A Tale of Two Lockdowns

As Melbourne contemplates the possibility of Stage 4 lockdown, the current situation offers us an interesting comparison: how does the first Stage 3 lockdown compare to the second?

If you know me or have read my posts before, you will not be surprised that I have looked at this through the lens of traffic volumes. If you know me personally, you’ll be even less surprised that I brought in the 2018 Victorian state election results into the mix.

Reading recent media reports on the Coronavirus, you may have noticed politics beginning to overshadow health-advice. It’s no surprise that many news sources are reporting citizens confused about the lockdowns as well as increasingly polarised opinions about the right course of action.

Arguably, the Victorian opposition has been unhelpful at best throughout the pandemic. I don’t think I have witnessed a single member for the opposition express a view that is consistent with the government. Even Albo is finding common ground with his conservative counterpart and we all know how much Albo “loves to fight Tories”. Is this mixed messaging impacting our driving behaviours?

But what does the data say? Well before we get into that, I am going to make a few assumptions:

  • People’s need to drive is unchanged between the first lockdown and the second.
  • Traffic generated by those outside the lockdown zones is not significant within the lockdown zones.
  • Traffic volumes for the week 11 May 2019 to 17 May 2019 is a representative traffic week to serve as the baseline.
  • Each weekday is comparable to past occurrences (eg: all Mondays within the sample are comparable).
  • Driving age is 20 – 79 (because it’s an easy division in ABS Table Builder)

Changes to traffic volumes

Whether it be confusion, malaise or straight-up disregard, the second lockdown has not been as effective as the first at keeping people off the road. Although traffic volumes have again dropped to better than 25% below normal, on all days in all council areas there has been an increase in traffic when compared to the first lockdown. There is a strong indication that for whatever reason, fewer people are staying home.

Traffic volumes in Stage 3 lockdown council areas

Proximity to hotspots

Given the geographic spread of the illness across Melbourne, it would be sensible to ask if proximity to hot-spots is a driving force for keeping people out of their cars. This was my first suspicion and on first glance, it might check out.

Change in traffic volumes between lockdowns by LGA

Hume and Brimbank council areas stand out as their traffic volumes returned almost to the same levels as the March lockdown. Additionally, they rank second and fourth on the active case tally, respectively. Other hot-spots like Wyndham, Melbourne and Moonee Valley have not performed quite as well as their active-case leaderboard companions.

In an attempt to quantify the proximity effect, I created a “perceived safety” score based on the distance between council areas and the number of active cases. I then compared this to the growth in traffic between lockdown events.

Distance and cases versus traffic growth

Perhaps unsurprisingly, this didn’t yield much useful information. The size of Mitchell Shire severely skews the scoring for that area. Additionally, as-the-crow-flies distances between council areas lose meaning when geographic features like Port Phillip Bay greatly distort the distance via a straight-line when compared to the trafficable route. Further, the large size of council areas makes it difficult to assess the impacts of known hotspots on behaviour.

In summary, proximity to outbreak areas may be influencing lockdown observance in Melbourne’s residents, but it is difficult to tell.


Lockdown traffic growth by state electoral district and 2018 election results

The first thing to note with this map is that the single worst district in Melbourne for growth in traffic volumes compared to the last lockdown is Frankston, a safe-ish state Labor seat. However, the traffic growth in and around Frankston may be somewhat explained by the recent completion of major road works in the area rather than by disregard for the lockdown. SCATS data suggests significant growth at the intersection of Thomspon / Frankston-Dandenong following its reopening.

SCATS intersection volumes at Thompson / Frankston-Dandenong

I have my doubts that only 1,210 vpd were traversing the intersection prior to its completion. It is more likely that vehicle detectors were not operational during construction. However, the reopening would have resulted in a major increase in intersection capacity and as a major intersection, capacity benefits would have been realised within the local area as well.

Circling back to politics and upon looking at a plot of Liberal electoral fortunes versus lockdown traffic growth, there does appear to be some correlation. Given the weakness of this relationship, I would be more likely to argue that those flouting lockdown rules by taking unnecessary trips are more likely to lean right politically rather than suggesting the opposition has had a major impact on lockdown behaviour. Further, I would suspect the vast majority of right-leaning voters in Melbourne are adhering to the rules as strictly as their friends across the political divide.

Two-party-preferred outcome versus district lockdown traffic growth

In the graph above, I have highlighted Tim Smith and Michael O’Brien’s districts as they are some of the more vocal critics of the Andrews government. However, their electorates very much fall within the middle of the pack for traffic growth. For this reason, I reiterate that I don’t think the opposition’s contrarian messaging has had a meaningful impact on lockdown behaviours.

Car ownership

Before I put this investigation to bed, I wanted to see if access to vehicles was a driving force in determining the growth of traffic between lockdowns. Perhaps boredom combined with the temptation of a car in the garage has led to more trips on our roads.

Car ownership versus lockdown traffic growth; Liberal vote indicated by bubble size

While boredom may play a role for some people, it is not a consideration for the wider population. What strikes me about this graph is the bubble-size. It turned out that the average 2PP Liberal vote is 45% above the red line and 30% below the red line. I believe this is the strongest indication yet that right-leaning voters are more likely to be making unnecessary trips than others.


With any overly aggregated data-source, there are only so many insights we can garner. It is self-evident that the residents of Melbourne are not adhering to lockdown restrictions as fervently as they did back in March 2020. Additionally, there is evidence to suggest that there is a correlation between those who lean right politically and those who are more likely to make unnecessary car trips during the second lockdown. It also appears there is a limited relationship between proximity to cases and additional on-road traffic and no relationship relating to car ownership.

If politics is truly playing a role in the observance of the lockdown, it is just as important now as it was at the start of this crisis that government and opposition seek common ground and deliver a strong unified message to their constituents for the common goal of stemming the virus. As is the mantra of this pandemic: we are all in this together.


  1. COVID-19 active cases from DHHS PowerBI dashboard accessed 18 July 2020
  2. Traffic volumes from VicRoads via DataVic accessed 17 July 2020
  3. 2018 Victorian State Election Results from Victorian Electoral Commission accessed 18 July 2020
  4. Car ownership and age by district the Census of Population and Housing (2016) published by the Australian Bureau of Statistics

Leave a Reply

Your email address will not be published. Required fields are marked *