Synthetic data generation is an important tool for privacy-preserving data sharing. While diffusion models have set recent benchmarks, flow matching (FM) offers a promising alternative. This paper presents different ways to implement flow matching for tabular data synthesis. We provide a comprehensive empirical study that compares flow matching (FM and variational FM) with a state-of-the-art diffusion method (TabDDPM and TabSyn) in tabular data synthesis. We evaluate both the standard Optimal Transport (OT) and the Variance Preserving (VP) probability paths, and also compare deterministic and stochastic samplers – something possible when learning to generate using variational flow matching – characterising the empirical relationship between data utility and privacy risk. Our key findings reveal that flow matching, particularly TabbyFlow, outperforms diffusion baselines. Flow matching methods also achieves better performance with remarkably low function evaluations (≤ 100 steps), offering a substantial computational advantage. The choice of probability path is also crucial, as using the OT path demonstrates superior performance, while VP has potential for producing synthetic data with lower disclosure risk. Lastly, our results show that making flows stochastic not only preserves marginal distributions but, in some instances, enables the generation of high utility synthetic data with reduced disclosure risk.

Generative Adversarial Networks (GAN) have been used in many studies to synthesise mixed tabular data. Conditional tabular GAN (CTGAN) have been the most popular variant but struggle to effectively navigate the risk-utility trade-off. Bayesian GAN have received less attention for tabular data, but have been explored with unstructured data such as images and text. The most used technique employed in Bayesian GAN is Markov Chain Monte Carlo (MCMC), but it is computationally intensive, particularly in terms of weight storage. In this paper, we introduce Gaussian Approximation of CTGAN (GACTGAN), an integration of the Bayesian posterior approximation technique using Stochastic Weight Averaging-Gaussian (SWAG) within the CTGAN generator to synthesise tabular data, reducing computational overhead after the training phase. We demonstrate that GACTGAN yields better synthetic data compared to CTGAN, achieving better preservation of tabular structure and inferential statistics with less privacy risk. These results highlight GACTGAN as a simpler, effective implementation of Bayesian tabular synthesis.

The transition to first parenthood is a critical juncture at which both paid and unpaid work become more gendered within couples. From a life course perspective, parenthood is not a discrete event but a process that couples traverse together across successive phases. Far less is known, however, about how the gendered division of labour and its implications for the within-couple pay gap evolve across subsequent phases of parenthood. Using longitudinal data from UK dual-earner couples (1996–2023), this study examines gender dynamics across four parenthood phases using couple-level fixed-effects and Karlson–Holm–Breen (KHB) mediation analysis. First, we find that the initial shock of parenthood is not temporary. Gendered divisions become entrenched in the established phase and re-intensify with subsequent births, while male partners’ labour contribution remains virtually unchanged. Second, the KHB decomposition reveals that the asymmetric division of labour significantly accounts for shifts in female partners’ relative hourly pay across phases: The female partner’s share of paid work operates as a suppressor that partially offsets negative direct effects, whereas her share of housework further reinforces relative pay disadvantage. This study contributes by (1) providing a multi-phase, couple-level account of how gendered arrangements unfold as families move through parenthood, bearing on ongoing debates between resource-based and gender-based explanations of couples’ division of labour, and (2) testing the mediation pathway linking parenthood phases to within-couple pay inequality and identifying when across the parenthood trajectory the reallocation of labour most strongly accounts for this inequality.

Clinical trial data including information about the health and side effects experienced by participants has typically been held securely behind the walls of pharmaceutical companies. It is only in the last decade that the concept of anonymisation for data sharing has evolved within the industry. To characterise the obstacles that need to be overcome in this area, this study reports on professionals' views and experiences of anonymisation for data sharing in the pharmaceutical industry. Semi-structured qualitative interviews were carried out with professionals working in technical anonymisation and/or data sharing management roles at major pharmaceutical companies, as well as with a contract research organisation specialising in data anonymisation for the pharmaceutical industry. Thematic analysis of the interviews to define processes and characterise barriers and facilitators relevant to anonymisation for data sharing was conducted. Fourteen participants were interviewed from seven companies. Results identified two themes relating to challenges:(1) developing a standardised methodology and(2) limited resources and data science capabilities. Results also identified benefits of data sharing policies and opportunities for improvement of automated processes. This study demonstrates challenges and opportunities for the development of standardised and automated approaches to quantitative risk assessment that are tailored to CSRs.

There is growing interest in synthetic data generation as a means of allowing access to useful data whilst preserving confidentiality. In particular, synthetic microdata generation could allow increased access to census and administrative data. An accurate understanding of the comparative performance of current synthetic data generators, in terms of the resulting data utility and disclosure risk for synthetic microdata, is important in allowing data owners to make informed decisions about the choice of method and parameter settings to use. Synthesizing microdata can present challenges as the data typically contains predominantly categorical variables that standard statistical methods may struggle to process. In this paper we present the first in-depth evaluation of four state-of-the-art synthetic data generators originating from the statistical (synthpop, DataSynthesizer) and deep learning (CTGAN, TVAE) communities and each capable of dealing with microdata. We use four real census microdatasets (Canada, Fiji, Rwanda, UK) to systematically validate and compare the synthetic data generators and their parameter settings in terms of the utility and disclosure risk of the resulting synthetic data using statistical metrics and the risk-utility map for visualization. Our analysis shows that the performance of the synthetic data generators considered depends on their parameter settings and the dataset.

Synthetic data generation is a promising approach for sharing data for secondary purposes in sensitive sectors. However, to meet ethical standards and legislative requirements, it is necessary to demonstrate that the privacy of the individuals upon which the synthetic records are based is adequately protected. Through an expert consensus process, we developed a framework for privacy evaluation in synthetic data. The most commonly used metrics measure similarity between real and synthetic data and are assumed to capture identity disclosure. Our findings indicate that they lack precise interpretation and should be avoided. There was consensus on the importance of membership and attribute disclosure, both of which involve inferring personal information. The framework provides recommendations to effectively measure these types of disclosures, which also apply to differentially private synthetic data if the privacy budget is not close to zero. We further present future research opportunities to support widespread adoption of synthetic data.

Paternal childcare engagement is a focus of work–family policy debates yet there is little consensus about what engagement means and how it might be measured. Drawing on Lamb’s (1986) classification of paternal involvement, we run confirmatory factor analysis on a sample of two-parent households from the UK’s Millennium Cohort Study to derive latent paternal engagement measures at nine months, three, five, seven and eleven years old. Structural Equation Modelling is used to explore the relationship between the engagement measures and parents’ employment hours. Employment hours have a significant association with paternal childcare engagement in the early stages of a child’s life, but it is paternal engagement in the previous time period that has a far stronger effect at every age. Specifically, paternal engagement in the first year of parenthood is important for fostering ongoing engagement until the child is at least age eleven, and this positive effect builds over time.

Synthetic data has a key role to play in data sharing by statistical agencies and other generators of statistical data products. Generative Adversarial Networks (GANs), typically applied to image synthesis, are also a promising method for tabular data synthesis. However, there are unique challenges in tabular data compared to images, eg tabular data may contain both continuous and discrete variables and conditional sampling, and, critically, the data should possess high utility and low disclosure risk (the risk of re-identifying a population unit or learning something new about them), providing an opportunity for multi-objective (MO) optimization. Inspired by MO GANs for images, this paper proposes a smart MO evolutionary conditional tabular GAN (SMOE-CTGAN). This approach models conditional synthetic data by applying conditional vectors in training, and uses concepts from MO optimisation to balance disclosure risk against utility. Our results indicate that SMOE-CTGAN is able to discover synthetic datasets with different risk and utility levels for multiple national census datasets. We also find a sweet spot in the early stage of training where a competitive utility and extremely low risk are achieved, by using an Improvement Score. The full code can be downloaded from github1.

Teaching datasets are a pivotal component of the data discovery pipeline. These datasets often serve as the initial point of interaction for data users, allowing them to explore the contents of a dataset and assess its relevance to their needs. However, there are instances where their viability is limited, particularly where source data is only accessible within restricted settings, such as trusted research environments (TREs). In response to this challenge, this paper proposes the production of synthetic datasets tailored for specific teaching purposes by utilising already cleared (and published) analyses as the basis for the synthesis. Unlike generic synthetic datasets, the datasets created are designed to solely reproduce the specific analyses. Crucially, the datasets can be generated without access to the original data. Two experiments with census data demonstrate the viability of the method and a live use case is described. Issues arising such as marginal disclosure risk are then discussed.

This study investigates the effect of personal income on attitudes towards climate change risk, considering the mediator of responsibility attribution (RA) for climate change and the moderator of educational attainment. Using a latent growth curve model applied to data from the UK Household Longitudinal Study dataset from 2009 to 2020, this study indicated that personal income growth heightens the likelihood of expressing concern about climate change while reducing the propensity for holding sceptical or paradoxical attitudes over time. Attributing climate change to personal behaviour mediated the income-attitude relationship. Significant differences in the mediation effect were observed between individuals who had received a university education and those who had not, even after controlling for covariates such as age, sex, political affiliation and employment status. These findings suggest that climate campaign emphasising RA can address attitude disparities across income and education strata, further promoting sustainable climate action and mitigation.

Introduction The World Health Organisation declared a global pandemic in March 2020. The impact of COVID-19 has not been felt equally by all regions and sections of society. The extent to which socio-demographic and deprivation factors have adversely impacted on outcomes is of concern to those looking to `level-up' and decrease widening health inequalities. Objectives In this paper we investigate the impact of deprivation on the outcomes for hospitalised COVID-19 patients in Greater Manchester during the first wave of the pandemic in the UK (30/12/19-2/1/21), controlling for proven risk factors from elsewhere in the literature. Methods We fitted Negative Binomial and logistic regression models to NHS administrative data to investigate death from COVID in hospital and length of stay for surviving patients in a sample of adult patients admitted within Greater Manchester (N = 10,372, spell admission start dates from 30/12/2019 to 02/01/2021 inclusive). Results Deprivation was associated with death risk for hospitalised patients but not with length of stay. Male sex, co-morbidities and older age was associated with higher death risk. Male sex and co-morbidities were associated with increased length of stay. Black and other ethnicities stayed longer in hospital than White and Asian patients. Period effects were detected in both models with death risk reducing over time, but the length of stay increasing. Conclusion Deprivation is important for death risk; however, the picture is complex, and the results of this analysis suggest that the reported COVID related mortality and deprivation linked reductions in life expectancy, may have occurred in the community, rather than in acute settings.

Open and practical exchange, dissemination, and reuse of specimens and data have become a fundamental requirement for life sciences research. The quality of the data obtained and thus the findings and knowledge derived is thus significantly influenced by the quality of the samples, the experimental methods, and the data analysis. Therefore, a comprehensive and precise documentation of the pre-analytical conditions, the analytical procedures, and the data processing are essential to be able to assess the validity of the research results. With the increasing importance of the exchange, reuse, and sharing of data and samples, procedures are required that enable cross-organizational documentation, traceability, and non-repudiation. At present, this information on the provenance of samples and data is mostly either sparse, incomplete, or incoherent. Since there is no uniform framework, this information is usually only provided within the organization and not interoperably. At the same time, the collection and sharing of biological and environmental specimens increasingly require definition and documentation of benefit sharing and compliance to regulatory requirements rather than consideration of pure scientific needs. In this publication, we present an ongoing standardization effort to provide trustworthy machine-actionable documentation of the data lineage and specimens. We would like to invite experts from the biotechnology and biomedical fields to further contribute to the standard.

Systematic observation is a promising unobtrusive method of assessing human behavior in urban environments without many issues typically associated with self-report measures (e.g., recall bias, low response rates). Improvements in video camera technologies make it more feasible for researchers to conduct systematic observation, which could reduce the time, labor, and cost to facilitate high-quality observational research in urban environments at scale. However, there are important ethical and information governance challenges driven by data protection laws, which discourage many researchers from using camera-based observation methods. The European Union General Data Protection Regulation is a leading global standard for data protection. Drawing on our experiences of conducting three studies using video cameras in public spaces, we discuss how to conduct this kind of research in line with General Data Protection Regulation requirements. The paper outlines issues concerning data protection, privacy, informed consent, and confidentiality, and how we addressed them. In doing this, the paper provides support for responsible use of camera-based observation methods, which will be of value to researchers, ethics committees, and funders. Outlining how to use video cameras responsibly will enable more research to be conducted that, in turn, will build the case for its benefits to researchers and society.

The effectiveness of approaches to bot detection varies, with real-time detection being almost impossible. As a result, this article argues that the general Twitter using public cannot be expected to judge which accounts are bots with certainty and therefore do not know to what extent they are being manipulated online. In this article, the challenge of detecting bots and fake accounts is demonstrated by constructing two distinct methods to bot detection. The first method takes a fixed criteria-based approach, by building on commonly cited identifiers for bots. The second method takes a more flexible, investigative approach in order to uncover bots involved in coordinated efforts to influence online debates. As well as profiling the specific mechanics of how each one operates, we argue that they can be compared against an evaluative framework that specifies a set of key criteria that bot detection methods should meet in order to perform. Here, we identify four key criteria on which these methods can be evaluated and then examine how they perform in terms of the key criteria of accuracy. The results of these methods are then compared and cross-checked against an existing and widely used bot detection service. The findings show that different bot detection methods can present significantly different results and that only confirmation from Twitter, through suspensions or announcements, can truly allow users to know whether an account is a bot or not. We argue that this development could have a significant effect on the level of trust that social media users have both in the information they receive through social media and also in the political process.

Introduction Federated Learning (FL) is a decentralised approach to training statistical models, where training is performed across multiple clients, producing one global model. Since the training data remains with each local client and is not shared or exchanged with other clients the use of FL may reduce privacy and security risks (compared to methods where multiple data sources are pooled) and can also address data access and heterogeneity problems. Synthetic data is artificially generated data that has the same structure and statistical properties as the original but that does not contain any of the original data records, therefore minimising disclosure risk. Using FL to produce synthetic data (which we refer to as "federated synthesis") has the potential to combine data from multiple clients without compromising privacy, allowing access to data that may otherwise be inaccessible in its raw format. Objectives The objective was to review current research and practices for using FL to generate synthetic data and determine the extent to which research has been undertaken, the methods and evaluation practices used, and any research gaps. Methods A scoping review was conducted to systematically map and describe the published literature on the use of FL to generate synthetic data. Relevant studies were identified through online databases and the findings are described, grouped, and summarised. Information extracted included article characteristics, documenting the type of data that is synthesised, the model architecture and the methods (if any) used to evaluate utility and privacy risk. Results A total of 69 articles were included in the scoping review; all were published between 2018 and 2023 with two thirds (46) in 2022. 30% (21) were focussed on synthetic data generation as the main model output (with 6 of these generating tabular data), whereas 59% (41) focussed on data augmentation. Of the 21 performing federated synthesis, all used deep learning methods (predominantly Generative Adversarial Networks) to generate the synthetic data. Conclusions Federated synthesis is in its early days but shows promise as a method that can construct a global synthetic dataset without sharing any of the local client data. As a field in its infancy there are areas to explore in terms of the privacy risk associated with the various methods proposed, and more generally in how we measure those risks.

In this paper, we review the methodology of one of the most comprehensive indices of gender equality, the Gender Equality Index by the European Institute for Gender Equality (EIGE). Building on Permanyer’s (J Eur Soc Policy, 25(4):414–430, 2015) critical analysis, we offer an extended critique of the EIGE’s current methodology, focusing on four interrelated issues: (a) the lack of transparency around the methodological decisions and the concomitant implicit theorising, (b) the continuing over-contribution of the ‘correcting coefficient’ to the index such that it predominantly captures achievement levels rather than gender gaps, (c) problems with the verification process and use of Principal Component Analysis, (d) issues arising from the aggregation and weighting of index components. Our analysis shows that in addition to the use of the correcting coefficient, other methodological choices (such as the use of ratios and geometric means) result in an unjustified penalisation of lower-GDP countries, reinforcing biased assumptions about gender equality progress in more affluent countries vis-á-vis lower-GDP countries in the sample. We call for greater transparency around theory, method and the relationship between the two while also proposing methodological improvements. These changes would bring the EIGE index closer to fulfilling its undoubted potential to provide a nuanced understanding of gender equality levels in the European Union and effectively inform policy development toward social change.

Linking administrative data to produce more informative data for subsequent analysis has become an increasingly common practice. However, there might be concomitant risks of disclosing sensitive information about individuals. One practice that reduces these risks is data synthesis. In data synthesis the data are used to fit a model from which synthetic data are then generated. The synthetic data are then released to end users. There are some scenarios where an end user might have the option of using linked data or accepting synthesized data. However, linkage and synthesis are susceptible to errors that could limit their usefulness. Here, we investigate the problem of comparing the quality of linked data to synthesized data and demonstrate through simulations how the problem might be approached. These comparisons are important when considering how an end user can be supplied with the highest-quality data and in situations where one must consider risk/utility tradeoffs.

Child disability and provision of unpaid care by parents may contribute to a family’s poverty status. In this paper, using data from the Family Resources Survey-2018/19, a sample of 5451 families with dependent children was analysed. Sub-samples were created based on the groups of interest: child disability vs no child disability within the family and unpaid care vs no unpaid care for children with disabilities. Logistic regression models show sociodemographic factors have similar effects on children’s material deprivation and after housing costs poverty across all groups. In contrast, having children with no disability has higher odds of after housing costs poverty.