Each record has a person weight attached and must be used when producing estimates. This weight is applied automatically by TableBuilder, but needs to be manually applied when using the Detailed Microdata (more details regarding applying weights can be found in the Using the Detailed Microdata section of this product).

For more information on the data available on this level, including interpretation points, refer to the Violence – Most Recent Incident chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) as well as the Data Item List available in the Data downloads section of this product.

The Violence MRI Index item must be used when any of the other data items on this level are used. It is not necessary to use all the categories in the Index item, but at least one of the categories in the Index item must be used in order for the data to be presented appropriately and in context.

The Index item contains a category for each of the types of violence identified above, plus a ‘not applicable’ category.

The Index item categories cannot be combined to produce data for any combinations of violence. For example, combining physical assault by a male with sexual assault by a male to produce data for any assault by a male. Information about the characteristics of the most recent incident is collected separately for each of the eight types of violence, and therefore must be presented separately.

Due to the detailed data available on this level, ‘total’ type categories have not been created for the Index item on this level and these must not be produced. This is due to the data being related to the most recent incident of each type of violence experienced and not of all experiences of violence. In addition, respondents may also have experienced more than one type of violence, and therefore responses to each data item may be different. For example, a respondent may report that the first person or service that they told about their most recent incident of physical assault by a male was the ‘Police’, and the first person or service that they told about their most recent incident of physical assault by a female was a ‘Counsellor or support worker’. Without using the Index item the respondent would appear in both the ‘Police’ and ‘Counsellor or support worker‘ categories and would no longer be presenting who the first person or service who was told. Another example is presented in the next section.

In addition, not all data items apply to all types of violence. For example, information about whether the most recent incident was reported to police is not collected for incidents of threat, only assault. Users should refer to the Data Item List for information about which MRI data items apply to which types of violence (under the Population column).

Additional notes:

• This Index item should not be used on its own to produce prevalence data. This is primarily due to the restriction of data to only experiences within the last 10 years. If prevalence data for violence types are required, then the Person level or Violence Prevalence level should be used.

Details regarding the applicable Index item categories to use with each of the detailed data items is presented in the Data Item List located in the Data downloads section of this product.

Where an Index category is not applicable to a data item, this data will be presented in the ‘Not applicable’ category of the data item.

Due to the low prevalence of some violence types, not all detailed data will be of reliable quality for analysis with other items and Relative Standard Errors (RSEs) or Margins of Error (MoEs) should be considered when interpreting the data.

As outlined above, in order to provide appropriate context and to avoid misrepresentation of the data, the MRI detailed data items must be used in conjunction with one or more types of violence in the Index item, and only for those types of violence that had information collected for the particular data item(s).

For example, a respondent had reported their most recent incident of physical assault by a male to the police and not reported their most recent incident of physical assault by a female to the police. Without using the Index item with the ‘Whether police contacted about most recent incident of assault’ data item and referencing that the data represents the most recent incident of a particular type of violence, not only could the data potentially be misinterpreted as representing all experiences of violence, this respondent would also be included in both the ‘Police contacted by respondent’ and ‘Police not contacted’ categories.

Additional notes:

• As data on this level is only related to experiences of violence within the last 10 years, timeframe data for when the most recent incident of each type of violence occurred should be produced using the timeframe of violence data items on the Violence Prevalence level. The timeframe data available on the MRI level should only be used to provide context for the other data items on this level, and never on its own as a means of obtaining violence prevalence estimates for specific timeframes.
• ‘Relationship to perpetrator of most recent incident’ information is, as stated, related only to the most recent incident of each type of violence experienced. Its main purpose is to provide additional context to the other data items on this level, and should not be used alone to obtain violence prevalence estimates for specific perpetrator types. If data about the prevalence of violence by specific perpetrators is required, the Violence Prevalence level should be used.

This level contains information on experiences of different types of violence by timeframe and perpetrator type.

This level contains up to 27 records per respondent, depending on how many types of violence they have experienced since the age of 15, and the resulting contribution to ‘total’ type categories (such as ‘Violence by a person’) in the Index item. Respondents who have not experienced any violence will have one record with ‘not applicable’ values against each data item, including the Index item.

Each record has a person weight attached and should be used when producing estimates. This weight is applied automatically by TableBuilder, but needs to be manually applied when using the Detailed Microdata (more details regarding applying weights can be found in the Using Weights and Producing RSEs and MoEs in Detailed Microdata collapsible of this product).

For more information on the data available on this level, including interpretation points, refer to the Violence - Prevalence chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) as well as the Data Item List available in the Data downloads section of this product.

Additional notes:

• The Violence Prevalence Index item must always be used when using other data items available on this level.
• There is a Field Exclusion Rule applied to this level on TableBuilder which prevents using more than one item on the level (other than with the Index item). For more details regarding Field Exclusion Rules and the reason for implementing this in the PSS, refer to TableBuilder Features Used in PSS section of this product.

There are 3 types of data items on this level: Index, relationship to all perpetrators, and timeframe. These types of items and their use are outlined in more detail below.

The Violence Prevalence Index item must be used when any of the other data items on this level are used. The reason for this is outlined in the Relationship and Timeframe subsections below, as well as the example provided at the bottom of this page. It is not necessary to use all the categories in the Index item, but at least one of the categories must be used in order for the data to be presented appropriately and in context.

The Violence Prevalence Index item can also be used on its own to produce prevalence estimates for each of the different types of violence.

The following table presents each category of the Index item and source of the data for the Index and other items on the level:

Categories which are identified as being derived combine information from the source categories identified in the table, and then re-evaluate the data for all the other data items on the level to determine: the most recent timeframe by each of the broad perpetrator types and all of the perpetrator types experienced, for each of the new combined categories. This would not be achievable by simply adding violence types together and using the other data items as this would count respondents, who had experienced multiple types of violence, more than once.

Note: This process of re-deriving the data for combined categories was appropriate for this level due to the nature of the data. It was not possible to undertake this process on other levels due to additional complexities of the data (which are referred to on the appropriate Level pages of this product).

The data item ‘Relationship to all perpetrators ever experienced violence since age 15 per violence type’ can be used in conjunction with the Violence Prevalence Index data item to obtain prevalence data (since the age of 15) for specific types of violence by specific perpetrator types. However, this item is not restricted to the most recent incident of violence and therefore cannot be used in conjunction with other violence data items.

Additional notes:

• This data item is designed only to be used in conjunction with the Violence Prevalence Index data item, and should not be used in conjunction with any of the timeframe of violence data items also available on this level, which capture information about when the most recent incident of violence occurred for a limited number of perpetrator types. This is due to the categories considered part of ‘Other known persons’ not having individual timeframes available for them. The timeframe data was collected only for the most recent incident across all ‘Other known persons’. For example, if someone experienced violence by a medical practitioner 2 to less than 5 years ago and the same type of violence by a friend or acquaintance 20 years or more ago, then the only reported timeframe information collected would be 2 to less than 5 years ago. If the ‘Relationship to all perpetrators’ data item was cross tabulated against the ‘other known person’ timeframe variable, both “medical practitioner” and “friend or housemate” categories would identify their timeframes as 2 to less than 5 years ago. As mentioned previously, a Field Exclusion Rule has been applied to this level on TableBuilder, which prevents users from performing this type of cross tabulation. However, Detailed Microdata users should also refrain from combining the relationship to all perpetrators data item with any of the timeframe of violence data items. For a demonstration of this issue, refer to the Examples of Violence Prevalence Level Data for Individuals subsection of this section.

When used in combination with the Violence Prevalence Index item, the timeframe data items provide information about when the most recent incident of each type of violence occurred, broken down by the following perpetrator types:

• Stranger
• Current partner
• Previous partner (lived with)
• Boyfriend/Girlfriend or date
• Ex-boyfriend/Ex-girlfriend (never lived with)
• Other known person (note below there are two versions of this group).

Timeframe data items are also available for the following aggregated perpetrator types, produced by combining two or more of the perpetrator types listed above:

• An ‘intimate partner’ timeframe item has been produced, which includes the perpetrator categories current partner, previous partner (lived with) and boyfriend/girlfriend or date. A second version has been created that also includes ex-boyfriend/ex-girlfriend (never lived with). The two versions have been produced to accommodate different user requirements. The 2016 PSS definition of intimate partner includes the ex-boyfriend/ex-girlfriend perpetrator type. However, in previous cycles of the PSS it was not possible to extract timeframe information for this category, therefore an intimate partner version which excludes ex-boyfriend/ex-girlfriend data was created to allow time series comparisons.
• A combined ‘current partner or previous partner (lived with)’ timeframe item has been produced, which identifies the most recent incident across both of these perpetrator types.
• Two ‘other known person’ timeframe data items have been produced, which identify the most recent incident across all of the known perpetrator types excluding current and previous partner, and boyfriend/girlfriend or date. In addition, one version incorporates ex-boyfriend/ex-girlfriend data. The two types have been produced to accommodate different user requirements. The 2016 PSS definition of other known person excludes the ex-boyfriend/ex-girlfriend perpetrator type (as it is included as an intimate partner perpetrator type). However in previous cycles of the PSS, ex-boyfriend/ex-girlfriend has been included as part of other known person perpetrator types. Therefore, an other known person version which incorporates ex-boyfriend/ex-girlfriend data was created to allow time series comparisons.
• A ‘known person’ timeframe data item has been produced, which identifies the most recent incident across all perpetrator types excluding stranger.
• A timeframe data item that identifies the most recent incident across all of the perpetrator types is also available.

Additional notes:

• The ‘other known person’ timeframe data items capture the most recent incident by any of the perpetrator types classified as an ‘other known person’. Timeframe data is not available for the detailed other known perpetrator categories in the Relationship to all perpetrators data item referred to above (such as friend or housemate, co-worker etc.). Therefore, it is not appropriate to use these items with the Relationship to all perpetrators data item.
• Limited timeframe data is available for the ex-boyfriend/ex-girlfriend perpetrator category, and therefore caution should be used when analysing and interpreting this data. If this perpetrator category is required for analysis, broader timeframes and types of violence may need to be used in order to obtain data that is sufficiently reliable for general use.
• The Field Exclusion Rule applied to this level in TableBuilder means that it is not possible to utilise more than one of the timeframe data items together in a table. For more details regarding Field Exclusion Rules and the reason for implementing this in the PSS, refer to TableBuilder Features Used in PSS section of this product.

The following examples demonstrate why it is important that the Index item is always used. In addition, as previously mentioned, Index categories should only be used as they have been set out, not by combining categories together.

Example 1

If someone identified in the survey that they had experienced:

• physical assault by a male co-worker 2 to less than 3 years ago, and
• sexual assault by a female aunt in the last 12 months,

then the timeframes would be reflected in their respective violence type, timeframe by broad perpetrator, and detailed perpetrator type, as per:

Note: If the Index item was not used, the data output for the timeframe items for this example would appear as:

• Other known person (incl. ex-bf/gf) – Less than 12 months ago, 2 to less than 3 years ago
• Other known person (excl. ex-bf/gf) – Less than 12 months ago, 2 to less than 3 years ago
• Known person – Less than 12 months ago, 2 to less than 3 years ago
• Most recent incident person – Less than 12 months ago, 2 to less than 3 years ago.

Example 2

If someone identified in the survey that they had experienced:

• physical assault by a female friend 2 to less than 3 years ago,
• physical threat by a male previous partner 20 years or more ago,
• physical threat by a male medical practitioner 10 to less than 20 years ago, and
• sexual assault by a male stranger in the last 12 months,

then the timeframes would be reflected in their respective violence type, timeframe by broad perpetrator, and detailed perpetrator type, as per:

Note: If the Index item was not used, the data output for the timeframe items for this example would appear as:

• Stranger – Less than 12 months
• Previous partner – 20 years ago or more
• Other known person (incl. ex-bf/gf) – 2 to less than 3 years ago, 10 to less than 20 years ago
• Other known person (excl. ex-bf/gf) – 2 to less than 3 years ago, 10 to less than 20 years ago
• Intimate partner (incl. bf/gf) – 20 years ago or more
• Intimate partner (excl. ex-bf/gf) – 20 years ago or more
• Current and/or previous partner – 20 years ago or more
• Known person – 2 to less than 3 years ago, 10 to less than 20 years ago, 20 years ago or more
• Most recent incident – Less than 12 months, 2 to less than 3 years ago, 10 to less than 20 years ago, 20 years ago or more.

The Partner violence level contains detailed information about men's and women's experience of violence (physical or sexual violence experienced since the age of 15) by their current partner and/or their most recently violent previous partner. It presents information about characteristics of the violence and the relationship, such as the duration of the relationship, periods of separation, how often violence was experienced, whether anyone was told about the violence etc., in order to gain greater insight into their experience. For the full list of data items available on this level refer to the Data Item List available for download under the Data downloads section.

The Partner violence level contains up to two records per respondent. Respondents that have experienced only one type of partner violence (either current partner violence or previous partner violence) will have one record, whilst respondents that have experienced both types of partner violence will have two, one for each type of partner violence. Respondents that have not experienced either types of partner violence will have one record with ‘not applicable’ values against each data item, including the Index item.

Each record has a person weight attached and must be used when producing estimates. This weight is applied automatically by TableBuilder, but needs to be manually applied when using the Detailed Microdata (more details regarding applying weights can be found in the Using Weights and Producing RSEs and MoEs in Detailed Microdata section of this product).

For more information on the data available on this level, including interpretation points, refer to the Partner Violence chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) as well as the Data Item List available in the Data downloads section of this product.

The Partner Violence Index item must be used when any of the other items on this level are used. It is not necessary to use all the categories in the Index item, but at least one of the categories must be used in order for the data to be presented appropriately and in context.

The Index item contains the categories of:

• current partner violence
• previous partner violence.

As detailed information about experiences of partner violence is collected separately for a current partner and for the most recently violent previous partner, data about a person's experience of partner violence must be presented and analysed separately for these two types of partner violence, and cannot be combined to form a ‘total’ partner category on the Partner violence level. This is due to respondents potentially experiencing violence by both a current and a previous partner, and therefore responses to each data item may be different. An example is presented in the next section.

Respondents may have experienced violence by more than one previous partner. The previous partner violence data available on this level only relates to the most recently violent previous partner, and therefore is not representative of all previous partners. It is important that previous partner data obtained from this level is correctly identified as being in reference to the most recently violent previous partner.

Additional notes:

• This Index item can be used on its own to obtain data for the number of people that have experienced current partner and previous partner violence, but the preferred method for obtaining partner violence prevalence data is using the partner violence aggregate data items which are available on the Person level, or the ‘Relationship to all perpetrators’ data item on the Violence Prevalence Level.

Details regarding the applicable Index item categories to use with each of the detailed data items are presented in the Data Item List located in the Data downloads section of this product.

Due to the level of prevalence of experiences of some types of partner violence, not all detailed data will be of reliable quality for analysis with other items and RSEs or MoEs should be considered when interpreting the data.

As outlined above, the Partner violence detailed data items must be used in conjunction with the Index item. Current partner violence and previous partner violence are treated as two separate experiences, and information about the details of the violence experienced are collected separately for each. For example, a respondent had reported that they had sought advice or support about the assaults/threats by a current partner, but had not sought advice or support about the assaults/threats by their most recently violent previous partner. Without using the Index item with the ‘Whether ever sought advice or support about violence by partner’ data item and referencing that the previous partner data represents the most recently violent previous partner, not only could the data potentially be misinterpreted as representing all experiences of partner violence, but this respondent would be included in the estimates for both the ‘Sought advice or support’ and ‘Did not seek advice or support’ categories.

Additional notes:

• If prevalence data is required for experiences of violence by a current and/or previous partner (including by the sex of the partner and timeframe), this can be obtained on the Violence prevalence level or using the aggregate data items that are available on the Person level (located on the SPS Level – PV Aggregates tab of the Data Item List).
• The ‘Sex of violent partner’ data item located on this level should not be used to produce perpetrator gender data on experiences of partner violence, due to the data not representing all experiences of previous partner violence (i.e. if a respondent identified experiencing violence by both a female and male previous partner, then only the most recent of the two partners would be represented in the sex item located on this level). Its main purpose is to provide additional context to the other data items on this level. If male and female perpetrator data is required for experiences of all partner violence, then the Violence Prevalence level or the PV aggregate data items located on the Person level should be used.

This level contains information about experiences of emotional abuse by a current and/or previous partner since the age of 15 (EMAB).

This level contains one or two records per selected respondent. Respondents that have experienced emotional abuse by one of the two partner types (current partner or previous partner) will have one record, whilst respondents that have experienced emotional abuse by both partner types will have two. Respondents who have not experienced emotional abuse by either partner type will have one record with ‘not applicable’ values against each item including the Index item.

Each record has a person weight attached and must be used when producing estimates. This weight is applied automatically by TableBuilder, but needs to be manually applied when using the Detailed Microdata (more details regarding applying weights can be found in the Using Weights and Producing RSEs and MoEs in Detailed Microdata section of this product).

For a full list of behaviours used for the 2016 PSS definition of emotional abuse and further information, refer to the Partner Emotional Abuse page in the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) as well as the Data Item List available in the Data downloads section of this product.

The EMAB Index item must be used when any of the other items on this level are used. It is not necessary to use all the categories in the Index item, but at least one of the two Index item categories must be used in order for the data to be presented appropriately and in context.

The Index item contains the categories of:

• current partner emotional abuse
• previous partner emotional abuse.

It also includes a ‘not applicable’ category which is applied to people who have not experienced emotional abuse by a partner. This Index item can be used on its own, but the preferred method for obtaining emotional abuse prevalence data is using the aggregate data items which are available on the Person level (located on the SPS Level – EMAB Aggregates tab of the Data Item List).

As detailed information about experiences of partner emotional abuse is collected separately for a current partner and for the most recent emotionally abusive previous partner, data about a person's experience of partner emotional abuse must be presented and analysed separately for these two types of partner emotional abuse, and cannot be combined to form a total partner category on the emotional abuse level. This is due to respondents potentially experiencing emotional abuse by both a current and a previous partner, and therefore responses to each data item may be different. If a total category was produced, data presented could be reflective of multiple responses rather than single response outputs.

Respondents may have experienced emotional abuse by more than one previous partner. The previous partner emotional abuse data available on this level only relates to the most recent emotionally abusive previous partner, and therefore is not representative of all previous partners. It is important that previous partner data obtained from this level is correctly identified as being in reference to the most recent emotionally abusive previous partner.

Additional notes:

• This Index item can be used on its own, but the preferred method for obtaining emotional abuse prevalence data is using the aggregate data items which are available on the Person level (located on the SPS Level – EMAB Aggregates tab of the Data Item List).

Details regarding the applicable Index item categories to use with each of the detailed data items are presented in the Data Item List located in the Data downloads section of this product.

Due to the level of prevalence of experiences of some types of partner emotional abuse, not all detailed data will be of reliable quality for analysis with other items and RSEs or MoEs should be considered when interpreting the data.

As outlined above, the Partner emotional abuse detailed data items must be used in conjunction with the Index item. Current partner emotional abuse and previous partner emotional abuse are treated as two separate experiences, and information about the details of the emotional abuse experienced are collected separately for each.

For example, a respondent had reported that they had experienced anxiety and fear as a result of the emotional abuse by their current partner, but they had experienced neither as a result of the emotional abuse by their most recently emotionally abusive previous partner. Without using the Index item with the ‘Whether ever experienced anxiety or fear due to emotional abuse’ data item and referencing that the previous partner data represents the most recently emotionally abusive previous partner, not only could the data potentially be misinterpreted as representing all experiences of partner emotional abuse, but this respondent would be included in the estimates for both the ‘Experienced anxiety and fear due to emotional abuse’ and ‘Did not experience anxiety or fear due to emotional abuse’ categories.

Additional notes:

• If prevalence data is required for experiences of emotional abuse by a current and/or previous partner (since the age of 15 or in the last 12 months), emotional abuse aggregate data items are available on the Person level (located on the SPS Level – EMAB Aggregates tab of the Data Item List).
• The ‘Sex of emotionally abusive partner’ data item located on this level should not be used to produce perpetrator gender data on experiences of partner emotional abuse, due to the data not representing all experiences of previous partner emotional abuse (i.e. if a respondent identified experiencing emotional abuse by both a female and male previous partner, then only the most recent of the two partners would be represented in the sex item located on this level). Its main purpose is to provide additional context to the other data items on this level.

The 2016 PSS detailed microdata file is a hierarchical file, with data available across six levels (for more detailed information please refer to the File Structure and Content section).

Data contained on the Person level can be combined with data across all of the other levels, however the other levels of the file are mutually exclusive and in most instances it would be conceptually inaccurate to attempt to merge characteristics, sum, or cross tabulate data items across the non-Person levels.

A number of aggregate data items have been created for use on the Person level, which are designed to assist microdata users with their analysis of the co-occurrence of different types of violence, abuse, stalking or sexual harassment. These aggregate items should only be used in conjunction with other aggregate and/or socio-demographic data items on the Person level. More information is provided in the Using Aggregate Data Items section.

This section provides a brief description about when it is appropriate or not appropriate to merge data across certain levels. For details regarding comparing data within levels, refer to the relevant Using Level pages located under the File Structure and Content section.

For more information on the technical details for combining data items across levels in the Detailed Microdata product, refer to the Identifiers and Copying Data Across Levels in Detailed Microdata section of this product.

Generally, the most appropriate instances where data should be combined across levels is where users combine one of the lower levels, which are related to peoples’ experiences of various types of violence (e.g. experience of most recent incident of violence, experience of partner violence, experience of partner emotional abuse, etc.) with items on the Person (e.g. age of respondent, sex of respondent) or Household (e.g. State/Territory) levels.

For example, Detailed Microdata users could merge data from the Person level with the Partner Violence level to cross tabulate ‘Whether ever experienced anxiety or fear due to partner violence’ (and restrict the output to either current or previous partner data or present both categories using the Partner Violence Index item) by the Sex of the respondent. For TableBuilder users, these two items could simply be selected with one placed in the row and one placed in the column positions in a table along with the appropriate category (or both categories separately) from the Partner Violence Index item.

Data items from the Person and Household levels can be combined with data from any of the lower levels. This would be appropriate for analysing differences in the way different socio-demographic groups experience violence, partner violence, and partner emotional abuse. For example, Detailed Microdata users could combine sex of respondent data from the Person level with ‘whether ever experienced anxiety or fear due to partner violence’ data from the Partner level in order to examine sex differences in the rates of anxiety/fear that are experienced as a result of partner violence. For TableBuilder users, these two data items could simply be selected with one placed in the row and one placed in the column positions in a table, and using the Index item to specify the type of partner violence (current, previous, or both).

There are three violence levels available in the PSS, Violence most recent incident, Violence prevalence and Partner violence. Each level produces different types of data:

• Violence most recent incident (MRI) presents detailed characteristics of the most recent incident only for each violence type.
• Violence prevalence presents data on the different types of violence which can further be broken down by perpetrator type and the timeframe for when the most recent incident of each type of violence occurred by the selected perpetrator type.
• Partner violence presents data relating to the characteristics of the relationship and violence experienced by a current partner and (most recently violent) previous partner.

Given the different reference periods and focuses, it is not considered appropriate to merge data across these levels.

For example, if a respondent had experienced physical assault by a male previous partner and sexual assault by a male previous partner:

• While the violence prevalence level will provide data on both violence types, the MRI information may not be related to the previous partner if another perpetrator type was more recent.
• As a person may have experienced violence by more than one previous partner, information on the Violence most recent incident and Violence prevalence levels may contain information about incidents by additional previous partners to the one presented on the Partner violence level.
• If the previous partner was the most recent perpetrator for both types of violence, the MRI data related to the two types of violence also cannot be combined as:
  • they may not be reflective of the same previous partner
  • are likely to have different responses to single response questions
  • it may not be possible to determine which answers are related to the MRI across the two violence types if they both occurred within the same timeframe
  • they do not represent responses for all experiences of violence by a previous partner.

Data on the Partner violence and Partner emotional abuse levels are able to be compared when reviewing data on the current partner only, with the assumption that respondents only have one current partner. Previous partner data is not able to be compared across levels as it is not possible to confirm in all cases that the previous partner is the same partner.

Aggregate items are available on the Person level (identified on the ‘SPS level – PV and EMAB’ sheet in the Data Item List located in the Data downloads section) which provide information on the co-occurrence of current partner violence and current partner emotional abuse. These can be used in conjunction with demographic information. More information is provided in the Using Aggregate Data Items section. Similar aggregate items were not able to be produced for the co-occurrence of previous partner violence and previous partner emotional abuse, as the most recently violent previous partner may have been different to the most recent emotionally abusive partner.

A number of high level aggregate data items have been included on the Person level. These aggregates are summary data items that have been derived from multiple records held on different levels and provide users with quick-access summary information.

The purpose of these aggregate items is to aid microdata users when producing simple cross tabulations about a person’s experience of violence/abuse/stalking by socio-demographic characteristics. These aggregate data items may also be used to analyse the co-occurrence of experiences without needing to merge information from each level of the file, and they can also be used to calculate prevalence rates. An outline of the types of aggregate items produced is provided below.

• Abuse Before The Age Of 15 Aggregates
• Partner Emotional Abuse Aggregates
• Partner Violence Aggregates
• Stalking Aggregates
• Violence Since The Age Of 15 Aggregates
• Multiple Experiences Of Violence And Co-Occurrence Aggregates.

Aggregate data items are listed in the Data Item List, available from the Data downloads section of this product.

Note: Due to current system limitations of TableBuilder, it is not possible to produce new aggregate data items. Should additional items be required, the Detailed Microdata product should be used, or customised data can be requested.

In the 2016 PSS, all Abuse before the age of 15 data are located on the Person level. The data items contain information about the prevalence of physical and/or sexual abuse (aggregates) and information about the characteristics of the first incident of physical abuse and first incident of sexual abuse. The aggregate items have the primary purpose of providing quick access to prevalence information, including prevalence by specific perpetrator types.

Note: Cross tabulations between the aggregates and first incident data items should not be undertaken. In addition, the first incident data should not be combined to produce aggregates – i.e. ‘Relationship to perpetrator(s) of first incident of physical abuse before age 15’ and ‘Relationship to perpetrator(s) of first incident of sexual abuse before age 15’ should not be simply combined to produce ‘Relationship to perpetrator(s) of first incident of sexual and/or physical abuse’ as this does not take into account that respondents may have experienced both.

For more details regarding interpreting this data, refer to the Abuse before the age of 15 chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

Several aggregates have been included in the 2016 PSS file which relate to a person’s experience of emotional abuse by a current partner and/or previous partner since the age of 15. These are derived from data on the Partner emotional abuse level and includes items such as ‘Whether experienced emotional abuse by a current partner and/or previous partner since age 15.’

For more information, refer to the Using the Partner Emotional Abuse Level section of this product as well as the Partner Emotional Abuse chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

Several aggregates have been included in the 2016 PSS file which relate to a person’s experience of violence by a current partner and/or previous partner since the age of 15. These aggregates are created from data on the Violence prevalence level. They include items such as ‘Whether experienced current and/or previous partner violence since age 15.’ The items have the primary purpose of providing quick access to data about the prevalence of partner violence.

For more information, refer to the Using the Partner Violence Level section of this product and the Partner Violence chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

In the 2016 PSS, all Stalking data are located on the Person level. The data items contain information about the prevalence of stalking and detailed information about the most recent episode (MRE) of stalking by a male and the most recent episode of stalking by a female. The experience must have occurred since the age of 15, and for some data it also needed to have occurred within the last 20 years. The aggregate items have the primary purpose of providing quick access to data about the prevalence of stalking by each gender and in total.

Note: Aggregate data items cannot be cross tabulated with most recent episode data items. In addition, most recent episode of stalking by a male data items cannot be combined with most recent episode of staking by a female data item to produce a most recent episode of stalking total.

For example, ‘Whether police contacted about stalking episode by most recent male stalker’ and ‘Whether police contacted about stalking episode by most recent female stalker’ should not be simply combined to produce ‘Whether police contacted about stalking episode by most recent stalker’ as this does not take into account that respondents may have experienced both.

For more details regarding interpreting this data, refer to the Stalking chapter of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

The 2016 PSS file includes numerous violence aggregates that are derived from information on the Violence Prevalence level, and include items such as ‘Whether experienced any violence since age 15’, (that is whether they have experienced any of the 40 different types of violence since the age of 15).

For the 2016 PSS violence aggregate data items, where a person has experienced more than one type of violence, they are counted separately for each type of violence they experience and only counted once in any aggregated totals. These aggregates can readily be cross tabulated with the other items on the Person level to produce high level summary tables.

For more information, refer to the Using the Violence Most Recent Incident Level, Using the Violence Prevalence Level and Using the Partner Violence Level sections of this product and the relevant Survey Content chapters within the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

The 2016 PSS microdata product includes several suites of aggregates relating to whether a person has had multiple experiences of various types of violence. For example, ‘Whether experienced violence more than once’ (i.e. whether a person has experienced one incident of violence or more than one incident of violence) and ‘Experience of multiple incidents of sexual assault’ (i.e. whether a person has experienced multiple incidents by the same person or multiple incidents by different people).

A suite of aggregates have also been created to capture information about the co-occurrence of current partner violence with current partner emotional abuse, such as ‘Whether experienced violence and/or emotional abuse by a current partner since age 15’.

Note: Multiple experiences of violence and co-occurrence aggregates cannot be cross tabulated with data items from the lower levels and would result in double counting (where a person has experienced violence more than once). Therefore, such cross tabulation should not be undertaken. Users wanting to undertake more detailed analysis about the characteristics of violence should use the data within each level and not across these levels.

For more information, refer to Using Data Items Across Levels section of this product.

This page outlines a number of the types of data items and special categories found within the PSS data files:

• Weights
• Index items
• Continuous items
• Not known and Refusal responses
• Not applicable categories
• Multiple response data items
• Items with different populations.

The 2016 PSS is a sample survey of private dwellings in Australia. Results from the survey are weighted in order to infer results for the total in-scope population. To do this, a ‘weight’ is allocated to each survey respondent, which indicates the number of persons in the in-scope population that they represent.

There is one benchmarked weight that has been produced for the 2016 PSS, which is the Person weight (PSSWGT). The weight for each person has been applied to their record(s) across each of the file levels for the Detailed Microdata product. This means that for the Household and Person levels, which contain a single record per respondent, the sum of the weights across all records will equal the estimated residential population of Australia (aged 18 years and over).

However, for the other levels, which may contain multiple records per respondent, this weight is repeated depending upon the number of records a person has on that level. As such, if users were to simply sum the weights for these levels, they would sum to a number in excess of the total estimated residential population (aged 18 years and over). To avoid any duplication of weights and obtain the correct weighted figure, the Index items must be used so that only one record per person per Index category is utilised in the final weight calculation.

For more information regarding the weighting process, refer to the Methodology page in the Personal Safety Survey, Australia: User Guide 2016 (cat. no. 4906.0.55.003).

Note: TableBuilder automatically applies the Person weight when creating tables. For the Detailed Microdata, the weight needs to be manually included when producing data.

Index items are included on all levels below the Person level to enable differentiation between various experiences, including the type of violence experienced (Violence Most Recent Incident level and Violence Prevalence level) and the type of partner (Partner Violence level and Partner Emotional Abuse level). This is particularly important as respondents may have had more than one type of experience. The Index items are based on either violence type or partner type depending upon which level is being used. Index items are an important method of ensuring the characteristics for each experience are presented appropriately.

With the exception of the Violence Prevalence level, ‘Totals’ are generally not included as part of Index items on each level. The different categories available within Index items represent different experiences and therefore cannot be combined or aggregated. This is deliberate. Conceptually it is not appropriate to combine data about the characteristics for the different violence/partner types, as actions a person may take could differ between incidents depending on the type of violence experienced or the partner involved and may relate to only the most recent incident of the type of experience. In addition, where a person has had multiple experiences, adding data for the different violence/partner types together would result in double counting persons who had experienced more than one violence/partner type.

For more details on the use of Index items, refer to the applicable level page under the File Structure and Content section.

Note: TableBuilder may provide a default Total when adding items into a table. Users should ensure these defaults are turned off for Index items. For more information, refer to the Using Different Types of Data Items in TableBuilder section under the Using the TableBuilder section.

The 2016 PSS microdata file includes a number of continuous variables which can have a response value at any point along a continuum. In order to analyse these continuous items in more depth, microdata users can calculate means, medians and other summary statistics, as well as produce custom ranges in TableBuilder and DataLab.

Users should note that some continuous data items are allocated special codes for certain responses, which are used for responses that do not represent the data being collected (e.g. the “Not known” category). When analysing continuous items, these special codes are to be excluded. TableBuilder automatically excludes these special codes, while Detailed Microdata users will need to take care to exclude these codes in their analysis. For more information, refer to the applicable section of the Using Different types of Data Items in TableBuilder and Using Different types of Data Items in Detailed Microdata sections.

The 2016 PSS allowed respondents the option to ‘refuse’ to answer particular questions. It also allowed for a ‘Not known’ response where respondents were unable to provide an answer (e.g. if they did not know a response to questions about their partner's income, education, etc.).

In instances of ‘Not known’ or ‘Refusal’ responses, data items will have a specific response category shown as either: “Not known”, “Refusal” or “Not known/Refusal”. Refer to the values used for each of these response categories in the Data Item List.

When using data items that contain these ‘Not known’ and/or ‘Refusal’ response categories, this should be considered in analysis and noted in any outputs. One option is to list these responses as a separate output category and note them as such. Users should note that sometimes these categories will need to be included in order to obtain the full population when creating a total. For example, excluding the ‘Not known’ or ‘Refusal’ categories from sources of current partner income data and then summing the remaining known categories to obtain a total may not produce the full ‘persons with a current partner’ population. Whilst the ‘Not known’ or ‘Refusal’ categories are unknown values, they still contribute to the population total.

There are a number of instances throughout the 2016 PSS file where certain items are only relevant to a particular respondent or sub-population of respondents – i.e. they are not relevant to the “All persons” population because not all respondents were asked the question(s). In cases where an item is not relevant to a particular respondent or sub-population, they will be included within the data item’s “Not applicable” category.

Given that a ‘not applicable’ category means that the data item in question is only applicable to a particular sub-population or is not relevant to all persons in the sub-population, this must be adequately explained in any outputs using such items. For example, data items relating to current partner violence should be described as a characteristic of those ‘who experienced current partner violence’, rather than a characteristic of all persons. Data about the type of injuries received should be described as characteristics of ‘those injured in their most recent incident of physical assault by a male’, rather than characteristics of all persons.

A number of data items produced from the 2016 PSS allow a respondent to provide more than one response. These are referred to as multiple response data items. In instances where a data item allowed for more than one response to be given this has been noted in the Data Item List as a “Multi-response item” directly beneath the data item name.

When using multiple response data items, the total number of responses for outputs will be greater than the total number of respondents. As such, where data items allow for multiple responses, components are not able to be aggregated to form a total. An example of such an item from the Violence Most Recent Incident level is ‘Physical assault behaviours experienced during most recent incident’. As it is possible that a person has experienced more than one type of physical assault behaviour, they will be counted separately for each type of behaviour experienced. If data users were to sum the individual categories to obtain a total, persons who experienced more than one type of physical assault behaviour would be counted multiple times.

Note: TableBuilder can automatically produce a Total which eliminates counting people more than once, whereas this process requires a manual calculation when using the Detailed Microdata in DataLab.

For more details regarding using this type of data item in the TableBuilder or Detailed Microdata products, refer to the Using Different Types of Data Items in TableBuilder and Using Different Types of Data Items in Detailed Microdata sections of this product.

Within the Data Item List, each item is provided with a clearly defined population to which it applies, which should be considered when analysing data. For example, most of the demographic data items located on the Person level relate to all persons and therefore data can be produced for the full population. In contrast, the data available on the Violence Prevalence level is restricted to just those people who have experienced violence. Persons and sub-populations for whom the data item does not apply will be found in the ‘not applicable’ categories as described above.

Care is required when cross tabulating data with differing populations, to ensure that the data produced from both items can be restricted to a comparable population when one or more items are not applicable to the full population.

For example, if a comparison was to be made regarding ‘Whether violence occurred for the first time while living with previous partner’ (population of ‘Persons who experienced violence by a previous partner’) and ‘Whether violence by partner occurred for the first time during pregnancy’ (population of ‘Women who experienced violence by a partner while living together’) the following steps could be followed:

• Restriction of table to Previous partner using Partner Violence Index item category 2
• Restriction of table using Female category from Sex data item located on the Person level
• Restriction of table using the ‘Violence experienced while living together’ category from the ‘Whether violence experienced while living together with previous partner’ data item
• Restriction of table using the ‘Violence occurred during pregnancy’ category from ‘Whether violence by partner occurred during pregnancy’.

The table produced would look like:

It may also be possible to produce a simpler table when one population is a subset of another (which is the case with these items with the exception of still needing the Index item to produce appropriate data) by removing the not applicable categories from the data items being compared.

But it is recommended to build the table with the full data item population restrictions in place to ensure that the data correctly references the population being analysed.

An example where it would not be possible to produce a comparable population between two items would be:

• ‘Whether violence experienced while living together with previous partner’ (population of ‘Persons who experienced violence by a previous partner’)
• ‘Whether ever left property/assets behind during temporary separations from violent current partner’ (population of ‘Persons who experienced violence by a current partner while living together and who moved away from home during temporary separations’).

As one data item is related to people who experienced violence by a previous partner, and the other is related to persons who experienced violence by a current partner, it is not possible to produce a comparable population. If a table was to be constructed with these two items it would look like:

If there was an overlapping population, the highlighted cells would contain data but it would be necessary to determine what that population is in order to appropriately report on the data.

A comprehensive list of the data items available on the PSS 2016 TableBuilder file can be found in the Data downloads section of this product. This Data Item List also includes the relevant population and classification details of all available data items. Additionally, the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) provides topic definitions and interpretation points.

Specific details regarding the following technical aspects of using data items available in the PSS TableBuilder are outlined below:

• Index items
• Multiple response data items
• Hierarchical classifications
• Continuous data items.

Some of these topics have also been covered more generally in the Notes on Data Items and Special Categories section under the File Structure and Content section and should be read along with the following information.

Each of the four lower levels includes an Index item. The Index item for each level below the Person level must be used when any data items from these levels are being used. This is in order to identify the different experiences for each person contained within these levels. For example, if a user is interested in analysing whether respondents ‘Ever experienced anxiety or fear due to partner violence’ by ‘Sex of respondent’, it would not be appropriate to simply cross tabulate these two items without an Index item being specified, as seen below:

This table would not be appropriate because respondents may have experienced anxiety or fear from a current partner but not anxiety or fear from a previous partner, and therefore they would end up in both categories if these items were cross tabulated in this fashion.

Instead, the Partner Violence Index item must be specified when cross tabulating these items to produce the individual responses related to each partner type:

As can be seen above, summing ‘Violence by a male’ and ‘Violence by a female’ together in TableBuilder using the default sum will actually produce the same total as the ‘Violence by a person’ category. This is because, when TableBuilder automatically produces the ‘Total’ this total only counts each person once in the cell. Taken at face value this suggests that it would actually be appropriate to sum these Index items together.

Indeed, using an Index item from any level alone it is possible to produce the same overall total as if it was derived, such as the derived ‘Violence by a person’ category. However, the main issue with using the summed ‘Total’ occurs when the Index items are used in conjunction with other items on the same level, such as when this Index item is cross tabulated with one of the various ‘Time Frame of Violence’ items (on the Violence Prevalence level).

For example, a respondent could have experienced violence by a male ‘Less than 12 months ago’ and by a female ‘5 to less than 10 years ago’. In the table below, the respondent would be included in the highlighted cells:

As this example highlights, the ‘totals’ (i.e. ‘Violence by a person’ derived for the Index item and the ‘Total’ column created by TableBuilder) are the same for the ‘Less than 12 months ago’ category (989.5). For reasons mentioned previously, the ‘Total’ for ‘When most recent incident occurred per violence type’ is also the same for both ‘totals’ (7236.6). However, for all other time frames the ‘totals’ differ.

The ‘Less than 12 months ago’ category is the same for ‘Violence by a person’ and ‘Total’ because it is the most recent time frame an incident can be reported. The ‘Violence by a person’ data has been produced to only count each person once when cross tabulated with time frame data items. Therefore for ‘Violence by a person’ a respondent will only be presented in their overall most recent incident of violence (even if they experienced multiple incidents of violence). As such, they are in the ‘Less than 12 months ago’ category, but not in the ‘5 to less than 10 years ago’ category (this is correct). On the other hand the ‘Total’ column derived by TableBuilder is simply the sum of the time frames for ‘Violence by a male’ and ‘Violence by a female’. This results in the respondent contributing to both the ‘Less than 12 months ago’ and ‘5 to less than 10 years ago’ populations (this is incorrect).

When this example is considered in the context of Index items located on the other levels which do not have ‘total’ type categories produced, this means that there is no possibility of producing ‘Total’ data for Index items without potential for respondents to end up in more than one category of the cross tabulated data item. Therefore, as advised previously, the default ‘Total’ must always be turned off for Index items. Examples include a default ‘Total’ that combines current partner and previous partner on the Partner Violence or Partner Emotional Abuse levels or combine violence types on the Violence Most Recent Incident level.

A number of data items produced from the 2016 PSS allow a respondent to provide more than one response. These are referred to as multiple response data items. An example of one of these items being used within the TableBuilder is shown below:

When a multiple response data item is tabulated, a person is counted against each category for which they have provided a response (e.g. a person who: ‘Had something thrown at them’, was ‘Slapped’ and was ‘Choked’ will be counted against each of these three categories).

Similar to a single response data item, a person not within the applicable population (as specified in the Data Item List) will fall into the ‘Not Applicable’ category (e.g. a person who did not report experiencing physical assault is not asked about physical assault behaviours experienced and is therefore considered ‘Not Applicable’ for this data item).

When using multiple response data items the total number of responses for outputs will be greater than the total number of respondents. Where data items allow for multiple responses, the components are not able to be aggregated to form a total. If users were to aggregate the total number of behaviours by manually summing these numbers, this would double count (or more) those persons who experienced more than one type of physical assault behaviour. Fortunately, the ‘Total’ that is automatically produced by TableBuilder only counts a respondent once and not by the number of categories responded. So, unlike Index Items, TableBuilder users should use this default total when working with multiple response data items.

As mentioned in the Notes on Data Items and Special Categories subsection under File Structure and Content section, in instances where a data item allowed for more than one response to be given this has been noted in the Data Item List as a “Multi-response item” directly beneath the data item name.

A number of items, particularly those associated with Standard classifications (refer to Appendix 2: ABS Standard Classifications page located on the Explanatory Notes tab of the Personal Safety Survey, Australia: User Guide 2016 (cat no. 4906.0.55.003)), have hierarchical classifications. These items are:

• Country of birth
• Country of birth of current partner
• Country of birth of Father
• Country of birth of Mother
• First language spoken as a child (4 digit)
• Main language spoken at home (4 digit)
• First language spoken by current partner as a child (4 digit)
• Main language spoken by current partner at home (4 digit)
• Family composition.

The classifications can be drilled down to very fine detail. However, it is likely that detailed data will not be able to be produced as a result of sample size not supporting the detail and suppression of tables with a large number of cells with small estimates. It is recommended that higher levels of the classifications be used initially and where estimates appear to support it, start moving to lower levels as necessary.

For more details regarding using data items with hierarchical classifications, refer to the Working with Hierarchies page of the TableBuilder, User Guide (cat. no. 1406.0.55.005).

TableBuilder includes a number of continuous variables which can have a response value at any point along a continuum. For technical information on using continuous items, see the Summation options, ranges and quantiles section of the TableBuilder, User Guide (cat. no. 1406.0.55.005).

In the 2016 PSS TableBuilder, continuous items appear under the ‘Summation Options’ tab. An example of this is the ‘Personal gross weekly income’ variable, shown below:

As the above example shows, in regards to continuous items, TableBuilder users have the option to output the sum:

The option to output the median:

And the option to output the mean:

Users should also note that, in order to analyse these continuous items in more depth, they can create custom ranges by selecting ‘Range’ under the item in question and entering the ranges and increments that they wish to apply for that item. There may be some restrictions in place regarding the lower and upper bounds, as well as the minimum ranges that can be produced. For details on these, refer to the Data Item List located in the Data downloads section of this product.

An example of one way that the aforementioned ‘Personal gross weekly income’ continuous item can been ranged in TableBuilder is shown below:

TableBuilder users should be aware that some continuous data items are allocated special codes for certain responses (e.g. 000 = ‘Not applicable’). When creating ranges in TableBuilder for such continuous items, these special codes will automatically be excluded and will not be included in these ranges (e.g. for the data item ‘Hours usually worked each week (all jobs - continuous)’, 0=“Not applicable”). However, note that labelling of ‘0’s in the Data Item List does not necessarily mean they are excluded from the ranges as they may still be important in some calculations (e.g. for the data item ‘Overall life satisfaction (continuous)’, 0=“Not at all satisfied”).

If the additional special code populations are required for analysis to build up the applicable population further, they can be accessed via the version of the data item located outside the summation options:

This data would present as:

The population used in the previous calculation examples are found with ‘A valid response was recorded’ with the remaining populations presented for categories as outlined in the Data Item List.

There are a number of features within TableBuilder which can be used to confidentialise data as well as to improve usability or understanding of the limitations of the data.

In accordance with the Census and Statistics Act 1905, all data in TableBuilder are subjected to a confidentiality process before release. This confidentiality process is undertaken to avoid releasing information that may allow the identification of particular individuals, families, households, dwellings or businesses.

Standard processes used in TableBuilder to confidentialise records include the following: 

• perturbation of data
• table suppression
• np (not published) of relative standard errors.

In addition, the PSS TableBuilder has utilised the following features:

• top coding and min/max values
• field exclusion rule.

To minimise the risk of identifying individuals in aggregate statistics, a technique is used to randomly adjust cell values. This technique is called perturbation. Perturbation involves small random adjustments of the statistics and is considered the most satisfactory technique for avoiding the release of identifiable statistics while maximising the range of information that can be released. These adjustments have a negligible impact on the underlying pattern of the statistics.

After perturbation, randomly adjusted individual cells will be consistent across tables, but the totals in any table will not be the sum of the individual cell values. The effects of perturbation on the data can vary, depending on the level of detail and the size of the estimates, with perturbation more likely to have a greater relative effect on the adjustment of smaller estimates compare to larger estimates.

However, please be aware that the effects of perturbing the data may result in components being larger than their totals or components may sum to be much lower than their totals. This includes determining proportions. The more detailed a table, the more likely the data will appear incongruent to their total due to the small populations involved. It is therefore important to also consider the RSEs when assessing the reliability of the data.

The introduction of perturbation in ABS statistics ensures that estimates produced in TableBuilder are consistent with published statistics. Error values produced using TableBuilder may differ slightly to published error values for the same estimates. These differences are negligible and both sources are considered valid.

Some tables generated within TableBuilder may contain a large number of cells with very low counts (excluding cells that have counts of zero). When this occurs, all values within the table are suppressed in order to preserve confidentiality. The following error message is displayed at the bottom of the table to indicate when table suppression has occurred:

Data items with a large number of categories are more likely to contain cells with low counts, as the data is spread out over many categories rather than concentrated in just a few. If table suppression occurs as a result, users may wish to try grouping categories together to produce cells with larger counts in them. In order to allow the most flexibility with the data, categories have not been pre-grouped and are presented individually. However, in data items where data across the categories is sparse, users may need to group some of the categories in order to produce robust estimates.

In addition, data items which contain a small number of categories and also include a special category, such as ‘not known’ or ‘refusal’ may also trigger table suppression due to these special codes generally containing minimal data. If a table is suppressed in these situations, it is recommended that the special codes be grouped with another category, or they be excluded from the initial table creation so that data for the other categories can be accessed. If a total population is then needed, this can be run in a separate table by grouping all categories together.

The RSE of an estimate may be suppressed and displayed as ‘np’ (not published). This occurs because the RSE cannot be estimated reliably, and in this case the RSE should be interpreted as being greater than 50%. Low prevalence experiences and data cross tabulated at finer levels of detail are more likely to have high RSEs as the estimates become smaller and as a result will be assigned an ‘np’. Users should therefore give consideration to the reliability of the data being produced when undertaking analysis.

For more details, refer to Using Weights and Producing RSEs in TableBuilder section of this product and Relative Standard Error page in the TableBuilder, User Guide (cat. no. 1406.0.55.005).

As mentioned in the ‘Continuous items’ subsection of the Using Different Types of Data Items in TableBuilder section, continuous items may have a minimum or maximum value applied as well as a minimum range restriction which limit the range that data is available within. Top coding may have also been applied to categorical items (such as age). These have been applied to minimise the risk of identification as well as for general usability purposes where data is sparse at lower and upper ranges. For details regarding these values, refer to the Data Item List located in the Data downloads section of this product.

Note: Minimum and maximum values applied for ranged data do not impact the calculation of means/medians which are based off of the source data.

In the 2016 PSS TableBuilder a Field exclusion rule has been applied to the Violence Prevalence level.

Field exclusion rules restrict the number of data items that can be added to the same table. A Field exclusion rule has been applied to all items on the Violence Prevalence level with the exception of the Violence Prevalence Index item.

If TableBuilder users attempt to use more than one of the items to which this Field exclusion rule has been applied, users will see the following message:

The main purpose of implementing this Field exclusion rule is to prevent the ‘Relationship to all perpetrators ever experienced violence since age 15 per violence type’ data item from being used with the timeframe data items. However, it also means that the timeframe items cannot be used together in the one table. For most purposes this would not be a problem, as TableBuilder does not allow concatenation of items, only nesting. Therefore it is only possible to build timeframe data as individual tables. However if cross tabulation analysis is required, this can be undertaken using the Detailed Microdata product or via a customised data request.

As mentioned in the Weights subsection of the Notes on Data Items and Special Categories section of this product, the PSS is a sample survey, and therefore a weight needs to be applied to the data to produce estimates for the entire in-scope population. TableBuilder automatically assigns this weight when producing tables.

In addition, when producing means, medians, and other calculations for continuous data items, where a person has multiple records contributing to the data table, (such as on levels below the Person level), TableBuilder utilises the weight that has been applied to each record a person has on the level. Therefore, for the lower levels of the file it is very important that the Index item is utilised to ensure that there is no duplication of a person’s weight in individual cell calculations.

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 by the scope of the survey) had been included in the survey.

One measure of the likely difference is given by the standard error (SE), which indicates the extent to which an estimate might have varied because only a sample of dwellings was included. There are about two chances in three (67%) that the sample estimate will differ by less than one SE from the figure that would have been obtained if all dwellings had been included, and about 19 chances in 20 (95%) that the difference will be less than two SEs.

Relative standard error (RSE) is a measure of sampling variability. The RSE is obtained by expressing the SE as a percentage of the estimate to which it is related.

Data users should note that TableBuilder automatically produces the RSE of the estimate. Users can output RSE values for a produced table by clicking on the ‘Options’ button, hovering over the ‘Relative Standard Error’ tab and selecting either ‘RSE’ or ‘Summation + RSE’. Selecting ‘RSE’ will simply display the RSEs in each cell of the table. On the other hand, selecting ‘Summation + RSE’ will show both the estimate and the RSE highlighted in red, as shown below:

For more information relating to RSEs, refer to the Data Quality and Technical Notes page of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003), and for using RSEs in TableBuilder the Relative Standard Error page of the TableBuilder, User Guide (cat. no. 1406.0.55.005).

Note that at this time it is not possible to produce Margin of Error values (MoEs) for proportions in TableBuilder, which were produced for Personal Safety, Australia 2016 (cat. no. 4906.0). The formula for calculating MoEs using proportions and RSEs is provided in Data Quality and Technical Notes page of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

A comprehensive list of the data items available on the 2016 PSS Detailed Microdata file can be found in the Data downloads section of this product. This Data Item List also includes the relevant population and classification details of all available data items. Additionally, the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003) provides topic definitions and interpretation points. Specific details regarding the following technical aspects of using data items in the Detailed Microdata is outlined below.

• Index items
• Multiple response data items
• Continuous data items.

These topics have also been covered more generally in the Notes on Data Items and Special Categories subsection under the File Structure and Content section and should be read along with the following information.

Each of the four levels that are below the Person level include an Index item, which must be included when looking at data on the lower levels. The Index item is used to specify which experience(s) the other data items relate to.

For example, if one is interested in data about whether anxiety or fear was experienced in the 12 months after the most recent incident on the Violence Most Recent Incident level, the Index item is required to specify which one (or more) of the types of violence this relates to, (e.g. physical assault by a male, physical assault by a female, etc.). Similarly, if one is interested in data about whether anxiety or fear was ever experienced due to partner violence on the Partner Violence level, the Index item is required to specify which of the two types of partner violence this relates to – current partner violence or previous partner violence. In this manner, all data on the lower levels must always be filtered through one or more of the categories contained on the Index item.

An example of using the Index item when analysing data on the Detailed Microdata files is: if analysis is being undertaken on whether respondents ‘Ever experienced anxiety or fear due to partner violence’, the Partner Violence Index item should be specified when producing anxiety or fear data. To do this, the following SAS code (or equivalent) can be used:

LIBNAME PSS16 “<Insert file's location name>” ACCESS=READONLY; /*Specifies directory path for where the PSS16 datasets are located*/

OPTIONS FMTSEARCH=(PSS16) MISSING = ‘0’; /*Applies formatting file to PSS datasets*/

PROC TABULATE DATA = PSS16.PSS16PAR;
CLASS PVQ49 TYPEPAR;
TABLE (PVQ49 ALL = ‘TOTAL’), TYPEPAR * PSSWGT = “ * SUM = “;
VAR PSSWGT;
RUN;

This code produces the following cross tabulation, where TYPEPAR is the Partner Violence Index item, PVQ49 is the ‘Ever experienced anxiety or fear due to partner violence’ item and PSSWGT is the Person weight.

In this example, there are 16,227,831 persons who did not experience partner violence. There are 431,485 persons who experienced current partner violence and 1,767,401 persons who experienced previous partner violence.

Note: Index item categories should not be added together to produce a total (such as would happen in SAS if a ‘PROC FREQ’ was to be run). If a total is appropriate (such as is the case on the Violence Prevalence level), these have already been derived for users to use in the Index item and data for the associated data items on the level have been produced to appropriately reflect the responses required for the new ‘total’ record.

For more details on the Index items available on each level and their use, refer to the relevant level pages located under the File Structure and Content section of this product.

A number of data items produced from the 2016 PSS allow a respondent to provide more than one response. These are referred to as multiple response data items. For example, for ‘Physical assault behaviours experienced during most recent incident’ respondents may have experienced multiple assault behaviours. In instances where a data item allowed for more than one response to be given this has been noted in the Data Item List as a “Multi-response item” directly beneath the data item name.

In addition, each response in a multiple response data item has been separated into a distinct item (represented by a letter at the end of the common SAS name). For example, in the above item, if a respondent had been slapped and beaten, they would have data in MRPABEHC = 3 (Slapped) and MRPABEHF = 6 (Beaten). However, note that for most items, categories which are considered to be special categories (such as ‘not known’, ‘refusal’, ‘not applicable’, etc.) are usually placed in position ‘A’ (as they generally cannot be answered in conjunction with another category). This should be confirmed prior to using multiple response items to ensure that correct placement is accounted for.

When using multiple response data items, the total number of responses for outputs will be greater than the total number of respondents. Where data items allow for multiple responses, the components are not able to be aggregated to form a total. For example, with the above physical assault behaviours item, the data may present as:

In comparison to the summed total of the individual physical assault behaviours (2.9 million), the actual total weighted estimate for the population aged 18 years and over who experienced physical assault by a male in the last 10 years is lower (2.4 million). The difference of 500,000 persons is due to persons that experienced more than one type of physical assault behaviour being counted multiple times when the categories are individually summed.

To produce the actual total weighted estimate for use with multiple response items, that only counts persons once in the total:

• the applicable Index item category could be used on its own
• an aggregate item reflecting the population wanted (i.e. ‘Whether experienced physical assault since age 15’) may be used
• a new derive could be created using the multiple response data which only counts a person once.

Any additional restrictions, as presented in the population for the data item in the Data Item List, should also be taken into account (for example, sexual threat by a male data is generally only available for female respondents and therefore the sex variable located on the Person level would need to be merged onto the Violence Most Recent Incident level and used in conjunction with the Index item category 3).

The Detailed Microdata file includes a number of continuous variables which can have a response value at any point along a continuum. When using the 2016 PSS Detailed Microdata file users should note that some continuous data items are allocated special codes for certain responses. When analysing continuous items, it is necessary to exclude the special codes. The special codes are used for responses that do not represent the data being collected (e.g. “Not known”). The codes vary, but will generally be 0, 96, 97, 98, 99 or variations of these. For example, the data item ‘Personal gross weekly income’ (INCFULLR) has reserved values of:

• 99996 for ‘No source of income’
• 99998 for ‘Not known’
• 99999 for ‘Refusal’.

The Data Item List provides the special codes for all continuous data items available on the 2016 PSS microdata file. Care should be taken to exclude these codes when categorising higher values for ranges, and when calculating means, medians and other summary statistics. Refer to the Data Item List in the Data downloads section of this product for details on what codes are used for each data item.

Every record on each level of the file is uniquely identified.

The identifiers ABSHID, ABSPID, ABSVID, ABSRID, ABSCID, ABSEID appear on all levels of the file. Where the information for the identifier is not relevant for a level, it has a value of 0.

Each household has a unique thirteen character/digit random identifier, ABSHID. This identifier appears on the Household level and is repeated on each level for all records pertaining to that household. The combination of identifiers uniquely identifies a record at a particular level as shown below.

The Household record identifier, ABSHID, assists with linking household characteristics such as geography (located on the Household level) to the Person records. When merging data with a level above (i.e. Household or Person), only those identifiers that are relevant to the level above are required.

Note: Levels below Household and Person should not be merged with each other. More details regarding this are provided in the Using Data Items Across Levels section of this product.

As the Household and Person levels each only contain one record per person, combining data from either of these levels with the lower levels is a simple process.

Identifiers (as outlined above) can be used on records at each level of the file to merge information from the Person level to the lower levels. Each person has a unique random identifier – ABSHID, which appears on all levels of the file.

There may be instances when a data item is not contained on the level of the file required by users. For example, information about the characteristics of a person's most recent incident of violence for up to eight types of violence is contained on the Violence Most Recent Incident level, while that person's geographic and demographic information is contained on the Household and Person levels respectively. To apply Person level characteristics to the Violence Most Recent Incident level, the Person level characteristics (on the Person level file: PSS16PER) must be merged on to the Violence Most Recent Incident level file (PSS16MRI). The Person level is merged on to this level using the person identifier (ABSHID and ABSPID). To do this, the following SAS code (or equivalent) can be used:

PROC SORT DATA = PSS16PER OUT = SORTED_PER; /* Create a sorted temporary dataset based on the Person level level */
BY ABSHID ABSPID;

PROC SORT DATA = PSS16MRI OUT = SORTED_MRI;
BY ABSHID ABSPID;

DATA MERGEFILE;
MERGE SORTED_MRI (IN = A) SORTED_PER (KEEP = ABSHID ABSPID AGECTB SEX IN = B);
BY ABSHID ABSPID;
RUN;

The KEEP statement includes all person data items specified to be merged on to the Violence Most Recent Incident level file. Person characteristics can be merged on to the Violence Most Recent Incident level to identify the characteristics of the people who reported one or more violent experiences since the age of 15, and these characteristics are applied to each 'type of violence' record on the Violence Most Recent Incident level. In the above example, age (item AGE) and sex (item SEX) are merged onto the Violence Most Recent Incident level and are applied to each record a person has on that level.

Both the Household and Person levels can be merged to lower levels at the same time by including the household and person files in the merge statement with the lower level.

For the most part, data that can be aggregated from a lower level to a higher level have already been produced. In most cases, as a result of the ‘most recent incident’ or ‘most recent previous partner’ approach to detailed data on the lower levels, as well as having differing numbers of records per selected person, it is not appropriate to produce an aggregated response (i.e. it is not possible to gain an accurate response to whether violence has ever been reported to police, as it is possible that the most recent incident of each type of violence for a person may not have been reported (data collected), but previous incidents may have been reported (data not collected)).

However there may be some data where aggregate items have not been produced and need to be created. If users want to present information from one of the lower levels as person level data (as per the aggregate items that have already been produced), the data requires a ‘flattening’ process for records. That is, that information from multiple experiences are presented as an aggregate response. In most cases, as a result of the ‘most recent incident’ or ‘most recent previous partner’ approach to detailed data, it is not appropriate to produce an aggregated response (i.e. it is not possible to gain an accurate response to whether violence has ever been reported to police, as it is possible that the most recent incident of each type of violence for a person may not have been reported, but previous incidents may have been reported). However it would be appropriate to produce a count of how many types of violence a person has experienced. So there should be careful consideration of the appropriateness of producing additional aggregate items.

The following SAS code is an example of copying information from a lower level to a level above:

PROC SORT DATA=PSS16PRV OUT=SORTED_PRV; /* Create a sorted temporary dataset based on the Violence Prevalence level */
BY ABSHID ABSPID;

DATA TOT_VIO (KEEP=ABSHID ABSPID VIONUM); /* File to create a count of violence types. Note: only items required for merging and new items created should be kept */
SET SORTED_PRV;
BY ABSHID ABSPID; /* This step will go through each Violence Prevalence record within each unique combination of ABSHID, ABSPID */
RETAIN VIONUM; /* holds onto the count set by the previous violence record for each selected person */

IF FIRST.ABSPID THEN DO; /* sets new counting value to 0 each time it gets to a new selected person */
VIONUM=0;
END;

IF VIOINDX IN (100,101, 110,111,200,201,210,211) THEN VIONUM=VIONUM+1; /* Starts a count of the number of unique types of violence experienced (i.e. the 8 core types of violence collected) */

IF LAST.ABSPID THEN OUTPUT; /* This outputs the last record including the totals found for each unique combination of ABSHID and ABSPID so only a single record for each selected person is retained*/
RUN;

PROC SORT DATA=PSS16PER OUT=SORTED_PER; /* Creates a sorted temporary dataset based on the Persons level */
BY ABSHID ABSPID;

DATA MRGFILES;
MERGE TOT_VIO SORTED_PER;
BY ABSHID ABSPID;

PROC FREQ DATA=MRGFILES; /* This procedure gives a weighted count of the data copied up from the Violence Prevalence level to the Person level now that the data is a 1 record to 1 record match */
TABLES VIONUM;
WEIGHT PSSWGT;
RUN;

The new variable VIONUM gives a count of the number of unique types of violence experienced per selected person on the Person level. This new item can then be analysed with any other item on the Person level.

As the 2016 PSS is a sample survey, in order to produce estimates for the entire in-scope population weight fields must be used in any calculations. There is one benchmarked weight that has been produced for the 2016 PSS; the Person weight (PSSWGT), which has been benchmarked to the estimated resident Australian population aged 18 years and over who were living in private dwellings (excluding very remote areas of Australia) as at February 2017.

The weight for each person has been applied to their records across each of the file levels for the Detailed Microdata product. This means that for the Household and Person levels, the sum of the weights for all records will equal the estimated residential population of Australia, noted above.

Users need to ensure that weighting is applied to the sample when producing data, so that results are representative of the in-scope population rather than just the sample.

Note: Respondents may have more than one record on any level other than the Household and Person levels, which if the weights are summed, will produce a total in excess of the true total population. To avoid this, as has been outlined in previous pages, the Index items must be used in order to utilise only one record per person per Index category. This will produce the population who experienced that category without any duplication of the weight.

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 by the scope of the survey) had been included in the survey.

One measure of the likely difference is given by the standard error (SE), which indicates the extent to which an estimate might have varied because only a sample of dwellings was included. There are about two chances in three (67%) that the sample estimate will differ by less than one SE from the figure that would have been obtained if all dwellings had been included, and about 19 chances in 20 (95%) that the difference will be less than two SEs.

Relative standard error (RSE) is a measure of sampling variability. The RSE is obtained by expressing the SE as a percentage of the estimate to which it is related. 

For more information relating to RSEs, refer to the Data Quality and Technical Notes page of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003).

Another measure of sampling variability is the Margin of Error (MoE), which describes the distance from the population value that the sample estimate is likely to be within, and is specified at a given level of confidence. The PSS MoE values are calculated at a confidence level of 95%. At the 95% confidence level, the MoE indicates that there are about 19 chances in 20 that the estimate will differ by less than the specified MoE from the true population value (the figure obtained if all dwellings had been enumerated). The MoE at the 95% confidence level is expressed as the standard error multiplied by 1.96.

Detailed Microdata users should note that RSEs and MoEs are not included in the Detailed Microdata. As such, users will need to manually calculate RSEs and MoEs themselves should they wish to output this data. Data users can calculate the relevant SEs (and subsequent RSEs and MoEs) using the delete-a-group Jackknife replication method. Further information can be found in the Data Quality and Technical Notes section of the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003), under the ‘Replicate Weights Technique’ sub-section.

Alternatively, users may be able to utilise the available features in their statistical package of choice which perform these calculations. For instance, the SURVEYFREQ procedure in SAS can be used to produce prevalence rates and their corresponding SEs that can subsequently be used to calculate the RSEs and MoEs.

Users of the Detailed Microdata should also be aware that all data produced through DataLab, including accompanying RSEs and MoEs, is unperturbed, and as such may not match the same data published in Personal Safety, Australia, 2016 (cat. no. 4906.0), or obtained through the TableBuilder product.

These files contain the data for the Detailed Microdata in SAS format.

PSS16HHD.sas7bdat contains Household level data
PSS16PER.sas7bdat contains Person level data
PSS16MRI.sas7bdat contains Violence Most Recent Incident level data
PSS16PRV.sas7bdat contains Violence Prevalence level data
PSS16PAR.sas7bdat contains Partner Violence level data
PSS16EMAB.sas7bdat contains Partner Emotional Abuse level data

These files contain the data for the Detailed Microdata in SPSS format.

PSS16HHD.sav contains Household level data
PSS16PER.sav contains Person level data
PSS16MRI.sav contains Violence Most Recent Incident level data
PSS16PRV.sav contains Violence Prevalence level data
PSS16PAR.sav contains Partner Violence level data
PSS16EMAB.sav contains Partner Emotional Abuse level data

These files contain the data for the Detailed Microdata in STATA format.

PSS16HHD.dta contains Household level data
PSS16PER.dta contains Person level data
PSS16MRI.dta contains Violence Most Recent Incident level data
PSS16PRV.dta contains Violence Prevalence level data
PSS16PAR.dta contains Partner Violence level data
PSS16EMAB.dta contains Partner Emotional Abuse level data

FORMATS.sas7bcat is a SAS library containing formats

A Test File has been created to allow researchers/analysts to become more familiar with the data structure and prepare code/programs prior to applying for or commencing a DataLab session. This aims to maximise the value of sessions by saving users’ time and resources once they enter the DataLab environment.

The Test File does not contain real data, and cannot be used for analysis. It mimics the structure of the actual microdata file in terms of the available data items and allowed values. All data on the file has been created through a randomisation process and is therefore false. Proportions within data items in the Test File will be similar to those in the real data. However, relationships between data items will not be intentionally maintained. It is extremely unlikely that a record in the Test File would match with a genuine record in the real data.

The Test File is available as a free download through the Data downloads section of this product. For further information users should email microdata.access@abs.gov.au or telephone (02) 6252 7714.

08/11/2018

This release includes corrections to two data items in the Data Item List.

Variables 'Whether experienced physical abuse before age 15' and 'Whether experienced sexual abuse before age 15' required the labelling of categories 1 and 2 to be reversed, so that 'more than once' experiences are category 1 and 'once' are category 2.

The associated TableBuilder and Detailed Microdata data products have also been corrected for this category labelling change.

The 2016 Personal Safety Survey (PSS) microdata is released as TableBuilder and Detailed Microdata (via the DataLab) products. Microdata files are released in accordance with the conditions specified in the Statistics Determination section of the Census and Statistics Act 1905. This ensures that confidentiality is maintained whilst enabling micro level data to be released. More information on the confidentiality practices associated with TableBuilder can be found on the Confidentiality page of the TableBuilder, User Guide (cat. no. 1406.0.55.005). Further information on the confidentiality practices associated with accessing Detailed Microdata via the DataLab can be found on the Using the Detailed Microdata page. For more information about confidentiality, see the ABS Confidentiality Series and How ABS keeps your information confidential pages.

For information on the institutional environment of the Australian Bureau of Statistics (ABS), including the legislative obligations of the ABS, financing and governance arrangements, and mechanisms for scrutiny of ABS operations, please see ABS Institutional Environment.

The 2016 Personal Safety Survey (PSS) collected information from men and women aged 18 years and over, and provides data on the nature and extent of violence experienced since the age of 15. It also collected detailed information about and provides data on men's and women's experience of: 

• current and previous partner violence and emotional abuse since the age of 15
• stalking since the age of 15
• physical and sexual abuse before the age of 15
• witnessing violence between a parent and partner before the age of 15
• lifetime experience of sexual harassment
• general feelings of safety.

The 2016 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 that has not been collected since 2012.

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, 2013 (cat. no. 4533.0)
• Bridging the data gaps for family, domestic and sexual violence (cat. no. 4529.0.00.002).

The statistics presented in this publication are indicative of the extensive range of data available from the survey and demonstrate the analytical potential of the survey results. Full details about all the information collected in the 2016 PSS are provided in the Data Item List which can be accessed in the Downloads tab of the Personal Safety Survey, Australia: User Guide 2016 (cat. no. 4906.0.55.003). Details about the data available specifically on the microdata products are available in the Data Item List which can be accessed in the Data downloads section of this Technical Manual.

For further details on the content and conduct of the survey please refer to the Personal Safety Survey, Australia: User Guide 2016 (cat. no. 4906.0.55.003). This guide includes information to assist users with interpreting and using the results of the survey, descriptions of the survey design and collection, and information on data quality.

The 2016 Personal Safety Survey (PSS) was conducted between 6th November 2016 and 3rd June 2017. The TableBuilder and Detailed Microdata were released approximately 10 months after enumeration was completed.

The microdata contains finer levels of detail for data items than what was published in Personal Safety, Australia (cat. no. 4906.0). For information on the level of detail provided, please refer to the Data Item List available in the Data downloads section of this Technical Manual.

Steps are taken to confidentialise the data made available on the microdata files in a way that maximises the usefulness of the content while maintaining the confidentiality of respondents selected in the survey. As a result, it may not be possible to exactly reconcile all the statistics produced from these microdata with each other, or with other published statistics. Further information about the steps taken to confidentialise the data is available through the How The ABS Keeps Your Information Confidential page.

Results from the most recent survey were published in Personal Safety, Australia (cat. no. 4906.0). Data from the earlier surveys can be found on the Past and Future Releases tab of that publication or by contacting the National Information and Referral Service on 1300 1350 70.

Much of the content of the survey is comparable to the 2012 PSS. However, there are some differences in sample design and survey methodology, as well as new content that has been added to the survey and some concepts having been revised in order to take into account new or emerging data needs.

There are comparability notes for each topic in the Personal Safety Survey, Australia: User Guide, 2016 (cat. no. 4906.0.55.003). This provides information regarding data comparability over time between the 1996 Women's Safety Survey, and the 2005, 2012 and 2016 PSS.

The information within this product should be referred to when using the microdata. It contains information including survey methodology, file structure, accessing and using the TableBuilder and using the Detailed Microdata, and the Data Item List.

Further information regarding the collection and interpretation of the data can be found in the Personal Safety Survey, Australia: User Guide 2016 (cat. no. 4906.0.55.003) publication.

Microdata products are available to approved users. Users wishing to access the 2016 PSS microdata should familiarise themselves with information available via the Microdata Entry Page.

The 2016 PSS microdata can be accessed using TableBuilder or Detailed Microdata (via the DataLab) products.

Any questions regarding access to microdata can be forwarded to microdata.access@abs.gov.au or phone (02) 6252 7714.