Digital Footprint Data Research Articles

A protocol for linking participants’ retailer ‘loyalty card’ records into the Avon Longitudinal Study of Parents and Children (ALSPAC)

Longitudinal population studies (LPS) have a long history of providing insights into how individual consumption patterns and other lifestyle choices affect health and socio-economic outcomes. LPS typically operate as research databanks, integrating rich and deep phenotypic data - covering diverse aspects of individual, family and household status - with genomic data and linked records on health and socio-economic outcomes. However, individual consumption and behavioural choices are traditionally studied solely using self-report methods which are prone to known biases. We propose to enrich LPS databanks with a new form of digital footprint data – individual shopping history records. These are collected by supermarkets through “loyalty” card schemes and can provide a new perspective on real world behaviours and history of consumption. However, as a novel class of data in the context of longitudinal research, our ability to assess the quality and completeness of the data is unknown, as is our ability to effectively triangulate between self-reported and linked data. This paper describes a protocol for linking individual level shopping history data into a LPS using Avon Longitudinal Study of Parents and Children (ALSPAC) as a testbed. The protocol covers the process of establishing participant fair processing, an ethical and legal basis for the linkage framework itself, and how these data will be integrated into the ALSPAC databank. It does not cover the subsequent research use of these data. The protocol was built on an extensive participant engagement and acceptability work and has been approved by the ALSPAC Law and Ethics committee.

Harnessing digital footprint data for population health: a discussion on collaboration, challenges and opportunities in the UK

Digital footprint data are inspiring a new era in population health and well-being research. Linking these novel data with other datasets is critical for future research wishing to use these...

Incorporating Digital Footprints into Credit-Scoring Models through Model Averaging

Digital footprints provide crucial insights into individuals’ behaviors and preferences. Their role in credit scoring is becoming increasingly significant. Therefore, it is crucial to combine digital footprint data with traditional data for personal credit scoring. This paper proposes a novel credit-scoring model. First, lasso-logistic regression is used to select key variables that significantly impact the prediction results. Then, digital footprint variables are categorized based on business understanding, and candidate models are constructed from various combinations of these groups. Finally, the optimal weight is selected by minimizing the Kullback–Leibler loss. Subsequently, the final prediction model is constructed. Empirical analysis validates the advantages and feasibility of the proposed method in variable selection, coefficient estimation, and predictive accuracy. Furthermore, the model-averaging method provides the weights for each candidate model, providing managerial implications to identify beneficial variable combinations for credit scoring.

The possibilities of the Teams information system in creating a digital footprint of first-year students

The article examines in detail the problem of assessing the involvement of first-year students in the educational process and the degree of assimilation of knowledge using digital technologies that make it possible to obtain a student’s digital footprint. The purpose of the article is to present the technology for obtaining digital footprint data when teaching first-year students at Yaroslavl State Technical University in the discipline “Information Systems Architecture” in the MS Teams environment and to show options for performing digital footprint analysis. Methodology. The formation of digital footprint data was carried out in the process of lecture surveys, work on laboratory assignments and passing an exam. To obtain a digital footprint, information systems MS Teams, MS Excel, MS Forms were used. Data obtained in the MS Teams environment was subjected to pre-processing, cleaning, merging tables, and adding (deleting) attributes. The generated tables were analyzed using classification and cluster analysis to search for non-trivial knowledge. Cluster analysis allows us to identify groups of students who successfully complete lecture and exam testing and those who have difficulties completing test tasks. The analysis allows us to identify the level of complexity of tasks and allows for their adjustment. A comparison of the success of completing tasks during the lecture course and in the exam gives an idea of the students’ ability to prepare for participation in testing and the ability to memorize large volumes of complex scientific and technical information.

Hybrid Approach to Predicting Learning Success Based on Digital Educational History for Timely Identification of At-Risk Students

Student retention is a significant challenge for higher education institutions (HEIs). The fact that a considerable number of dropouts from universities are primarily due to academic underperformance motivates universities to develop learning analytics tools based on models for predicting learning success. However, the scalability of such models is limited since students’ academic performance and engagement, as well as the factors influencing them, are largely determined by the educational environment. The article proposes a hybrid approach to forecasting success in completing an academic semester, which involves creating a set of predictive models. Some of the models use historical student data, while others are intended to refine the forecast using current data on student performance and engagement, which are regularly extracted from available sources. Based on this approach, we developed an ensemble of machine learning models and the Markov-process model that simultaneously address the tasks of forecasting success in mastering a course and success in completing a semester. The models utilize digital footprint data, digital educational history, and digital personality portraits of students extracted from the databases of Siberian Federal University, and the resulting ensemble demonstrates a high quality of the forecast. The proposed approach can be utilized by other HEIs as a framework for creating mutually complementary forecasting models based on different types of accessible educational data.

Using social media metrics and linked survey data to understand survey behaviors

Introduction & BackgroundLinking social media and survey data at the individual level has the potential to add evidence to a variety of research questions. To make this data openly available to others, social media data need to be converted into useful metrics that minimise issues of disclosure while maximising utility. This research explores linkages of Twitter data and survey data in the Understanding Society Innovation panel, focusing on the usage of non-disclosive metrics created from Twitter data alongside the similarly anonymised survey data. Objectives & ApproachThe Innovation Panel asked for consent to link Twitter data to survey responses and data has been collected from the Twitter API. However, Twitter’s unstructured nature necessitates creating measures that can be used jointly with linked survey data. We have developed a framework to create social media metrics that can be combined with survey data that also remove any disclosive data, so these data can be widely shared for maximum utility. The current research analyses these data to understand what the metrics look like through presentation of descriptive statistics. We also begin to show these data may be used in combination with survey data through inclusion of a set of metrics in logistic regression models predicting attrition and measurement of mental health. Relevance to Digital FootprintsSocial media is a prevalent aspect of social life and leaves a substantial digital footprint. However, there are a number of limitations to these data, including a lack of understanding of who is producing the data, and having the ability to relate these to a variety of specific (and possibly higher quality) measures for a representative sample of the population. Linkage to surveys address these problems and can lead to new research opportunities using digital footprint data. ResultsWhile small sample sizes impact the power of some analyses, the methods developed are illustrative of ways to use this novel data source. Results show that there is high variation in the created metrics, and initial analysis shows that the inclusion of a set of user-level Twitter data is not significantly related to attrition. However, more accounts followed on Twitter and the number of user retweets are significantly related to higher levels of mental distress on the GHQ scale. Conclusions & ImplicationsOverall, there is some evidence that social media helps to understand survey outcomes, perhaps more so on measurement outcomes. This study provides an initial start on how to use these curated linked social media and survey data, and we note there are other social media networks that we can apply this strategy to; for example, LinkedIn, particularly with changes made to Twitter (X).

Foodinsecurity.london: Developing a food-insecurity prevalence map for London - a machine learning from food-sharing footprints

Introduction & BackgroundThe ability of policymakers to positively transform food environments requires robust empirical evidence that can inform decisions. At present, there is limited data on food-insecurity in the UK that can be used to inform interventions by local authorities, due to the prohibitive costs and logistical challenges of administering longitudinal surveys. This study builds on existing research and a key pilot study developed in partnership between Olio - a food-sharing app with 7 million registered users as of 2023, the University of Nottingham and Havering Council in 2020, which resulted in the world’s first map prototype of food-insecurity. Objectives & ApproachOur approach leverages Machine Learning methods applied to unprecedented food-acquisition behavioural data and open area-level deprivation statistics to model and predict individuals' experience of food-insecurity across London. We used Olio’s extensive network of users to distribute 2,849 surveys, asking respondents across London about their experiences of food-insecurity. The survey was distributed online, adapting the US Department of Agriculture Food Security module. Respondents were asked about their experiences, including (1) eating smaller meals or skipping meals, (2) being hungry but being unable to eat, and (3) not eating for a whole day, because they could not afford food or because they could not get access to food. Using the household, rather than the individual-level version of the food insecurity module helped shed light on the experience of vulnerable groups - such as children. Relevance to Digital FootprintsThe survey responses provided a ground truth about users' experiences of destitution. Deprivation metrics and digital footprint data in the form of food-acquisition behavioural data were then used in a Random Forests Machine Learning model to predict whether households were experiencing food-insecurity, achieving high accuracy. Food-sharing data from almost 50,000 London-based users active on Olio’s platform were then used to identify relevant food-seeking behaviours and aggregate recognised instances of food-insecurity at neighbourhood (MSOA) level. Conclusions & ImplicationsTo identify and rank relevant socio-demographics and food-seeking behaviours most informative for describing food-insecurity an extensive variable selection analysis was performed. The resulting SHAP (SHapley Additive exPlanations) values showed that a combination of food solicitation and the general deprivation of an area were important predictors of food-insecurity.

Development and use of a co-produced short mood survey to collect ground truth in digital footprints research

Introduction & BackgroundTo use digital footprint data for mental health and well-being research we often need to collect concurrent, high-quality measures of ground truth. Delivering frequent surveys to participants using an ecological momentary assessment (EMA) methodology is one way to collect such data. However, existing surveys tend to be long, not focused on momentary states or rely on rating images which are not platform agnostic. Here we present a five-item test-based survey designed with participants and validated for use in EMA studies to collect data about momentary changes in mood. We describe its methodological development and how it has been used to investigate music listening on Spotify as a digital footprint of mood. Objectives & ApproachThe survey is based on the circumplex model of affect. It was co-produced with a participant advisory group (N=5), who gave feedback on the length, content and delivery of the survey. It was then piloted in a group of N=98 participants to assess statistical validity, and congruence with the 20-item Positive and Negative Affect Schedule (PANAS). Following this it was delivered in a wider sample (N=150) four times a day over a two-week period using an EMA app on participant’s phones. Relevance to Digital FootprintsEMA is an increasingly popular method for collecting ground truth to support the interpretation of digital footprint data. This newly developed and tested mood survey offers an opportunity to reduce participant burden for collecting mood data in EMA studies which will support the collection of high quality and high time-resolution ground truth for digital footprints research. ResultsTogether with participants we selected four emotions across the axes of arousal and valence, as well as rumination which participants considered important in their music listening behaviors. Factor analysis of pilot data showed that the questions represented two factors of positive and negative affect. The ratings on a 0-10 scale of the emotions ‘cheerful’ and ‘relaxed’ explained 44% of the variance in positive affect, and ratings of ‘worried’, ‘sad’ and ‘frustrated’ explained 40% of the variance in negative affect. Delivery of the questionnaire in a wider student sample (N=150) four times per day for two weeks allowed for the opportunity to assess typical response rates in a realistic EMA setting. On average participants completed 3 out of the 4 surveys a day. Conclusions & ImplicationsThe co-created, short mood survey for the collection of ground truth in digital footprint studies was validated across two independent samples, and shown to allow for good response rates in a two week study. Future testing on wider samples will provide opportunities to validate the survey and assess its effectiveness across demographic groups and different sample types.

The dynamics of emotion expression on Twitter and mental health in a UK longitudinal study

Introduction & BackgroundAn estimated 4.95 billion people used social media in 2023, with the average user active on around seven platforms for over two hours per day. This widespread use leads to abundant digital footprint data around interactions with social media. These data can be collected continuously and reflect real behaviour of users in naturalistic settings. These strengths have led researchers to propose the use of social media data in digital phenotyping, where digital footprints can be used to quantify and predict health conditions. Mental health assessment in particular could benefit, as existing approaches, such as self-report questionnaires and inpatient assessment, are unable to perform the real-time monitoring that digital phenotyping could potentially achieve. Digital phenotyping models for mental health require careful consideration of what aspects of social media data to include. Including all data users generate could result in models that are overfitted and difficult to explain. Studies are required that explore the relationship between specific aspects of social media data, such as the time course of expressed emotion, and gold-standard measures of mental health. Objectives & ApproachWith participants’ consent, we linked Twitter data to self-reported measures of mental health from the Avon Longitudinal Study of Parents and Children. We performed sentiment analysis using three different approaches—LIWC, VADER and RoBERTa—to estimate the amount, variability and instability of positive and negative emotional content in each participant’s Tweets over a one-year period. We explored the association between these measures of emotion expression and self-reported scores of depressive symptoms, anxiety symptoms and wellbeing. These mental health measures are the Short Mood and Feelings Questionnaire, the Generalized Anxiety 7 and the Warwick Edinburgh Mental Wellbeing Scale. Relevance to Digital FootprintsOur research is highly relevant to digital footprint research, as it involves the use of digital footprint data (i.e. Twitter data) to predict mental health outcomes. Conclusions & ImplicationsThe results of our analysis will inform the development of digital footprint based phenotyping for mental health that could one day provide information to supplement clinical assessments.

Estimating hyperlocal traffic CO2 by customizing spatial relationship: An analysis from Digital Footprint data

Estimating hyperlocal traffic CO₂ by customizing spatial relationship: An analysis from Digital Footprint data

Exploring the destination network in the context of tourism mobility: a multi-scale analytical framework

ABSTRACT Tourist mobility has attracted considerable attention from tourism academics and industry practitioners. This study explores the destination network, which is constructed based on tourists’ digital footprint data collected from online travel diaries, by proposing a multi-scale analytical framework. In particular, to analyze the tourism destination network comprehensively, the proposed framework integrates traditional quantitative analysis (gravity center model, statistics of flow directions) and three scales of network analysis (macroscopic, mesoscopic and microscopic network metrics), to reveal the structural patterns of the underlying pattern characteristics; furthermore, by using the multiple regression quadratic assignment procedure (MRQAP), the framework also explores the formation mechanism of the destination network by measuring the correlation and causality between the destination network and the corresponding transportation network, combining other potential driving factors. Based on illustrative case study data, which contains 18, 939 travel diaries and 51, 603, 044 transportation records from Chengdu-Chongqing economic circle in China, the empirical results show that the destination network is dynamically changed during four-year period, and the transportation network has great potential in explaining the formation of destination network. The dynamic characteristics of destination network and the exploration of formation mechanism offer the opportunities for improving destination planning and management.

Внедрение адаптивного обучения в университете: кейс УрФУ реализации дисциплины «Иностранный язык»

Digitalization of learning environment and learning individualization are among the key global trends of education transformation. Digital tools allow designing individual learning path for each student, and improving learning efficiency by considering entrance levels of skills and knowledge, personal specifics of information perception, and speed of acquiring new knowledge. Learning analytics tools are used to analyze student digital footprint data in order to not only forecast student success of failure at the following stages of the course, but to design the most optimal path to desired learning results through the use of adaptive learning tools. The current research was aimed at developing and testing an adaptive learning method for teaching of a University foreign language course. The adaptive learning method uses analysis tools to form unique individual learning trajectory for each student. Comparative analysis of different learning models was performed using data for 5,154 students. Conclusions were made on the advantages of adaptive learning and mixed learning model with active teacher’s role in student success monitoring. Successful experience of testing the adaptive learning method for teaching foreign language demonstrates practical value of research results, and allows their further use for implementing adaptive learning in higher education institutions.

Can We Measure the Structural Dimension of Social Capital with Digital Footprint Data?—An Assessment of the Convergent Validity of an Indicator Extracted from Digital Footprint Data

Network analytical metrics often seek to capture the structural dimension of social capital, but such data collections using traditional social research tools often suffer from biases like interviewer effects and are usually only suitable to study small groups of participants. Digital sources of social relations might offer great potential for facilitating such measures though, because they digitally store unprecedented amounts of relational data, free from the limitations associated with self-reported data. This study therefore compares individual node degrees collected through a contact diary (i.e., overall-social capital), and a counterpart extracted from digital footprint data from the social media network, Facebook (i.e., social media network-social capital). The findings suggest that researchers conducting empirical studies involving the concept thus should not ignore social media network-social capital as a practical alternative measure of overall-social capital; it provides a sound approximation but only after controlling for other influential factors. In particular, our results highlight that the usability of the digital social capital metric is conditional on the three-way interaction between the variables gender, age, and social media network-social capital. Thus, the evidence from our study, in turn, also intimates that individuals act heterogeneously in the digital sphere with respect to their networking behaviour.

Exploring income and racial inequality in preparedness for Hurricane Ida (2021): insights from digital footprint data

Abstract Preparedness for adverse events is critical to building urban resilience to climate-related risks. While most extant studies investigate preparedness patterns based on survey data, this study explores the potential of big digital footprint data (i.e. population visits to points of interest (POI)) to investigate preparedness patterns in the real case of Hurricane Ida (2021). We further investigate income and racial inequality in preparedness by combining the digital footprint data with demographic and socioeconomic data. A clear pattern of preparedness was seen in Louisiana with aggregated visits to grocery stores, gasoline stations, and construction supply dealers increasing by nearly 9%, 12%, and 10% respectively, representing three types of preparedness: survival, mobility planning, and hazard mitigation. Preparedness for Hurricane Ida was not seen in New York and New Jersey states. Inequality analyses for Louisiana across census block groups (CBGs) demonstrate that CBGs with higher income have more (nearly 8% greater) preparedness in visiting gasoline stations, while CBGs with a larger percentage of the white population have more preparedness in visiting grocery stores (nearly 12% more) in the lowest income groups. The results indicate that income and racial inequality differ across different preparedness in terms of visiting different POIs.

Digital Competence Profiles in Education

Purpose of research. The trend of personalization of learning forms a demand for the development and implementation of software tools for processing the digital footprint of the learner, to manage his/her educational trajectory. In combination with decentralized identification technologies, such software tools assume a more active role of the learner in the creation of his/her personal learning trajectory. Additional challenges for the developed software tools are the functional requirement to ensure the independence of learner profile data from individual educational services, while at the same time complying with the standards of information security of working with data. The purpose of this research is to expand the understanding of the possibility of modifying online learning by including distributed solutions in the field of information technology application in education.Materials and methods. The paper describes in detail the solutions developed by the authors for software implementation of a digital profile of learner's competencies - a decentralized application designed to store data of learning activity and confirmed qualifications aggregated from various external educational platforms and services. When forming these solutions, the authors proceeded from the advantages of using a decentralized approach to software development. The methodological basis of the research is represented by a set of approaches and technologies that support the logic of the software development process - distributed registry technology, digital identification approach, credentials storage standard.Results. The authors substantiate the choice of solution for software implementation of digital competence profile: SSI approach will provide decentralized identification of digital footprint data beyond the boundaries of individual educational platforms and services; application of the standard of verifiable VC credentials will create conditions for decentralized storage of learning activity data. The results of the study may be of practical interest for the administration of domestic educational organizations, administration of educational platforms and educational services implementing online learning programs for effective and secure management of each individual student's digital footprint data, creating an individual educational trajectory in his/her interests and confirming competencies.Conclusion. Implementation and scaling of the digital competence profile are aimed at ensuring the identification, safe storage, and effective management of the learner's digital footprint data, which can be used to create an individual learning trajectory. The digital competence profile will allow to realize the function of confirming any forms of activity, the nature of interaction of the learner with the functions of the educational service, as well as allow the learner to control and manage the use of the information placed in the digital profile. As part of the whole, the digital profile is considered in the overall architecture of the information system, which implies the interaction of the software application with other subsystems, including in the structure of the educational platform. The scheme of interaction demonstrated for the digital educational platform “Digital Tutor” developed at the Plekhanov Russian University of Economics. Additional research requires tools for controlling the activity and motivation of the learner to formalize his/her active role in the issues of personal learning trajectory management.

Identifying drivers of food insecurity through linked data- the Priority Places for Food Index

Introduction & Background15.5% of all UK households are food insecure; either unable to afford to eat, skipping meals or reducing meal sizes despite being hungry. Drivers of food insecurity include both access to and affordability of food, with those most in need often left unable to access healthy and affordable food. Taking a place-based approach to understand the drivers of food insecurity allows for targeted support from government, third sector and private organisations to mitigate growing food insecurity in the UK.
 Objectives & ApproachThis research presents the methodology for the co-production and construction of the Priority Places for Food Index (PPFI) and supporting dashboard, co-designed by the Consumer Data Research Centre (CDRC) and consumer champions Which? in response to the 'cost-of-living crisis'.
 The PPFI equally weights measures of access to affordable food and indicators of barriers to affording food across seven domains; Proximity to supermarket retail facilities, Accessibility of supermarket retail facilities, Access to online deliveries, Proximity to non-supermarket food provision, Socio-economic barriers, Fuel Poverty and Family Food for support. The PPFI uses open data combing traditional census data metrics, with government data (e.g., Healthy start voucher and free-school meals uptake), digital footprints data (web-scraped delivery addresses and food bank item request data) and scaled-survey data (fuel poverty, propensity to shop online).
 Relevance to Digital FootprintsDigital footprint data can complement traditional data sources to provide a more nuanced view of health inequalities. These data are typically less timely to collect than traditional data collection methods (census, survey) allowing a more reactive response to emergent issues such as the cost-of-living crisis.
 ResultsThe PPFI interactive map and underlying data have been published via the CDRC https://priorityplaces.cdrc.ac.uk/.
 Conclusions & ImplicationsWe demonstrate the value of data linkage across individual and population level data to provide localised insight into food insecurity and identify where digital footprints data can improve gaps in the current evidence base. We also reflect on the value of co-production and stakeholder engagement in creating a policy ready interactive map which has facilitated the lobbying of targeted practical support and policy change to address food insecurity.

Predicting health related deprivation using loyalty card digital footprints

Predicting health related deprivation using loyalty card digital footprints

Carbon foot printing school meals: data linkage and engagement activity

Introduction & BackgroundFood production is a substantial contributor to greenhouse gas emissions and climate change. A more sustainable diet is often a healthier one, so making lower carbon food choices serves to benefit the planet and person. In order to understand the carbon footprint of food choices, linkage of recipe information to carbon footprint data and transaction records is required. To inform positive change insights from data linkage must be communicated to the target audience, in this case schoolchildren.
 Objectives & ApproachSchool dinner recipe information and transaction records for school meals at five schools for a six week period were acquired. Carbon footprint estimates were calculated for each recipe, using published data. An automated dashboard was created in order that these calculations could be replicated by catering teams. Carbon footprints were appended to the school transaction records for meal choices. An interactive web game was created in ‘top trump’s' style using a selection of the recipes, with carbon footprint and popularity ranking, generated from the transaction records.
 Relevance to Digital FootprintsTransactional meal sales data from schools are digital footprint data. In this work we link these digital footprint data to detailed recipe information with estimated carbon footprints from an open data source.
 ResultsThe Consumer Data Research Centre Carbon Calculator and The Planet Plates game were created. The Carbon Calculator is being used in a number of settings to support food procurement and recipe development. The Planet Plates game has been used in Leeds Schools to empower schoolchildren to make positive changes to lower the carbon footprint of their meal choices. The children were engaged with all the activities and not only learned about sustainability of their food choices, but about how data they generate can be used anonymously for public good.
 Conclusions & ImplicationsData linkage of digital footprint data is a powerful tool for behaviour change to tackle some of the world’s most pressing challenges. Methods and insights should be shared widely and made accessible to a range of stakeholders wherever possible.

Education for out-of-school children: Unpacking driving factors of vulnerability via digital-learning footprint data in East Africa

Introduction & BackgroundApproximately 617 million children and adolescents worldwide do not possess the foundational skills to live healthy and productive lives. Sub-Saharan Africa is profoundly affected, resulting in social and financial dependency and raising vulnerability to forced marriage, female genital mutilation, and mental health issues. Contextual factors are considered critical in this crisis, yet have received little attention due to a currently insurmountable deficit in traditional census and survey data. Digital footprint data offers a potential route to filling this information gap - particularly in education- and how a learner’s environment can impact digital learning and future well-being.
 To address the crisis, the ‘Global Learning’ XPRIZE competition challenged teams to develop software empowering marginalised out-of-school children to learn literacy and numeracy skills. Five finalist teams tested their technology with 2041 children using handheld tablets in 172 remote villages in Tanzania.
 Objectives & ApproachOur study examined factors that can predict improvements in learning outcomes, building on the digital footprint data collected from the children participating in the device intervention in the form of app usage and locational and activity data. Additional geospatial features were engineered based on village coordinates, distance to local amenities, services and transport, variables serving as additional potential indicators of isolation and connectedness. These data were linked with child-level factors, including household composition and literacy levels.
 After comparative assessment of machine learning regression models, tree-based models (XGB, RF) were used to establish the optimal predictive performance for literacy and numeracy. Variable importance using SHAP was used to determine which specific contextual variables should be considered before deploying digital interventions to support education and well-being.
 Relevance to Digital FootprintsUtilising digital footprint data to quantify the influence of geospatial and contextual data in digital interventions can offer comprehensive insights to understand and address factors impacting learning outcomes in this context.
 ResultsPrior school attendance, home reading environments and high familial literacy were found to be predictors of higher learning outcomes after a technology-based learning intervention. Environments featuring an unemployed caregiver and few siblings were surprisingly consistent positive predictors, suggesting accompanying and focussed caregiver support as valuable for effective development via digital interventions. Proximity to police stations and health centres were revealed as key predictors, indicating the importance of social and physical connectedness in positive learning environments.
 Conclusions & ImplicationsTargeted improvement of EdTech provision with out-of-school learners promises to help the development of learning outcomes in remote villages and reduce vulnerabilities to correlated social and health issues. However, positive outcomes can only be achieved when appropriate supporting environments surround interventions.

Investigating the Relationship between Timing of Tweets and Mental Health, Well-being and Sleep Quality in a UK birth cohort

Introduction & BackgroundSocial media use has been proposed as a cause of worsening mental health and wellbeing over the last decade, but its role in mitigating some of the effects of social distancing during the pandemic showed that it also has the potential to improve these outcomes. Whilst existing research disagrees on the degree to which social media use harms or helps, there is growing consensus around the need to move from global measures of social media use to specific measures of types of social media use. These new measures can enable an exploration of proposed mechanisms and causal pathways linking social media use and mental health and wellbeing. A commonly proposed mechanism is nighttime social media use reducing sleep quality, and consequently harming mental health and wellbeing.
 Objectives & ApproachWe aimed to investigate the relationships between the time Twitter users post content and their mental health, wellbeing and sleep quality using direct measurements of Twitter use linked to standardised mental health measures in a well-characterized cohort.
 This study uses approximately 1.5 million Tweets harvested between January 2008 and March 2023 from 622 participants in the Avon Longitudinal Study of Parents and Children (ALSPAC). These Tweets have been linked to questionnaire data collected on six occasions spanning April 2019 to May 2021. These questionnaires included standard measures of depressive symptoms, anxiety symptoms, mental wellbeing and difficulty sleeping.We have taken two approaches to explore these relationships, using circular statistical methods novel to social media data analysis to account for day/night cycles. The first approach used mixed effect models to investigate the association between the time a Tweet was posted and the mental health, mental wellbeing and sleep quality of the poster. The second approach explored the relationships between the mean hour participants post Tweets in a given time period, and their mental health, mental wellbeing and sleep quality.
 Relevance to Digital FootprintsThis research is highly relevant to Digital Footprints, due to its use of data directly extracted from a social media site. The methodologies employed in analysing this alongside more traditional epidemiological survey data provides an example of how digital footprint data can complemented by high quality ground truths.
 ResultsThere was evidence that the timing of Twitter activity was predictive of the mental wellbeing and sleep quality of participants, even after adjustment for demographic, educational and socio-economic covariates. However, the hour a Tweet was posted at explained very little of the variation in the mental wellbeing or sleep quality of the participant who posted it (0.1% and less than 0.1% respectively). There was weak to no evidence that the timing of Twitter activity was predictive of the depressive and anxiety symptoms of participants.
 Conclusions & ImplicationsWhilst this study found evidence that the hour participants post on Twitter is predictive of their mental wellbeing and sleep quality, the amount of variation explained by these models suggests that this is not a clinically relevant risk factor. This study supports arguments in the literature that the use of social media has a very small and insignificant effect on mental health, wellbeing and sleep quality.