We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
ABS Census Test
Last registered on November 05, 2019


Trial Information
General Information
ABS Census Test
Initial registration date
November 04, 2019
Last updated
November 05, 2019 9:38 AM EST

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Behavioural Economics Team of the Australian Government
Other Primary Investigator(s)
PI Affiliation
PI Affiliation
PI Affiliation
Additional Trial Information
On going
Start date
End date
Secondary IDs
The Australian Bureau of Statistics (ABS) is responsible for the delivery of the Census of Population and Housing (the Census) every five years. Historically, a single ‘night of enumeration’ has been highlighted as the point in time to complete the Census. However, the ABS has always permitted early and late completion of the Census. To provide responders with greater flexibility and encourage greater participation, ABS/BETA conducted a test to advise responders they have a number of days to complete the Census, giving them a ‘response window’. We will deliver a two-arm cluster randomised trial to evaluate differences in the framing of a ‘response window’.
External Link(s)
Registration Citation
Greenwell, Harry et al. 2019. "ABS Census Test." AEA RCT Registry. November 05. https://doi.org/10.1257/rct.4984-1.0.
Experimental Details
We tested two approaches to framing the response window.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Proportion of households that commenced the survey prior to October 18 (i.e. up to and including October 17).
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Proportion of households that completed the survey prior to October 18 (i.e. up to and including October 17).
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
This was a cluster randomised trial. Clusters are small statistical areas called mesh blocks that consist of approximately 30 dwellings.
Participants were in the Australian selected statistical areas in New South Wales and Queensland. Participants were aware that they were invited into the Census test, but they were not be aware that they were part of a trial.
The intervention consisted of two versions of the invitation letter delivered by post.
Experimental Design Details
Not available
Randomization Method
Mesh blocks were stratified by: the difficulty of enumeration in the 2016 census, number of dwellings, enumeration mode in 2016 census (drop-off or mail-out), and mesh block category (residential, primary production etc). Mesh blocks were selected from strata in groups of four and randomly allocated by computer algorithm to condition in a one-to-one ratio.
Randomization Unit
The cluster was a small statistical unit called a mesh block.
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
Sample size: planned number of observations
approx 28,544
Sample size (or number of clusters) by treatment arms
446 clusters per arm
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
With an estimated ICC of 0.01, the effective sample size was 20,789 giving a minimum detectable effect of h = 0.04.
IRB Name
BETA Committee of Peers
IRB Approval Date
IRB Approval Number
BETA ETH2019-02
Analysis Plan

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information