Estimating temporal variation in transmission of COVID-19 and adherence to social distancing measures in Australia

May 15, 2020


A number of data streams provide information on mobility before and in response to COVID-19 across Australian states/territories. Each of these data streams represents a different aspect of population mobility, but they show some common trends — reflecting underlying changes in behaviour. We use a latent variable statistical model to simultaneously analyse these data streams and quantify these underlying behavioural variables.


Nick Golding, Freya M. Shearer, Robert Moss, Peter Dawson, Lisa Gibbs, Eva Alisic, Jodie McVernonDavid J. Price, James M. McCaw

Technical report posted online 15 May 2020. Access here.

Key messages

Assessment of adherence to social distancing measures

  • An analysis of trends in population mobility data streams up to 11 May was performed to assess adherence to social distancing policy.
  • This analysis suggests that adherence to social distancing measures may have decreased in the past four weeks (Figure 2).
  • Two waves of a national survey were conducted to assess how Australians are thinking, feeling and behaving in response to social distancing measures (Figures 3 and 4).
  • We used a statistical model to analyse the survey data and estimate a 37% increase (from 2.78 to 3.80) in reported daily non-household contacts between 3 April and 6 May.

Estimates of current epidemic activity

  • Due to very low case incidence, estimates of the effective reproduction number (Reff) using existing methodologies are becoming increasingly unstable, i.e., based primarily on model assumptions rather than actual case data.
  • To overcome this limitation, we report preliminary estimates of local transmission potential (Figure 7) from a new method which estimates components of the effective reproduction number. This method uses both daily incident case counts and outputs from an analysis of population mobility (Figure 1).

Forecasts of the daily number of new confirmed cases

  • Estimates of local transmission potential were input into a mathematical model of disease dynamics which was projected forward to forecast the daily number of new confirmed cases.
  • We report Australia-wide (Figure 8) and state-level (Figure 10) forecasts of the daily number of new confirmed cases up to 1 July, assuming that local transmission potential remains at its current estimated level.

Forecasts of the daily number of new confirmed cases under alternative future transmission scenarios

  • A scenario analysis was performed to assess the potential impact of increased transmission following the relaxation of social distancing measures from 11 May.
  • We project the daily number of new confirmed cases in Australia up to 1 July for three future scenarios: one where local transmission potential increases from 11 May to 1.1, one where it increases to 1.2, and another where it increases to 1.5 (Figures 9, S8, S9, and S10).

Technical reports and summaries