Compiling information on human trafficking and slavery is not a straightforward exercise—a major challenge being the availability, consistency and completeness of data. Data are not always collected and where it is recorded, must be sourced from different data collection systems administered by different agencies with different agendas. Further, data are not always recorded in a systematic fashion, either within or between collection agencies, and may be limited in detail or scope (Aronowitz 2009). The latter is particularly relevant to victim-based data, due to the need to preserve anonymity and the frequent reluctance of victims to communicate their experiences. It can also be affected by the administrative purpose of the data recorded and the logistics of broader data collection.
This section examines the nature and type of administrative and other data on human trafficking and slavery collected by Australian Government and non-government agencies identified as the likeliest sources of data for the monitoring program ie the AFP, CDPP, DFAT, DIBP, DSS, the Australian Red Cross and other prominent service provders such as Salvation Army Samaritan Accommodation. While international data sources are likely to be valuable in propulating aspects of the framework (specifically contexual factors), for reasons of practicality and to fulfil the purpose of the monitoring activity (ie the Australian situation), primary data providers will be confined to Australian-based agencies.
The section discusses the primary issues affecting the collection and utility of human trafficking and slavery data in Australia, as it relates to the conceptual intention of the proposed framework and what constitutes the manageable development and use of a centralised data collection. The following discussion is written from the perspective that the AIC will be the administrator and data custodian for the monitoring program if it is established.
Data collection and utility issues
There are two elements determining data availability—whether specified data are currently collected (or proposed for future collection) and the suitability of existing data to populate the conceptual framework. Information provided by consulted agencies was mapped against the framework’s conceptual elements to denote data availability and identify information gaps (see Table 15).
Much of the human trafficking and slavery data that is collected by these agencies falls within the conceptual elements of Incident and Response. Such a confined scope was not unexpected. Comparable monitoring developments and datasets described earlier also found that the majority of available data related to these themes (Aronowitz 2009; Banks & Kyckelhahn 2011; BKA 2010; Eurostat 2013; National Rapporteur on Human Trafficking in Human Beings 2010). These data represent what is largely recorded in administrative datasets and ultimately reflects the purposes of capturing that data; that is, for operational and statistical reporting, performance monitoring and similar purposes. Hence, the majority of data collected in Australia (and elsewhere) focuses on the nature of the crime (ie the type of trafficking or exploitation), the persons involved in the trafficking event (victim and offender characteristics), criminal justice responses (investigations, arrests/charges, convictions) and service provision specifics.
Qualitative material (such as client case files and court transcripts) may fill the gaps left by operational and other administrative data, particularly about the trafficking event and additional detail on recruitment method, migration pathways and means of exploitation. Mining such resources can produce a more developed picture of indicator themes that cannot be easily addressed by administrative data (see, for example, analyses undertaken in Simmons et al. 2013) but the cases available for analysis may not be representative of all trafficking and slavery events. Further, this method of data extraction is a time-intensive undertaking and at variance with the ultimate intent of establishing a data collection-based monitoring program.
Data describing Context and Impacts and Outcomes are largely missing, similar to human trafficking data collections administered elsewhere. While some contextual information may be drawn, for example, from case file notes prepared by service providers, or obtained from summary reports describing regional issues and anti-trafficking programs, baseline data is generally not available, cannot be shared and/or mostly presented in narrative format that precludes easy or reliable conversion into data items. Little to no information is consistently collected on Impacts and Outcomes, other than the potential to gauge reoffending and outcomes for persons who received support through the STPP. Stakeholders noted the importance of including these elements in the conceptual framework but conceded that it was unlikely such information would be routinely collated, outside of what service providers might record as part of their ongoing support relationship with clients.
Populating the conceptual element Risk is dependent on the availability and scope of data used to monitor other conceptual elements. At present, basic risk profiles might be developed using victim characteristics compiled by service providers (which appear to be the most detailed), although customary demographic characteristics would need to be matched between sources and corroborated with those recorded by law enforcement and the CDPP. Measuring actual risk—incidence and prevalence—is less straightforward, particularly when using victim numbers as the numerator. Relying on victim data from one source is an unreliable measure as it only captures the population who have come into contact with that particular agency. Equally problematic is aggregating victim numbers from different sources. The current absence of formal data linkages or similar arrangements between relevant agencies prevents the identification of unique individuals, leading to the well-recognised potential for double counting.
Data consistency, completeness and comparability
Consistency refers to the uniformity of data recorded over time and by different data custodians. Resource constraints and periodical modifications to improve or streamline data collections affect the consistency (and comparability) of time-series data. A number of agencies specified that consistency issues with data collected in the past would prevent them from providing historical data.
Inconsistencies in data also arise where standardised data recording practices—in the form of data entry rules and defined data fields—are not followed or cannot be applied to existing data recording tools, potentially compromising aggregation of data from different cases and different time periods. A number of agencies consulted for this project explained that they use, or are introducing, standardised data entry inventories to specifically collect or improve their current data collection on human trafficking and slavery.
|Context||Risk||Incident||Response||Impacts and Outcomes|
Incidence/prevalence—Offender population group charged/arrested
Exploitation—Type (as per Criminal Code offences recorded) and Industry
Referrals (into AFP and to other agencies)
Action and means—Number of offenders per event, number of victims per event, role of offenders, method of coercion and force
Exploitation—Type (as per Criminal Code offence(s))
Offender—Gender, age, birth country
Criminal justice response—Referrals/ referring agency
Brief and committal outcomes, plea
Convictions and other outcomes, length of time between filing and disposition
Appeals filed/appeal outcome
Re-entry into criminal justice system: offenders
Victim—Age, country of origin and visa statusb
|Australian Red Cross||n/a||+||+||+||O|
|Incidence/prevalence—Population group comprises persons referred by the AFP to the STPP. Number of clients, number of new clients, number of exited clients
Population group does not capture:
|Victim—Gender, age, country of birth, nationality, marital status, number of dependents employment status, visa at referral, current visa status
Exploitation—Type of trafficking/exploitation, location, industryc
|Type of support provided and length of support provision
Personal and systemic barriers to support—recorded where identified
|‘Positive’ outcome of support and other assistance—caseworkers to add ‘from their perspective’|
|Salvation Army||O||+||+ O||+||O|
|Client information on life history and circumstances that preceded the trafficking event||Prevalence count—Population group comprises number of clients receiving support. Includes (a) clients who have reported to the AFP and/or state/territory police and (b) clients who have not reported to police||Victim demographics
Action and Means and Exploitation data potentially from assessment tools and case notes. Generally not recorded in database
|Type of support provided to clients|
a: Excludes the element Research and Evaluation
b: Demographic data on persons referred to the AFP or those transferred to visas under the Human Trafficking Visa Framework varies in detail and the extent to which it is recorded in administrative databases
c: As determined by the AFP
Note: n/a Data not available + Data available and electronically recorded O Data might be available and/or is mostly not in electronic format
Further, some were aiming to develop a separate dataset that would integrate relevant information presently housed in separate databases. Overall, careful data auditing and follow-up analysis is recommended for the proposed data development phase to assess the consistency and reliability of requested data items.
Tied to consistency is data completeness. Where data entry is undertaken by multiple persons or in multiple settings, and particularly where data entry guidelines are absent, there is the additional risk that data are being recorded in variable levels of detail or completeness. It appeared that variability in completeness tended to affect the inclusion of useful contextual or additional explanatory material, although on occasion, the population of specific fixed data fields too.
Resources and time pressures are a consistent factor that affect the completeness of data but that also has impact when data, by necessity or procedure, are recorded at different junctures (usually as it became available). For example, some of the data recorded by support programs are subject to a client’s willingness to provide information about themselves and their circumstances. That information may never be provided, or is only provided at different stages of engagement. Similarly, operational data is recorded at different stages as the assessment, investigation or prosecution progresses. Episodic data compilation is not unique to the collection of information on human trafficking and slavery, nor does it necessarily affect the quality of the chronicle of data eventually contained in the finalised record. It does, however, require the application of consistent rules as to how complete a record needs to be before the data contained in that record can be extracted or subsequently used.
Finally, a critical step in collating similar data from multiple sources is to appreciate the ‘source and origin of data…and method of collection’ (Aronowitz 2009: 32) and to recognise the potential impact on data comparability. Data and information collected by the agencies consulted is used for a variety of purposes. These uses include program, agency and interagency committee reporting requirements, submissions and media releases, and contributions to other reporting mechanisms such as criminal history and sentencing databases. Depending on the agency, this information is also vital in providing evidence for case management, either for victim support or investigative purposes.
A significant issue in aggregating human trafficking and slavery data is the absence of any formal mechanism to link data collected in different systems. Data collections administered by the consulted agencies did not include a formal linkage key or unique identifier variable and there was limited scope to incorporate such an identifier. This not only affects the capacity to crossreference data across collections administered by different agencies but even to reconcile data within individual agencies.
The biggest gap is the ability to link data on offenders and incidents to data collected on victims. Victim support services, for example, collect information on victim characteristics but little on the trafficking and exploitation situations experienced by their clients. By contrast, law enforcement authorities collate information on offenders and some detail on the specifics of the offences being investigated or prosecuted but minimal information on victims, although this was an area targeted for enhancement for at least one agency interviewed for this project.
Formal data linkage arrangements have become relatively commonplace in Australia, primarily involving linkage between health, welfare and some criminal justice administrative data. The establishment of linkage arrangements are, however, dependent on the quality and comparability of the data. It is also dependent on the establishment of strict protocols around the protection of privacy. One option recommended to manage privacy considerations is to create a master dataset that uses a project specific identifier, rather than to establish direct data linkage between source data collections (see, for example, AIHW 2005).
This option, if acceptable to data providers, still requires a process to identify individual cases across data sources. In the absence of a routine method of reconciling data collected across administrative data holdings (eg between data recorded by the AFP and CDPP), some collaborative arrangement involving affected data providers and the data custodian is needed to confidently link cases. The simplest approach is to use a variable (such as the victim’s name) that is likely to be recorded across data holdings but may still require some substantial back and forth between contributing data providers before reconciled data can be delivered to the data custodian. Linking these cases to any data that can be provided by non-government agencies is less straightforward, due to variation in what is recorded and in what format, and consent issues (see below).
Data linkage limitations also create issues of double counting. This predominantly affects victim data; for example, where victims are accessing multiple services at once and there is no formal information sharing between services or agencies. Occasional crossreferencing between support services does occur, but it is not standard process and as noted above, is only ever undertaken when the victim consents. The risk of double counting victims of human trafficking and slavery is recognised and even the most carefully prepared estimates acknowledge potential overestimation where crossreferencing is not comprehensively enacted (see, for example, Lebov 2009).
The second set of considerations included the conditions and logistics of data provision, specifically:
- consent and privacy provisions;
- the resources required for data extraction and preparation; and
- current information-sharing arrangements and recognition of formal agreements between data providers and data custodian.
Consent and privacy
The primary concern raised by stakeholders on the issue of consent and privacy related to the provision of victim data. This concern was largely expressed by non-government agencies who indicated that that the release of victim data to third parties would be limited to general demographics and the type of exploitation experienced. Any release of additional information required the consent of their clients. Some government entities were similarly reticent about the provision of more detailed victim data because of the small number of victims.
At the core of these reservations is the requisite to preserve anonymity. One method commonly employed to reduce the chance of identification is to collect aggregate rather than unit record data. However, even where victims have consented to their information being included in a monitoring program, there is the real chance that only a small population of cases exists, which then potentially exposes individuals to being identified. This can be dealt with to some extent through strict adherence to rules around the treatment of small cell sizes by suppressing any data where the population is five or less or, for population of more than five, including caveats around the representativeness of the population considered and constraints on how case material is used or publicised. Other monitoring programs administered by the AIC adhere to the princples outlined above regarding the publishing of aggregate data and statistics. Additional protocols include:
- restricting access to the dataset to designated staff only;
- seeking approval for access to data from the AIC’s Human Research Ethics Committee; and
- entering into data sharing arrangements with external data providers.
The term resources is used in this section to refer to the systems in place to record and report on data and the capacity to provide data on an ongoing basis to a monitoring program. There was, not unexpectedly, variation in the type of databases used by the different agencies consulted but significantly, data extraction was considered by most (but not all) as a relatively straightforward procedure. In a few cases, however, information was not routinely entered into a database, with material retained in paper format. This was largely a resource issue based on time and staffing constraints. It was noted that where this occurred, the flexibility of the system operated could allow for future inclusion of data items not currently recorded or an alternate arrangement might be made, resources permitting, to construct a dataset that collated information of specific use to both the monitoring program and the collection agency. A number of agencies were also reviewing their current data recording tools, either to improve consistency in definitions and related data entry rules or to introduce new formats to assemble data pertinent to human trafficking and slavery. These developments would also improve the process of compiling monitoring data.
Contributing data to the proposed monitoring program was not considered an overly onerous undertaking if certain provisions were in place. Annual or biennial data transmission dates were preferred, provided data preparation did not coincide with end of financial year (and similarly intensive) reporting timeframes. The use of specially prepared data templates, with agreed data definitions for specified variables, would provide the necessary guidance in preparing data for transmission to the data custodian.
Similar kinds of arrangements have been adopted by the AIC in the administration of its other monitoring programs. These have included:
- formal undertakings with data providers that outline the purposes and specifics of the data provision/custodial arrangement;
- the development and piloting of data collection templates and data specifications in consultation with data providers (and an option to provide data in a different format if the template cannot be readily populated);
- clear data transmission guidelines (describing the what, when, how and who to of data transmission) and;
- changes to reporting practices to reduce burden, such as the move to biennial data collection and reporting.
Additional practices undertaken or recommended for the immediate future include examining the ‘lessons learned’ from these monitoring programs, specifically around the value of the variables collected and the efficiency of the procedures used to collect and collate the data.
The stakeholders consulted with differed in terms of the form in which they could provide the data. Some agencies indicated a preference for providing aggregate data, particularly where it is victim data being supplied, while others stated it would be less resource intensive to provide raw or unit record data.
An important point to make here is that data provision arrangements for a monitoring program on human trafficking and slavery is predicted to be a less fluid process than those for other AIC monitoring programs. For monitoring programs such as NHMP and NDICP, an identical set of standardised data items are transmitted by each of the primary data providers. For a monitoring program on human trafficking and slavery, variable combinations of data items will need to be sourced, from a mixed group of data providers, who operate distinct systems of data recording and retrieval. To cater for this, individual data templates will probably need to be created for each data provider (to populate different indicators) and careful reconciliation of data (particularly victim data) with and between provider agencies.
The agencies consulted with identified a range of formal and informal (ie by request) information-sharing arrangements. Most of these extended to interagency record reconciliation, co-case management of clients and for investigative purposes. One agency consulted was currently examining protocols for existing data sharing arrangements.
There was collective support for some form of formal agreement, such as a Memorandum of Understanding, between the data custodian (the AIC) and data providers. This type of arrangement would specify conditions around data provision (eg transmission timeframes, data format), ongoing requests, data storage and data use. It was also seen as a suitable mechanism for mitigating issues around privacy and confidentiality.