The Australian Bureau of Statistics (ABS) publishes Census, income, and socioeconomic statistics for statistical areas, not for individual addresses. Geocodio's Australian demographics, income, and housing appends find the area that contains each address and return the ABS figures for that area. For street-level matches, most fields come from the Statistical Area Level 1 (SA1), the smallest area the ABS uses for detailed Census data.

Geocodio's au-demographics, au-housing, and au-income appends add that data to an API request in one step, for single addresses or batches. If you're uploading a spreadsheet, you can choose any or all of these appends in the data appends step of the upload process.

Every figure describes an area. Geocodio doesn't return data about individual people or households.

Which Census data is available?

You can request one, two, or all three in the same request, such as fields=au-demographics,au-housing,au-income. Each counts as its own lookup.

These appends return statistics only, not the area codes. To get the SA1, SA2, and other codes too, add au-census to the same request. It counts as its own lookup.

What do the Census appends return?

Each category, or append, starts with a meta object that names the source, the year of the data, and the area that answered (geography and geography_code). Dollar amounts are in Australian dollars, and meta says so with "currency": "AUD".

• source: the publisher, "Australian Bureau of Statistics".

• The year of the data. au_demographics and au_housing have census_year, which is 2021. au_income holds three blocks with different vintages, so each block names its own (see au-income).

• geography: the area the values describe, sa1, sal, or poa. Personal Income is always sa2.

• geography_code: the ABS code for that area.

• currency: "AUD" on au_housing and on the census and personal_income blocks of au_income. Every dollar amount is in Australian dollars.

After meta come the tables, each keyed by its name. Every table has its own meta object with the ABS table_id and the universe (who or what the table counts), such as "Persons" or "Occupied private dwellings". The rows that follow are objects with a value and a percentage:

• value is the count, median, average, or score the ABS published.

• percentage is the row's share of the table's universe and appears only on rows that are counts.

• Medians, averages, and SEIFA results have a value only, under Total (or Score and Decile).

The full response schema is in the API docs.

Which area does the data describe?

The statistics describe the area an address is in, not the address itself. Geocodio uses the most detailed area the match supports: the SA1 for street-level matches, or the suburb or postcode when the match is only that precise. meta.geography names the level, and meta.geography_code gives the ABS code for that area. For street-level matches, it's the same SA1 that au-census returns.

Personal Income is the exception. The ABS publishes it for SA2s only, so personal_income is filled for street-level matches only, from the SA2 the address is in (meta.geography is sa2). For suburb and postcode matches it's null with no warning, because a suburb or postcode centroid can sit in a different SA2 from the address. For the few SA2s the ABS leaves out of Personal Income in Australia, such as Jervis Bay, it's null and _warnings says why.

A bare postcode query, such as 3030, returns one result for each suburb in the postcode. Each of those results reads the Postal Area for the whole postcode, not its own suburb, because the query didn't name a suburb.

An SA1 figure describes a few streets. A postcode figure can describe tens of thousands of people. If your list mixes match levels, group or filter on meta.geography before you compare areas. For more on what each level means, see Australian Census Geographies.

Which year is the data from?

• Demographics, housing, and census income come from the 2021 Census.

• Personal Income in Australia is the 2022-23 financial year.

• SEIFA is SEIFA 2021, built from the 2021 Census.

These appends have no year option and always return the newest data Geocodio has loaded. au-demographics2021, au-housing2021, and au-income2021 aren't valid fields: Geocodio skips them and adds a warning. Each meta names its year, so check it if your analysis has to stay on one year.

Demographics

Geocodio returns the categories exactly as the ABS publishes them, including the ABS's terms for Aboriginal and Torres Strait Islander peoples.

Housing

Tenure counts dwellings, not people. The tenure table counts occupied private dwellings, such as houses and apartments with someone living in them. A row of 90 means 90 homes, not 90 residents. The table keeps the ABS table name, but Geocodio returns the five tenure totals only, not the ABS breakdowns by landlord type or dwelling structure.

Income and SEIFA

au_income holds three blocks, one per ABS release. Each has its own meta, because each release has its own year and may use a different geography level:

There are two personal income medians. The census median is weekly, from 2021, and comes from the SA1, suburb, or postcode. The personal_income median is annual, from 2022-23 tax data, and comes from the SA2, so it's returned for street-level matches only.

What are the four SEIFA indexes?

The ABS builds SEIFA (Socio-Economic Indexes for Areas) from Census answers about income, education, jobs, and housing. Each index ranks every area in Australia using a different mix of those answers.

Which index should you use? To find disadvantaged areas, use IRSD, which measures disadvantage only. For a fuller picture of both ends, use IRSAD. The indexes can rank the same area differently, so pick the one that matches your question.

• Score: the ABS standardizes the scores so that across all SA1s in Australia the average is 1,000 and the standard deviation is 100. For every index, a lower score means a less advantaged area. Geocodio returns scores as whole numbers. Scores are for ranking areas, not for arithmetic.

• Decile: the area's rank, split into ten equal groups. Decile 1 is the lowest-scoring 10% of areas, and decile 10 is the highest-scoring 10%. The ABS ranks each area type separately across Australia, SA1s against SA1s and suburbs against suburbs, so deciles from different match levels can't be compared.

Why don't the rows add up?

The ABS makes small random changes to its counts to protect privacy, so rows can add up to a little more or less than their total, especially in small areas. Here is the au_demographics result for 1 Martin Place, Sydney NSW 2000, cut down to its "Total persons" table:

{
  "results": [
    {
      "formatted_address": "1 Martin Pl, Sydney NSW 2000",
      "accuracy_type": "rooftop",
      "fields": {
        "au_demographics": {
          "meta": {
            "source": "Australian Bureau of Statistics",
            "census_year": 2021,
            "geography": "sa1",
            "geography_code": "11703164412"
          },
          "Total persons": {
            "meta": {
              "table_id": "G01",
              "universe": "Persons"
            },
            "Total": {
              "value": 223
            },
            "Male": {
              "value": 118,
              "percentage": 0.529
            },
            "Female": {
              "value": 109,
              "percentage": 0.489
            }
          },
          ...
        },
        ...
      }
    }
  ]
}

Male (118) and Female (109) add up to 227, but the Total is 223. (For the same reason, the tenure shares in au_housing add up to close to 1 rather than exactly 1.) Use the Total row when you need the population, and treat small differences as expected.

Why is a value null?

When the ABS doesn't publish a median, an average, or a SEIFA score, usually because the area has very few people or private dwellings, Geocodio returns null for it, never 0. Treat null as unknown. A 0 is a real count of zero, and the percentage of a 0 or unpublished total is null.

For example, the SA1 for 29-55 College Crescent, Parkville counts 1,765 people but no occupied private dwellings. Every tenure row is 0 with a null percentage, and its median rent, household and family income medians, average household size, and every SEIFA score are null:

{
  "fields": {
    "au_housing": {
      ...
      "Tenure and Landlord Type by Dwelling Structure": {
        ...
        "Owned outright": {
          "value": 0,
          "percentage": null
        },
        "Owned with a mortgage": {
          "value": 0,
          "percentage": null
        },
        "Rented": {
          "value": 0,
          "percentage": null
        },
        ...
      },
      "Median rent ($/weekly)": {
        ...
        "Total": {
          "value": null
        }
      },
      ...
    }
  }
}

If the ABS publishes no data at all for an area, that append is left out of the result, and _warnings explains why. You aren't charged for an append that's left out.

Cost

Australian geocoding and coordinate lookups cost 1 credit each. Each census data append adds 1 credit per address, so an Australian address with all three appends costs 4 credits in total. An append with no data for an address is left out and isn't billed.

Australia is available on all plans, including Pay-As-You-Go. See Country support.

Related guides

• Australian Census Geographies: SA1 and SA2 codes, to join any other ABS table

• Census: Demographics, Census: Income, and Census: Housing (the US equivalents)