Home » News

What can tweets about contact tracing apps tell us about attitudes towards data sharing for public health?

Birds eye view of a crowd of people on street

What can tweets about contact tracing apps tell us about attitudes towards data sharing for public health? (Part 1)

The ongoing COVID-19 pandemic means governments have been looking for technological solutions in order to reduce the spread of the virus. Contact-tracing apps are now being used, from Singapore’s ‘TraceTogether’ to ‘StopKorona!’ in North Macedonia. As restrictions on movement are eased in many countries, these apps aim to identify if an individual has been in contact with an infected person through Bluetooth and/or GPS signals. This provides alerts to users and creates early warnings of new outbreaks.  As these apps have been adopted, a huge amount of online discussion has followed about the benefits and concerns around sharing personal data for the benefit of public health. 

So much of this conversation seems novel. Several months ago, most in the UK would have gawked at the possibility of a government app privy to information about who they come into contact with. Yet, the phrase “we are in unprecedented times” has been difficult to escape in recent weeks.  

For me, the onset of the pandemic has coincided with a new research position with the LifeInfo project, under the supervision of Michelle Morris‘ expertise in Health Analytics. This project focuses on people’s attitudes towards sharing their lifestyle data – from supermarket loyalty card to fitness apps – and linking this to health records to drive research into the risk factors of non-communicable diseases such as diabetes, heart disease and certain cancers. Access to these data could have immense benefit as millions of yearly deaths can be attributed to poor diet and physical inactivity.  

Holly Clarke

Leeds Institute for Data Analytics

Holly Clarke is an Intern at Leeds Institute for Data Analytics, applying data science solutions to solve complex, real-world challenges. She is working for the LifeInfo project with Michelle Morris, researching attitudes towards novel lifestyle and health data linkages and how access to this information could improve public health. 

You can follow Holly on Twitter: @HollyEClarke

At the heart of this project is also the vital recognition that we must understand people’s concerns about such initiatives and adapt research accordingly. Part of my role is analysing free-text survey responses about the circumstances under which people would share different types of lifestyle data for health research and factors that might impact their decision to do so.  While the conversation about contact tracing apps and their place in our lives is certainly novel, many of the words and topics about these apps mimic those that come out of my analysis.  

This made me wonder how I could tap into the conversation about contact tracing apps and the insights this could give about data sharing, privacy, surveillance and public health. For the past two weeks I have been scraping tweets about coronavirus apps and will continue to do so as they are developed, trialed, and used in countries around the world.   

This is the first of a series of short blog posts about attitudes towards contact tracing apps and data-sharing for public health. Using text analysis and Natural Language Processing (NLP), I will be answering questions about the conversation around these apps. What topics are prevalent and how do people feel about sharing their data? How does this look in different countries and what role does context play? How does this relate to more general attitudes about data-sharing for public health benefit and what might the impacts be going forward? Twitter is by no means a direct expression of public opinion, but analysing tweets can give us important insights about people’s attitudes, news stories that shape narratives, and shifts in opinion over time.  

So, are people talking about contact tracing apps?  

The first thing to establish is whether people actually care about contact tracing apps. Here, the answer is an undeniable “yes!”. A total of 12,593 tweets were collected on the topic of COVID-19 apps produced during the two-and-a-half-week period between 24 April 2020 and 12 May 2020 (and limiting collection to those in the English language)1. Governments need around 60% of the population (80% of UK Smartphone users) to enable contact tracing apps for them to be effective, which could influence many people to consider their relationship with data-sharing that haven’t given much thought to it before.  

Tweets about coronavirus apps have gone from relatively low numbers (just 203 tweets on 25 April, the first full day of collection) to peaks of over 1300 tweets per day on the 27 April and 5 May. These peaks can be linked to the ‘COVIDSafe’ app release in Australia and the announcement that the NHS ‘track and trace’ app was to be trialed on the Isle of Wight in the UK (see figure 1a).  

Time series graph showing daily counts of tweets about COVID-19.  Shows a peak following the release of the COVIDSafe app in Australia on 26 April and a peak following the UK Government announcement on 4 May re testing of 'track and trace' app on the Isle of Wight.
Figure 1:  Time Series of Covid-19 app tweets showing (a) the number of tweets about Covid-19 apps per day (top) and (b) the number of geo-coded tweets that were produced in different countries highlighting Australia and the UK (bottom), 25 April –11 May 2020.  

Where are people talking about COVID-19 apps? 

Some tweets are geo-located, indicating the country and even city the tweets were produced. Although these tweets make up only a small proportion (2.5%) of the total tweets collected, they act as a sample to indicate where people were tweeting about COVID-19 apps.  

Graph showing locations of Geocoded Tweets about COVID-19 app - UK and Australia have the most tweets.
Figure 2: Locations of geo-located tweets about Covid-19 apps showing (a) the number of tweets collected from each country (top) and (b) their locations on the world map (bottom), 25 April –11 May 2020. 

Most tweets are shown to be produced in the UK and Australia. In these countries contact tracing apps have been nationally introduced and promoted (in the case of Australia) or locally trialed (in the case of the UK). Canada and the US currently constitute only a small proportion of tweet locations; however, this could change in the forthcoming weeks as these countries are yet to announce apps. 

India is the third most popular country for tweet locations where the contact tracing app ‘Aarogya Setu’ has been introduced with associated controversies about personal privacy. Many more tweets about this app have likely been created but in languages other than English, so are not included within the dataset. This is important to consider as insights gained from analysing  tweets will reflect a majority Western perspective. Some of the first countries to introduce contact-tracing apps are non-English speaking (for example South Korea) and additionally have restriction on access to Twitter (in the case of China).  

Over 80% of the geo-located tweets were produced in two countries – the UK and Australia.  Yet, this is not consistent across time.  As shown in figure 1b, during the first week of data collection the conversation was dominated by the Australian context (shown in blue), and this is consistent with the first peak of tweets related to the roll-out of the Australian contact tracing app. Following this, the second week of data collection shows the conversation has shifted towards the UK context (shown in orange) as the NHS app is trialed in the Isle of Wight.  

What do tweets say? 

Next week’s blog post will focus what people are saying about Covid-19 apps, whether attitudes are positive or negative, and if this differs based on the country and context.  The wordcloud below gives an initial insight into the current conversation around these apps.  Two findings stand out. First, context appears to play a large role in shaping the conversation. Words referring to key places and actors (both technological and state) are frequently included in tweets. These include ‘government’, ‘nhs’, ‘apple’, ‘google’, ‘India’, ‘Australia’ and ‘Isle [of] Wight’.  Second, it is striking that the words ‘privacy’ and ‘trust’ are amongst the most frequent words used, showing data management and personal privacy to be at the forefront of discussion.  

Word cloud of most frequent words included with COVID-19 app tweets.
Figure 3: Word cloud of most-frequent words included with Covid-19 app tweets2 

1Search terms included any reference to ‘corona/coronavirus/covid/covid-19 app’ as a single phrase and inclusive of alternative punctuation and spacing  

2Note: common ‘stop words’ are excluded, for example ‘is’ or ‘and’, also the words ‘corona’, ‘covid’ and ‘app’ are excluded as these were the search terms and thus highly frequent. 

CDRC to adopt key role in powerful new COVID-19 data alliance

CDRC to adopt key role in powerful new COVID-19 data alliance

The Consumer Data Research Centre will work through its parent organisation Leeds Institute for Data Analytics to provide a new COVID-19 data alliance with scientific expertise and access to global academic research networks.

Leeds Institute for Data Analytics (LIDA) has worked alongside consortium-leader Rolls-Royce to develop the concept and will take a founding position in a new alliance of data analytics experts challenged with finding new, faster ways of supporting the response to COVID-19 and subsequent global recovery.

Early alliance members are Leeds Institute for Data Analytics, IBMGoogle CloudThe Data CityTruataRolls-Royce and ODI Leeds. The alliance will be facilitated and co-ordinated by innovation specialists, Whitespace.

Together the initial wave of members brings all the key elements of open innovation; data publication, licensing, privacy, security; data analytics capability; and collaborative infrastructure, to kick off its early work and grow its membership.

Emergent will combine traditional economic, business, travel and retail data sets with behaviour and sentiment data, to provide new insights into – and practical applications to support – the global recovery from COVID-19. This work will be done with a sharp focus on privacy and security, using industry best practices for data sharing and robust governance.

As part of LIDA’s involvement in Emergent, researchers will have the opportunity to access these data sets using collaborative platforms which have been established by CDRC.  The academic community will be encouraged to articulate and engage in projects to help understand the changes we are seeing in human activity and social behaviour as a result of COVID-19.

Emergent models will help get people and businesses back to work as soon as possible by identifying lead indicators of economic recovery cycles. Businesses small and large around the world, as well as governments, can use these insights to build the confidence they need to take early decisions, such as investments or policies, that could shorten or limit the recessionary impacts from the pandemic.

The alliance is voluntary and insights will be published for free.

Professor Mark Birkin, who leads both the Consumer Data Research Centre and Leeds Institute for Data Analytics commented:

“Increasing numbers of academics and other commentators are now recognising the potential for commercial organisations to share important data to help in the battle against COVID-19.

An established investment in data sharing capability and analytics capacity makes LIDA ideally placed to lead such conversations.

We are delighted to bring our skills and expertise as a founder member in the Emergent consortium, which offers such enormous potential to deliver benefits to society – and which are so badly needed at this difficult time.”

Connecting business and the academic community

The Consumer Data Research Centre was created in 2014 from a substantial award in the ESRC Big Data Network.  Leeds Institute for Data Analytics at the University of Leeds was then established from the union of the CDRC (Leeds) with the MRC Centre for Medical Bioinformatics.

Since then, both LIDA and the CDRC have been actively promoting the mutual benefit of collaborative projects between corporate partners and the academic community, with researchers working in cross industry teams to undertake scientific research that produces real world insights.

The COVID crisis has further highlighted the importance of these types of collaboration, with governments and their advisers seeking real world insights into mobility, behaviour and human contact networks.

LIDA will be utilising its extensive network – which includes the ESRC Business and Local Government Data Centres, the Alan Turing Institute, Doctoral Training Centres in Data Analytics and Society (ESRC) and Artificial Intelligence (UKRI) – to connect partners with academic experts from multiple institutions and disciplines.

Providing a secure infrastructure

LIDA and IBM will be providing the infrastructure to enable alliance partners to share and compute their data.

Where there is a need to use secure data, partners will be granted access to LIDA’s ISO accredited infrastructure, which will enable them to perform analysis in a safe and controlled environment. Partners using the LIDA infrastructure will be supported by project management and technical support teams from the Consumer Data Research Centre.

For projects using public data, partners will use IBM’s environment and any non-sensitive data will be shared via emergentalliance.org.

Join Emergent

Caroline Gorski, Global Director, R2 Data Labs, the Rolls-Royce data innovation catalyst which started the alliance, said: “We want the global economy to get better as soon as possible so people can get back to work. Our data innovation community can help do this and is at its best when it comes together for the common good.

“People, businesses and governments around the world have changed the way they spend, move, communicate and travel because of COVID-19 and we can use that insight, along with other data, to provide the basis for identifying what new insights and trends may emerge that signify the world’s adjustment to a ‘new normal’ after the pandemic.”

The first challenges have already been issued by the alliance, including one to identify lead indicators of economic recovery which businesses can use to build the confidence they need for investment or activities that will shorten or limit any recessionary impact from the virus.

Emergent hopes to rapidly expand its network of data owners and has set up a website for potential members to register their interest at emergentalliance.org.

CDRC (Leeds) also encourages prospective academic participants to contact us directly at k.r.norman@leeds.ac.uk to receive further updates.

Analysing student eating habits

Analysing student eating habits

Scientists have for the first time used anonymous data from pre-payment food cards to get a unique insight into the eating habits of first year university students.

Data scientists from the Consumer Data Research Centre at the University of Leeds have been able to build a detailed picture of what 835 students ate, and when, by analysing the data linked to their pre-payment food cards.

The cards revealed what they were buying in the campus refectory and associated food outlets.

The analysis gives the most accurate picture to date of first year student diets. Many previous studies have used food diaries, but their accuracy can be variable because they rely on the student remembering exactly – and being honest about – what they have eaten.

Dr Michelle Morris, a University Academic Fellow in Health Data Analytics based at Leeds Institute for Data Analytics, said understanding student diet had public health implications.

Previous studies in the UK and the US have shown that “fresher” students are at risk of weight gain, probably as a result of the lifestyle changes that come with starting university.

In the US, they talk of the “Freshmen 15”, the 15lbs (6.8kg) that students put on. In the UK, research indicates the average student gains 7.7lbs (3.5kg).

The findings, Assessing diet in a university student population: A longitudinal food card transaction data approach, have been published in the British Journal of Nutrition.

The study, which pre-dated the coronavirus outbreak and followed the students aged 18 to 24 over their first semester (12 teaching weeks), revealed student eating habits which clustered around seven dietary behaviours:

  • Vegetarian: with popular purchases being salads, breakfast cereals, yoghurt and fromage frais and a notable absence of meat products
  • Omnivores: which included the most average amounts of all products purchased, with above average amounts of ice cream, desserts and cakes, breakfast cereals and fish.
  • Dieters: with above average purchases of soups, pasta, noodles and salad.
  • Dish of the Day: which included above average purchases of meat and meat products.
  • Grab and Go: which included above average purchases of sandwiches, crisps, nuts and eggs.
  • Carb Lovers: with bread, cheese, egg products and pasta being among the top picks.
  • Snackers: with confectionery, crisps, nuts being above average choices.

Dr Morris, said the dietary patterns were ranked on the basis of “healthfulness”, with vegetarian the most healthful and snackers being the least.

She added: “Our analysis shows that although some students followed one dietary pattern throughout the semester many switched between them.

“Some students moved from a more healthy to a less healthy pattern; for example,  some vegetarians switched to an omnivore diet; and vice versa with some of the students who started off as snackers – the least healthful diet – did move to the Dish of the Day which offered a more balanced range of food options.

“Worryingly perhaps, the most popular move was from a dieter pattern, to the snacking pattern.”

Females were found to be heavily represented among the vegetarians (88%) and dieters (80%) while the men dominated the dish of the day (84%) and grab and go (62%) diet patterns.

This information could be used to target information about healthier eating to students.
Dr Michelle Morris, Leeds Institute for Data Analytics

Dr Morris said the most popular dietary pattern amongst the slightly older students, those aged between 20 and 24, was the omnivore pattern of eating – that could be due to the fact that they may already have lived away from home and settled into a more varied dietary pattern.

She said: “The information from this analysis reveals the pattern of the students’ eating habits, and how that changes over time. That is information that could be used to target information about healthier eating to students.

“Research has shown that adult eating habits take root early in adulthood. So, time spent at University is a great time to encourage healthy eating behaviours that could remain with them for life.”

The research was funded by the Economic and Social Research Council through a Strategic Network for Obesity grant. Maintaining the anonymity of the students was of utmost importance at all stages of the research.

Notes to editor

For further information or interview requests, please contact University of Leeds Media Relations and Communications Officer David Lewis via d.lewis@leeds.ac.uk

Home Working and Horizon Scanning

Home Working and Horizon Scanning

Work has been transformed by the coronavirus crisis with remote working now the norm for millions of workers. But distance from the office is also providing some opportunities to take a wider perspective of the data landscape and to scan business horizons using data sources that we might have overlooked or never investigated in detail.

The CDRC Data Store remains open for business, and our Open and Safeguarded data products are available as normal. Our Secure labs are closed for the duration of the crisis, but we are still accepting Secure data applications for access when things return to normal.

For students, our Masters Dissertation Scheme is still running with a record number of projects for students to complete in the coming months using business and CDRC data. The scheme gives Masters students registered at any UK university a unique opportunity to engage with horizon scanning or other business problems using novel datasets and interesting business perspectives on applied problem-solving. In the past, many participating students have carried out work at the businesses office, but this year students are being offered opportunities to work with businesses through homeworking for the duration of the crisis. The Scheme still brings together the best of academic and business perspectives upon applied problem-solving. Academic supervisors similarly gain the opportunity to collaborate on potentially high impact research with the business community.

So… if you are a Master’s student interested in collaborating with business, but can no longer do this through fieldwork or primary data collection, why not click here to see if any of the CDRC projects interest you? A number of the organisations that we work with are very keen to use part of their homeworking to coach students in the workings of business, especially if you have relevant skills and ways of working to offer!

We also have the CDRC Data Store which has a wide range of data sets available, some of which may be very useful in your dissertation or current research.

Prioritising food establishment inspections

Prioritising food establishment inspections

Populations who frequently eat fast food and live within close proximity of unhygienic food establishments may be at higher risk of contracting foodborne illness than those who do not eat takeaways regularly – but which food establishments are most likely to be unhygienic?

Recent research by CDRC PhD student Rachel Oldroyd uses logistic regression to identify ecological determinants of non-compliant food outlets in England and Wales.  Rachel’s recent paper in Health & Place highlighted:

  • A clear gradient of association is observed between increased deprivation and the probability of non-compliance.
  • Food outlets in the most deprived areas are 25% less likely (OR = 0.75) to meet hygiene standards than those in the least deprived areas.
  • Takeaways, sandwich shops (OR = 0.504) and small convenience retailers (OR = 0.905) are less likely to be compliant than restaurants.
  • Food outlets in large conurbation areas are less likely (OR = 0.678) to meet hygiene standards than those located in cities and towns.
  • Outlets in deprived and urban areas, especially takeaways, sandwich shops and convenience stores should be prioritised for inspection.

You can read the full paper here.

Household Mobility – Where and how far do we move?

Flow Map

In the first article in this series I discussed why it is important for us to understand household mobility, outlined the sources currently available to researchers and highlighted the potential of using commercial data as a possible alternative to census or admin data.

Using large scale commercial data sets, such as the Whenfresh/Zoopla Property Transactions and Associated Migration (available via the CDRC), is exciting as it can answer questions and enrich our understanding of mobility patterns and population change in the UK.

The Consumer Data Research Centre has partnered with online property search provider Zoopla and data insight consultancy Whenfresh to obtain data about the characteristics of properties which have been sold in England and Wales. For the 2014 calendar year, visualised here, there were over 900k unique property transactions.

Where do we move?

This image highlights some larger UK cities and the flow of households between them. When a line becomes more opaque on one side it means that the flow of house movers is directed towards the respective city by a majority of people.

1. Over 68% of moves occurred within the same postcode area (e.g. the postcode area starting LS covers Leeds, EX covers Exeter, etc.) and 34% of moves occurred within the same postcode district (e.g. LS1 covers much of Leeds city centre).
2. Largest number of moves within the city:

  • London – 6057
  • Bristol – 2523
  • Nottingham – 1938
  • Leeds – 1427
  • Sheffield – 1387
  • Manchester – 1342

3. London accounts for 7.1% of all moves in England and Wales (either in to, out of or within the city) and is very well connected to the greater South East region, towns on the south coast and other key cities. London and the greater South East has been termed an ‘Escalator Region[1]’, whereby young people move in to gain skills and training before eventually moving on.

How far do we move?

  1. On the whole, households don’t tend to move very far, the overall median distance of a move in England and Wales is 3.2 miles (we use median distance as opposed to average distances as there are a small number of moves across very large distances which skew the reporting)
  2. People who move within the East of England move further than anywhere else (2.8 miles) whereas the median distance of moves in London is only 1.6 miles. These values are not surprising, given the settlement patterns in the East of England are very different to those in London: the move from Cambridge to Norwich is 64 miles, compared with just 30 miles between Hayes in the far east of London to Dartford in the far west of London.
  3. People moving within the regions of the north of England travel shorter distances than those in the south. The median distance of move within the North East is 1.61 miles, within the North West is 1.73 and within Yorkshire and Humber is 2.01 miles. This is compared to 2.42 miles in the South East and 2.3 miles in the South West.

How connected are our cities?

CityReceived households from % of places in England and WalesSent households to % of places in England and WalesMore households moving into or out of city?
London39%60%Out (-4157)
Bristol23%28%Out (-376)
Norwich20%15%In (+148)
Nottingham18%22%Out (-265)
York17%14%In (+81)
Leeds15%16%Out (-238)
Swindon15%12%In (24)
Manchester15%18%Out (-396)
Birmingham15%21%Out (-560)
Southampton15%19%Out (-169)
Reading14%25%Out (-288)

Norwich appears to be an attractive destination with 20% connectedness and a net gain of 148 households. Nottingham, York, Leeds, Manchester, Birmingham and Southampton are all well connected cities, appearing in the top ten. One surprisingly well connected city for inflows was Swindon, which received households from 172 other places. The surprising entry in terms of outflow connectivity is Reading, which was connected to 25% of other places.

What data are used in these visualisations?

The Consumer Data Research Centre has partnered with online property search provider Zoopla and data insight consultancy Whenfresh to obtain data about the characteristics of properties which have been sold in England and Wales. For the 2014 calendar year, visualised here, there were over 900k unique property transactions. Attached to these transactions data are the Royal Mail redirection service data which provides details about the forwarding postcode for over 212k households who moved. Those records for which there is a forwarding address (and as such an origin-destination link) represent 19% of the 1.1m residential property sales made in 2014[2].  There are a number of unique advantages to using these data:

  • They are timely and easy to update. The data presented are for the calendar year 2014 but other date ranges could be specified.
  • They provide good temporal coverage. Instead of a yearly snapshot of transitions they report the exact date that a property was sold. This allows us to start to understand seasonal trends in mobility.
  • The geography provided is extremely detailed. Sales (origins) are at address level while destinations are delivered as postcodes. This allows us to start to assess mobility patterns at a very small scale, answering questions about communities rather than the more common administrative units used in previous work.

Consumer data like those used here have a role to play in the future of understanding mobility patterns, and population change more generally.

The visualisations were created by Herwig Scherabon[3], an expert in data visualization and information design. Interpreting and visualising flow information is particularly difficult because of the dimensions of the data. Having origins and destinations dictates a need for some way of linking the two together, and the volume of data soon becomes unmanageable. This is why design decisions, like being selective in the number of data entries shown or deciding how opaque crossing lines should be, is important when trying to get a message across.

About the author

Dr Nik Lomax is a University Academic Fellow at the University of Leeds, his research focuses on the way in which demographic behaviour changes over time and how people interact with the areas in which they live and work. Much of his work focuses on the dynamic processes involved in migration but he is also interested in the social implications of changing demographic composition: household formation, social exclusion and population ageing for example. Areas are shaped by changing economic conditions, policy interventions and social attitudes, which in turn has an impact on demographic behaviour. Modelling and explaining these complex interaction is key to the work the work which he does.

[1] Reference to Tony Fielding’s work here

[2] HMRC data

[3] http://scherabon.com/

Demographic profile of pedestrian flows: linking Huq mobile geo-data and SmartStreetSensor footfall data

Birds eye view of a crowd of people on street

Demographic profile of pedestrian flows: linking Huq mobile geo-data and SmartStreetSensor footfall data

Recent research conducted by Consumer Data Research Centre (CDRC) at UCL discussed the potentialities and limitations of football data and mobile geo-data in the context of retail location analysis.

Footfall data derived through SmartStreetSensor project is collected using a network of sensors. This data provides information about the volumes of pedestrian flows, but lacks the contextual insights on the origins of the population.

The geo-data provided by Huq Industries captures around 6% of the average footfall, but maintains a continuous list of locations visited by the users over long periods of time. This information can be used to study the demographics and consumer behaviours of the population.

The preliminary research summarised in this case study sugges linking the contextual information derived from geo-data to footfall counts in order to create a comprehensive understanding of the magnitudes and demographic profile of the pedestrian flows in the retail centres.

Please find the full report here

What’s happened to UK migration since the EU referendum – in four graphs

In this recent Conversation article, CDRC’s Nik Lomax explains what has happened to UK migration since the EU referendum. 

Many of the analyses of why a majority of British voters opted to leave the European Union in a referendum in June 2016, have pointed to a desire to control immigration as a key driving factor. However, surveys since the referendum show fewer people are now concerned about the issue than they were before the poll.

But what has actually happened to immigration in the three years since the UK voted for Brexit?

Decline in net migration

The latest migration estimates published by the Office for National Statistics (ONS) show a steady decline in net migration – the number of immigrants entering the country minus number of emigrants leaving the country – in the three years since the EU referendum in 2016.

The UK saw a net gain of 311,000 migrants in the year to June 2016, which dropped to a net gain of 212,000 migrants in the year to June 2019. This means that while more people are still arriving in the UK than leaving it, the net figure has gone down.

This trend is driven by both sides of the equation. Alongside a decline in the number of people immigrating to the UK – which fell from 652,000 to 609,000 per year in the three year period – the number of people emigrating rose from 341,000 to 397,000. However, the headline figure masks substantial differences between migration from within and outside the EU during this time.

UK long-term international migration UK.
Office for National Statistics – Long-Term International Migration (LTIM), LTIM with preliminary adjustments based on Department for Work and Pensions and Home Office data

Increase from outside the EU

There has been a fall in EU migration since the referendum. In the year ending December 2015 there was a net gain of 218,000 EU citizens. Following a steep decline covering the time of the EU referendum in June 2016 and the immediate aftermath, the figure for the year to June 2019 was 48,000 – its lowest level during the whole of the 16 years covered by the latest ONS data.

EU immigration fell considerably during this time, from 304,000 to 199,000 per year, while emigration of EU citizens increased steadily from 86,000 to 151,000. The net decline can be seen for the EU as a whole, but is most striking for the so-called EU8 group: Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, Slovenia. The UK’s net gain of people from these countries was 80,000 in the year ending December 2015, falling to around zero in the year ending June 2019.

In contrast, net migration from outside the EU has steadily risen over the same time period, from 164,000 to 229,000 in June 2019, continuing a trend which began in 2013. This has been driven primarily by an increase in immigration rather than a drop in people leaving.

Net migration by citizenship.
Office for National Statistics – Long-Term International Migration (LTIM), LTIM with preliminary adjustments based on Department for Work and Pensions and Home Office data

While the UK is unable to put limits on the number of EU citizens arriving under free movement rules while it remains in the EU, it can control migration from outside the EU. Yet, it is this type of migration that has increased consistently.

Uncertainty in employment markets

The International Passenger Survey, one of the data sources upon which the latest ONS publication is based, asks for the reasons that people migrate, with employment and study consistently the most common answers. The latest migration data reveal a decline in EU citizens migrating to the UK for work-related reasons, which include looking for a job or to take up a job offer. Work-related reasons are the most common for EU citizens, and more migrate for a definite job than to look for work.

International Passenger Survey Data, Office for National Statistics

The fall in the number of EU citizens migrating to look for a job is most apparent when comparing the year before the EU referendum against one year after it. The fall in those migrating to take up a definite job is most apparent when comparing the year ending December 2017 with the December 2018 figure. This might reflect uncertainty in the immediate post-referendum period, meaning EU citizens were less prepared to migrate speculatively but still willing to move to take up definite employment. There is consistent evidence that the number of national insurance number registrations (required to work in the UK) for EU nationals has been falling since a peak in 2015.

Attraction of British education

For migrants from outside of the EU there was a similar decline in the number of immigrants looking for work over in the three years since June 2016 (from 24,000 to 8,000), largely driven by a more restrictive migration regime. However, the numbers migrating for a definite job increased from 51,000 in the year to June 2016 to 74,000 in the year to June 2019.

Among migrants from outside the EU the most common reason for migrating to the UK was to undertake formal study – with the number giving this reason up from 113,000 in the year to June 2016 to 157,000 in the year to June 2019. This rise, combined with the rise in those migrating for employment has contributed to the net gain of migrants from outside the EU.

International Passenger Survey Data, Office for National Statistics

Given that international students generally stay in the UK for a defined period of time while studying for a course, there was considerable debate about if students should be included in the governments now abandoned net migration target. However, the debate will continue if similar targets are pursued after the 2019 election.

Where are British citizens going?

Net migration for British citizens remains fairly stable, with a net loss in each of the past 16 years. So where do these British citizens go? This is a surprisingly difficult question to answer comprehensively as the data are not routinely collected, rather estimates are constructed from various sources.

In 2006, the Institute for Public Policy Research, drawing on individual country census and other data sources, reported that around three quarters of all Britons living abroad live in 10 destination countries: Australia, Spain, US, Canada, Ireland, New Zealand, South Africa, France, Germany and Cyprus. An update in 2008 showed that UAE and Switzerland had taken Cyprus’s place at number 10.

Recent research published by demographers Guy Abel and Joe Cohen broadly confirms the top nine destinations using 2010 data, although Italy comes in at tenth spot in their work.

An eye on the future

The latest 2018-based National Population Projection from the ONS take into account trends in migration over the past 25 years. This helps put things in to perspective, as over the longer term, trends tend to fluctuate less than in the short term, where they are influenced by events such as economic conditions or Brexit.

The principal projection has factored in a decline in net migration over the next six years, with a fall to 190,000 annually from 2025. It remains to be seen if the current short-term trend for an overall decline in net migration seen in the latest estimates will continue, or indeed accelerate depending on the outcome of the Brexit negotiations. If it does, then there is a more radical low migration projection variant, which assumes a much lower annual 90,000 net gain by 2025.The Conversation

Nik Lomax, Associate professor in Data Analytics for Population Research, University of Leeds

This article is republished from The Conversation under a Creative Commons license. Read the original article.

We’re hiring: Research Data Scientist/Data Engineer at the CDRC in London

We are seeking someone with a MSc or PhD in computer science/information science or an allied discipline with a strong programming component, with awareness of quantitative methods and social science datasets. The postholder will research improved ways of concatenating and conflating consumer Big Data sources with conventional social surveys and framework datasets. The postholder will research future data acquisition strategies, including development of appropriate preservation and custodianship, consistent with data user priorities.

Further information on the post and person specifications.

What does supermarket loyalty card data reveal about food purchase behaviours?

What does supermarket loyalty card data reveal about food purchase behaviours?

Dr Michelle Morris, UAF LIDA

This week LIDA and CDRC researchers presented two posters at the 13th European Nutrition Conference in Dublin, showcasing some recent work with the large UK retailer: Sainsbury’s.

The first poster presents PhD results from Vicki Jenneson, a student in our Data Analytics and Society Centre for Doctoral training. Results reveal that households in Leeds purchase, on average, 3.5 portions of fruit and vegetables daily. This is higher in affluent and rural areas and with 22% of households purchasing more than 5 portions per day. Conversely in poor, urban areas 18% purchase less than 1 portion per day.

The UK recommendations are to consume 5 portions of fruit and vegetables per day per person. It may be that people get their fruit and veg at school or buy them at a work canteen. However, the transaction data offer a novel and objective measure of fruit and veg purchases.

The work additionally revealed variation in purchasing according to the time of year and the age and gender of loyalty card holders. For the full abstract visit here,or view the poster here.

The second piece of work, from CDRC Research Fellow Stephen Clark, reviews the UK dietary recommendations, the Eatwell Guide, compared with loyalty card purchases for Yorkshire and the Humber. As a proportion of the weight of all purchases, fruit and vegetable purchases are encouragingly close to the recommendations, with 31% purchased compared with 39% recommended. Surprisingly purchases of starchy products, such as bread and pasta, were below the recommended with 17% purchased compared to 37% recommended. Meat and plant based protein products were similar to recommendations and more than twice as many dairy products are purchased compared to recommended. Perhaps unsurprisingly, sweet and savoury snacks like chocolate and crisps exceed recommendations with 17% of purchases by weight on these food, compared to 3% recommended. For the full abstract visit here,or view the poster here.

We are excited to be collaborating with Sainsbury’s on this work and by the potential of these types of transaction data to understand the food purchasing behaviours of our population. We accept there are limitations to these data as they may not capture all food consumed and that individuals may buy from multiple retailers. However, compared to limitations of self-reported data such as recall bias, in addition to the burden of completing a food diary, limiting the scale of data collection, these novel data sources offer great potential in future research and policy making.

The work described here is in the early stages, full academic papers are forthcoming.