Analysing the Data -
Methodology 2 - Census Benchmarking
In this section we compare the number of the police searches for each racial appearance group by ‘benchmarking’ those numbers against the total populations of those groups as identified in the census. This generates a search rate.
The starting point in this analysis is to acknowledge that the concept of race is socially and politically constructed. A person’s racial and ethnic identity can include one or more of their racial and ethnic ancestries. Furthermore, nationality or place of birth, does not necessarily indicate either identity or ancestry. Importantly, racial or ethnic appearance is a separate concept again. However, while the concept of race is socially constructed, the experience of racial discrimination is very real. Consequently, while acknowledging the ambiguities involved in this analysis, the methodology presented here lays an experimental foundation for evidencing systemic racism in policing.
In addition to ambiguities in the concept of race itself, there are compounding ambiguities as a consequence of the questions asked in the Australian census. At this point, the Australian Census does not collect ethnic identity or ethnic appearance. Instead, it records ‘ancestry’. One of the key ambiguities in the Census is that it allows people to record their ancestry as ‘Australian’. In 2021, 27% of Victorians described themselves as being of ‘Australian’ ancestry. What does this mean? Other than First Nations people, all Australians are immigrants. This 27% of people could be White appearing or appear like any other racial group. This is a critical issue that the census needs to address to enable effective research into systemic racism in Australia. The present study makes an assumption that people of ‘Australian’ ancestry group are White appearing. We make this assumption based on the ‘normalisation’ of whiteness in Australia that results in White people being more likely to consider themselves ‘Australian’ without qualifying this label than other groups (see Stanaway et. al. 2024)
A second key ambiguity is that the census allows people to record up to two ancestries. This means that the number of people of different types of ancestries is 127% greater than the total population in Victoria. To manage this issue we have divided the total number of people with a particular ancestry type by the total responses not the total population to find a percentage in the population figure. This will also generate errors. Because of these accounting errors, we offer these findings as preliminary only.
Racial appearance estimate calculations from the 2021 Census
This estimation has used ABS’s Tablebuilder tool to extract the ancestries of people resident in Victoria to create an estimated ‘racial appearance’ category from the 2021 Census. We create data using two assumptions.
Assumption 1: To estimate the number of people of ‘white’ appearance (Caucasian) we have added Australian, New Zealander, and North-West European, and Jewish categories together. The assumption made here is that the police can distinguish visually between Europeans who have ‘Mediterranean’ ancestry from other Europeans. Because Victoria Police have merged Mediterranean with Middle-Eastern, this analysis swamps the policing experience of people who appear Middle-Eastern with people who may be being treated as ‘White’. To remedy this problem we recommend that Victoria Police in future separate Middle-Eastern from Mediterranean.To estimate the number of people of ‘Middle Eastern appearance’ we have added North African and Middle Eastern (excluding peoples of Sudan, South Sudanese and Sudanese categories and Jewish) and including Afghan and Hazara groups. To estimate the number of people of ‘African appearance’ people we have added Sub-Saharan African with Sudanese categories (excluding South Africa). To estimate the number of ‘Pacific Islander appearing people’ we have used the Oceania (excluding New Zealander, Australian Aboriginal and Australian). To estimate the number of Asian people we have added South-East and North-East Asian categories. To estimate the number of people of Aboriginal appearance, we have used the Aboriginal Australian and Torres Strait Islander category.
Assumption 2: By way of comparison, in this estimate of the number of people of ‘white’ appearance (Caucasian) we have added Australian, New Zealander, and North-West European, Southern and Eastern European, and Jewish categories together. We assume here that the police are likely to ‘racialise’ all European appearing people as ‘white’ (Caucasian) and will, for example, record people of Italian and Greek ancestry as being of white (Caucasian) appearance. This produces a different set of data that you can see here.
2.1 Are the Victoria Police search rates of different populations proportionate to their population?
Because many Aboriginal people are not of ‘Aboriginal appearance’ in order to establish the total number of searches of Aboriginal and Torres Strait Islander people by police in 2022 and 2023, we added the people who police perceive to be Aboriginal to people the list of people who were recorded as being of Indigenous Status. Where people are listed as being of Indigenous status and Aboriginal appearance we counted them only once. Indigenous Status is recorded if a person has been arrested and detained by the police and is based on their self-identification. Figure 6 below sets out the search rate for racial appearance groups in Victoria as a proportion of their percent in the overall Victoria population. This means that the numbers disclose the search rate by Victoria Police compared with their percentage in the population. For example 3.6 means that the group is 3.6 times more likely to be searched than if the search rate reflected the groups proportion in the population. The groups included in this analysis are those who collectively experienced at least 200 searches during 2022 and 2023.
Figure 6 provides preliminary evidence that in 2023 Aboriginal and Torres Strait Islander Victorians are approximately 10 times more likely to be searched than their proportion in the population would predict. This means they are over-represented in the search data by 10 times. People who police perceive to be African are over-represented almost 6 times. People who police perceive to be Middle-Eastern are 1.2 times more likely to be searched than their proportion in the population, and people who are perceived by police to be Pacific Islanders are 4 times more likely to be searched than their proportion in the population. The preliminary evidence is that people perceived to be Asian, Indian, South American and people perceived to be Caucasian are underrepresented in the 2022 and 2023 Victoria Police search data. Tables 3 and 4 below provides the figures used to generate Figure 6.
Figure 7 below compares the search rates of different groups with the search rate of people perceived to be White.
Figure 7 reveals that in 2023, police were about 11 times more likely to search an ATSI person than a White person (based on their respective size in the population), they were about 6 times more likely to search a person they perceived to be African than a person they perceived to be White, were almost 1.2 times more likely to search a person they perceived to be Middle-Eastern/Mediterranean and were almost 4 times more likely to search a person they perceived to be Pacific Islander than a person they perceived to be White. They were less likely to search people they perceived to be Asian, South American or as being from or having ancestors from the Indian sub-continent than White people. The data to create this Figure was drawn from Tables 3 and 4 below.
2.2 How does Victoria compare with NSW?
The ABC obtained search data from NSW police between 2020 to 2024. The data they obtained, when compared to NSW 2021 Census data revealed the following: Aboriginal and Torres Strait Islander people are searched at 5.6 times the rate of White people. People perceived to be African are searched at 4.3 times the rates of White people. Pacific Islanders are searched at 3.0 times the rate of White people. People perceived to be Middle-Eastern are searched at 2.9 times the rate of White people. All other groups are searched at a lower rates to White people. The NSW findings are consistent with the findings for Victoria Police, although the disparities for Victoria Police appear to be greater for all groups except Middle-Eastern. The difference in finding for Middle-Eastern probably because Victoria Police combine ‘Middle Eastern’ appearance with ‘Mediterranean’. In NSW, these groups are separated and people who appear to be Mediterranean are searched at a lower rate than White people. You can see the analysis of the NSW data here. The ABC’s report on the data is here.
Discussion
This data reveals that people police perceive to be Aboriginal, African, Middle Eastern/Mediterranean and Pacific Islander are more likely to be subjected to a search without warrant than is expected from their respective proportions in the population. This disproportionality is indicative of systemic racism against these populations. When compared with people perceived to be white people, people perceived to be Aboriginal and Torres Strait Islanders are 11 times more likely to be searched than white people, people police perceive to be African are 6 times more likely to searched and people the police perceive to be Pacific Islanders are 4 times more likely to be searched than people perceived to be Caucasian. Using Assumption 1 (where police are able to visually distinguish between Caucasians and people who have Italian or Greek ancestry, Middle-Eastern/Mediterranean perceived people are 1.2 times more likely to be searched than White people. However, using Assumption 2 (police are likely to view all Europeans as ‘White/Caucasian’) Middle-Eastern appearing people are about 5 times more likely to be searched than White people. Data from NSW, where search records of Middle Eastern appearing people is separated from search records of Mediterranean appearing people and other Victorian data sources, indicates that Assumption 2 is likely to give a more accurate reflection of the experience of Middle-Eastern appearing people in Victoria. People perceived to be Asian, South American or from the Indian sub-continent are less likely to be searched than people police perceive to be Caucasian.
A census benchmarking analysis of search data from Victoria Police reveals systemic racism toward (that is the disproportionate over-policing of) all non-White groups, except Asians, South Americans and Indians.