A Project of IVT, ETH Zurich and WWZ, University of Basel
This work is licensed under Creative Commons.
Contact: Joseph Molloy (joseph.molloy@ivt.baug.ethz.ch)
Previous and future reports can be found at: https://ivtmobis.ethz.ch/mobis/covid19/en
7 Dec:
8 June:
31 May:
25 January:
14 December:
Previous news (click to expand)
18 November:
2 November:
6 October:
28 September:
24 August:
11 August:
6 August:
13 July:
29 June:
15 June:
25 May:
18 May:
11 May:
4 May:
27 April:
20 April:
13 April:
On March 16, 2020, 3700 participants who completed the MOBIS study between September 2019 and January 2020 were invited to reinstall the GPS Logger and Travel Diary App ‘Catch-My-Day’, developed by MotionTag. This voluntary recording of their mobility behaviour allowed us to track the impact of the various special measures during the unfolding pandemic. The pandemic is still going on one year later and many participants are still tracking.
The results are shown in comparison to those of the first weeks of mobility data from the original MOBIS Study which were recorded between 1st September and 1st November 2019, and thus serve as a baseline well before the pandemic hit Switzerland. Only trips inside Switzerland are currently considered, although data on cross border travel is available.
Participation decreased from about 1’300 participants to around 500 by the start of the second COVID19 wave in autumn/fall 2020 for any number of good reasons, such as a new smartphone, operating system updates, etc.. About 250 rejoined the panel after a second invitation in October 2020. We are very grateful for their engagement. Still, we happily agreed, when LINK offered to recruit more participants to the panel. This further increase of our sample allows us to complement the existing core. By mid-January a total of 393 additional participants had joined via LINK.
For the MOBIS study, participants were only eligible if they used a car at least 3 days a week - which skews the sample away from the Swiss general population. We did not impose a similar condition on the LINK-recruited participants as we are now aiming towards a more representative sample of the population. However, this means that the sample as of 2021 is no longer comparable to 2019 and 2020, as the mobility behaviour is quite different between the two groups. As such, for any analyses which makes comparisons to the pre-pandemic era, only link participants who meet the MOBIS-criteria are included.
The number of tracking participants each day used to calculate the average daily values, includes all participants who recorded tracks before or after that date. This allows the consideration of those who stay at home while still allowing for survey dropouts.
The GPS Travel diary used, Catch-My-Day (for iOS and Android) can have a 2-3 day delay before the tracks are available for analysis. The scaling by active participants accommodates for this, but the results of previous reports may change when the report is updated. The scales are calculated against the representative sample we obtained as part of the MOBIS recruitment process.
The colors in the graphs below are selected to indicate transport mode groupings. The greens indicate active modes, and the blue/purples public transport. brown is Car, and black is the total. These colors are consistent through following transport mode related graphs.
Here, the average daily distance travelled by participants is presented, differentiated by gender. To aid readability, a 7-day rolling average is used. The clear reduction in travel caused by the lockdown at the start of the pandemic is visible, as well as the gradual increase over the following months. The amount of travel is slowly recovering to pre-pandemic levels, as seen during the relatively normal period before the second wave in Autumn 2020.
[1] “Download chart data”
llustrated below is the number of participants who are ‘mobile’ on a particular day. That means that they logged some travel in the Catch-my-day app, even a short walk. The downward spikes indicate the weekends. The 7-day rolling average is given by the black line, which shows relative stability since the end of the first lockdown. This is the case even during the second partial lockdown. Various small variations are driven primarily by public holidays where people remain at home.
[1] “Download chart data”
Here we see the breakdown by travel mode. The values are given relative to the pre-pandemic behavior, calculated based on the average from September and October 2019. Particularly evident is the large increase in cycling observed during the lockdown, which was sustained throughout the summer of 2020. Public transport usage collapsed during the lockdown and recovered much more slowly than other modes. It is still at around only 50% of pre-pandemic levels.
Also presented in this section is a stacked version of this graph, with public transport modes grouped together. This graph makes it evident how the overall mode share has changed, with driving and cycling taking modal share away from public transport.
[1] “Download chart data”
The following two figures present regression estimates based on a Poisson model. Such a model can be used to estimate the average proportional change in a variable of interest while controlling for confounding factors. We control for weather effects and person fixed effects to absorb unobserved heterogeneity that is constant across time. The vertical lines or the colored bands mark the 90%-confidence intervals.
[1] “Download chart data”
This section presents additional indicators by mode, in addition to the daily distance that matches the previous graphs in section 5. Again, these graphs present the percentage change when compared to the baseline period in 2019. Particularly interesting is the variation in trip distance. Walk trips during the lockdown were much longer, but this behavior wasn’t sustained afterwards. On the other hand, bus trips have become shorter since the start of the pandemic, potentially driven by home-office trend.
[1] “Download chart data”
The graphs below present the change in average daily kilometers by various sociodemographic variables. The lines have been smoothed to improve readability. For some demographics, be aware that there are however only a small number of participants (see the distributions section). The difference by household size during and after the lockdown are particularly interesting, as is the behavior of the 25-35 age group in summer 2020. During the first lockdown higher income groups reduced their daily travel more.