Guidance

Work Plan - Crowd Movement and Behaviour Study

Updated 8 September 2021

Introduction

There will be a series of sporting and cultural events taking place during April/May 2021 which will involve audiences and visitors in attendance. The principal aim of these events is to collect a scientific evidence base to inform the planned re-opening and operation of such venues and events in the context of COVID-19 transmission risk. This is known as the Event Research Programme (ERP). 

It is clear that in the context of developing a risk-based model for assessing events taking place in different settings, the focus is very much on capturing quantifiable information, and one of the key aspects relates to data capture and analysis of the motion and movement of attendees. As a result, this work plan has been developed.

Work Plan

We have identified that there are three phases of activity to which we can contribute: 

1. Scoping Phase (late March - early April)

2. Pilot Event Observation and Reporting Phase (mid April - mid May)

3. ‘Deep Dive’ Analysis and Modelling (May onwards). 

Scoping Phase (late March - early April)

In order to provide a clear indication of the type and scale of data collection activities that will provide the required evidence base across up to 12 venues, a scoping phase is required. This scoping phase will inform our capacity to contribute to the subsequent phases. We will undertake three different tasks during this phase:

Development and sign-off of Evaluation Framework

The evaluation framework will be focused on identifying the risk of virus transmission associated with attendance at events concentrating on spectator behaviour and venue operational response. A key input to this will be a Public Health view on which locations and activities represent a higher risk (e.g., time spent queuing in concourses vs time spent queuing outside the grounds). As a result this will be informed by the work of other teams engaged in the ERP and we are seeking co-ordination and sign-off via DCMS. 

A quantitative framework will be developed, to form the basis for data capture. Simplistically, it will be a series of event day activities, the location in which they take place, a measure of the number of people involved in that activity, and a metric on the resultant conditions/outcome. This metric would be related to key indicators associated with transmission (e.g., density of people in a space; number of interactions within distance thresholds; propensity for wearing face coverings; level of vocal activity etc.). 

NB. The technological means used to collect data would indicate the level of granularity and coverage that can be applied to each of the metrics, however the aim is to develop a ‘checklist’ of all aspects against which we can ensure that there are common threads of data capture that can be applied across all venues to allow for comparability.

Site visits and discussions with venues

The core activity for this scoping phase is to spend time with each of the venue operators – and on site at each of the venues – to identify the potential approach to data collection necessary during the pilot events being held there. We will use the ‘checklist’ derived from the Evaluation Framework as the basis for these discussions, and we will look to understand to what extent data collection against each of these aspects can robustly be provided by:

1. The venue’s existing systems (e.g. turnstiles, ticketing system, CCTV) 

2. Additional systems implemented for the pilot (e.g. wearables, apps, automatic counters)

3. Third Party sources (e.g. Transport Data, Local Authority CCTV, Mobile Network data) 

4. On-the-ground observations (e.g. personnel with handheld equipment, temporary camera installations)

We acknowledge that our work will not be taking place in isolation and that we will be liaising with other members of the research team who also have data collection needs. From a spectator behaviour perspective, there is a key interface with the work being undertaken by the University of Edinburgh and University of Surrey teams related to spectator perceptions – and ensuring that there is the capacity to cross-reference between their survey outputs and this data collection activity. 

This activity will include a site visit, as previous experience indicates that working from plans alone does not allow for a sufficient understanding of the physical arrangements to deliver a robust data collection and observation plan. Note several of the venues are existing MS clients – in these cases we have more familiarity with their existing systems and their capacity to accommodate an overlay for any given event.

Pilot event observation and reporting phase (mid April – mid May)

The precise scope of our activities during the Pilot Event Observation Phase will be determined as an output of the Scoping Phase. However, it is possible to provide an indication of our approach unrelated to data collection and deployment of technology.

Plan observation and event-day data capture

Following the scoping phase, we will establish a working relationship with the venues to ensure sufficient interaction during the planning process, as well as on the event day and during debriefing. This ‘observer’ role will be key in establishing why decisions were made, what information they were based on and the extent to which different guidance influenced them. If needed, this can include supporting the organisers during (or in advance of) key SAG - or equivalent - meetings. 

At this stage, some work will be undertaken to understand how to extract useable data from the technology sources that are already in place and to be deployed. Where necessary we would undertake some initial analytics on the accuracy of those outputs in so far as their ability to provide a robust set of metrics. This may include some testing work using data collected at any historic events that may have taken place with attendees. There will also be an activity to ensure that the event-day deployment of technology is in place – i.e. ensuring existing sources are set up to capture the views/data that are needed and also ensuring that any additional technology is in situ, set up and working properly.

Event-day attendance

We assume that up to three members of our Movement Strategies observation team will be in attendance on the event day at each venue. This includes presence in the Control Room as well as ‘roaming’ observers. As the team would act in accordance with the Public Health requirements, it may be that access will be limited to certain areas – e.g., if there is strict segregation between Hospitality/VIP and GA areas then we would ensure that we remain in alignment with this. If, as plans emerge, it appears that more people will be required to attend to ensure sufficient coverage we would address this as part of the scoping phase. 

We will use the evaluation framework to log qualitative findings on-site.

Our team will be present as observers only and will not engage in any event operations. 

Analysis, evaluation and reporting of observations

We will spend time post event digesting the observations undertaken, analysing the quantitative outputs (given some of the technologies under consideration this task may in some instances be quite involved), and completing the evaluation according to the new framework. For the analysis, we will also investigate the quality/accuracy/validity of any techniques used to automatically capture metrics as part of this study.

We will develop outputs that would cover the findings across all the events, but we will also include a separate assessment covering each individual venue/event. This will allow us to present the overarching messages and recommendations that are of general application, but also provide direct feedback on location/event specific conditions and observations that would be of relevance in specific settings and of value to each of the individual venue operators.

Triangulation analysis of perceptions and behaviour

We will collaborate with the University of Edinburgh and University of Sussex team to triangulate the data on attendee adherence from their survey responses with the observational data on attendee behaviours collected as part of this exercise. We will focus on behaviours associated with transmission (e.g., density of people, number of interactions, wearing of face coverings). It is proposed that this is a collaborative exercise delivered through workshops between the two teams to collate the data efficiently and identify where strong conclusions and correlations can be seen.

Contribution of drafting to the main deliverable

The priority deliverable of the ERP is a report. While we envisage that the main authors will sit within DCMS, we will contribute to the content of the report and ensure that any messages or recommendations emerging from our analysis can be backed up by a clear audit trail within our work.

Deep dive’ analysis and modelling (May onwards)

Following the submission of the priority deliverable, the ERP will continue to drill down further into the data and information collected as part of the Pilot Events to develop a more detailed understanding of the factors which influence risk, and to develop a scalable approach to embedding this into public guidance. 

During this phase, we would continue to support with extended behavioural data analysis, including additional cross-referencing with the work undertaken by all other parties feeding into the programme – providing a more complete and robust scientific base for any conclusions being drawn. We can also offer our support using our crowd modelling tools and capabilities. We have extensive experience in the development and delivery of microsimulation models – a spatial representation of people movement within a given space – based on ‘seeding’ the individuals within the model with different behaviours and/or goals. These use stochastic methods and can therefore be used to provide a quantified (and visual) representation of the impacts of different behaviours captured during the observation phase when applied to venues (or adjacent streets/transport nodes) of different size, scale or layout. Simulation outputs can then be reported against different risk factors (in this context related to public health and disease transmission as opposed to historic applications of crowd safety and fire/smoke attenuation) and fed into the team evaluating the relative level of risk of different event types, with different crowd typologies within different settings. This would all be underpinned by the evidence collected during the pilots, with the opportunity to supplement this with other data emerging once venues re-open or from other global studies of a similar nature. MS has developed and hold existing models for a number of the UK’s most high-profile venues, including Wembley Stadium and the O2 Arena. It may be of interest to use one of these as a case study to identify how these techniques can be deployed before considering a wider roll-out. 

Research questions

We have developed three main research questions in support of the work plan detailed above:

1. How do attendees move and behave at pilot events?

2. What is the typical frequency and duration of person-to-person contact at different event types?

3. To what extent do attendees adhere to behavioural guidance designed to reduce the transmission of COVID-19 during these events?

Key indicators

To address the research questions, as part of the evaluation framework, we plan to collect data on a select set of ‘key indicators’ associated with disease transmission. These will consist of the behavioural, environmental and other measures originally identified in the science framework for opening up group events. The key indicators we plan to collect data on are further detailed below.   

Environmental measures

We will seek to collect qualitative and quantitative data on two measures related to environmental considerations arising from people movement within the venue: 

1. Clustering of people flow (‘bottlenecks’) and dwell time

2. Number of people and seating layout

Behavioural measures

We wish to collect qualitative and quantitative data that can be used to assess a variety of measures related to attendee and staff behaviour in the venues during the pilot events: 

1. Physical distancing (e.g., proportion adhering to 2m/1m+/or other rule, derived population densities, etc.).

2. Physical distance maintained over time for both attendees and staff (e.g., number of interactions over time, interaction duration and cumulative duration of interactions).

3. Volume / area per person in viewing accommodation, circulation areas and other facilities (i.e., crowd densities).

4. Forms of greeting (e.g., types such as handshakes, fist-bumps, hugs, etc., and proportions).

5. Wearing of face coverings (e.g., proportion wearing them, proportion wearing them correctly, types of coverings worn). 

6. Hand hygiene (i.e., use of hand sanitizer or soap).

7. Singing, chanting and shouting, and other active or passive audience behaviours (e.g., dancing, swaying, sitting, leaning, clapping/cheering, grouping, milling etc.)

8. Proportion of the audience standing compared to seated, over time, and during distinct event phases.

9. Use of toilet facilities and concessions, and associated queuing.

10. Sharing of food and drink.

11. Consumption of alcohol.

12. Pre- and post-event activities (i.e., circulation from transport hubs to and from the venue, and behaviours during ingress and egress.)

Data collection requests from venue operators/event organisers

We have compiled list of initial information requests from event organisers and venue operators at the individual pilot events. We need to understand the scope and extent to which we (Movement Strategies) can: 

• Access video footage from the venue’s internal and external CCTV cameras, and the areas covered. 

• Temporarily position our own cameras at pre-defined locations around the venue to supplement data collection.

• Gain access to pre-defined areas within the venue on the day(s) of the pilot event(s). A small team of MS staff will require such access (e.g. 3-4 people).

• Access ticketing data, such as group sizing and seating allocation.

• Access ticketing scans from the venue (e.g., turnstiles or any other existing technology that counts flows or occupancies).  

• View planning documents created or used by the venue or event organisers to plan venue capacities and operations (e.g., SOPs, social distancing guidelines, SG02). 

• Access location data from venue wi-fi networks.

• Temporarily install beacon devices to enable location monitoring and/or deploy wearables.

• Access logs from any building access or crowd monitoring system during the event (e.g., footfall counters, scans of staff or visitor passes, thermal screening, etc). 

• Access/record any current microphones in audience areas that can give a volume level (dB) reading, and whether it is possible to access these records after or during the event. 

• Access concession sales data by location to understand transaction times alongside queues. 

• Access venue cleaning logs by location. 

• Access CCTV footage of the areas surrounding the venues aka ‘Zone Ex’ (e.g., transport hubs, car parks, entertainment venues, walkways to/from venues). Note: large venue operators might have agreements with local stakeholders to share external CCTV footage, so multiple permissions may be required. 

• Temporarily position our own cameras at pre-defined locations surrounding the venue (within ‘Zone Ex’) to supplement data collection. 

• Determine if it is possible to access other third-party data sets and local transport data during or after the event via the event organiser, venue operator or third-parties.  

We will discuss the above information requests with the venue operators and event organisers to establish the best options for data collection on a site-by-site basis.