Multilevel regression and poststratification (MRP) is widely used to estimate opinion in small subgroups and to adjust unrepresentative surveys. Yet, even flexible MRP models contain errors generated by non-response and model misspecification. We propose a principled, data-driven method to leverage observable errors on auxiliary quantities with known marginal distributions—for example, election outcomes—to improve estimates of policy attitudes. Our method leverages the correlation between auxiliary variables and outcomes of interest to calibrate MRP estimates to these known marginal distributions. We illustrate our approach using a pre-election poll measuring support for an abortion referendum. We find that the method reduces county-level error by nearly two-thirds relative to traditional MRP. We also show how our calibration approach can be used to generate estimates for smaller nested geographies, such as precincts, even in the absence of poststratification data at this level. Our approach provides a framework for fully incorporating known population data to improve estimates of public opinion in small subgroups, providing scholars another tool to study representation.

This qualitative study explores how wage and payment practices inflict both economic hardship and psychological distress among tea garden workers in Sylhet and Moulvibazar. Using an interpretivist approach, semi-structured interviews were conducted with 12 purposively sampled workers in Sylhet. Transcripts in Bengali and the local Sylheti dialect were evaluated using reflexive thematic analysis. The findings reveal six key themes centered on the psychological impacts of wage insufficiency, unpredictable payment schedules, disproportionate quota penalties, and opaque deductions. Participants identified low base wages as the primary driver of chronic food insecurity and debt cycles. Furthermore, systemic issues like delayed payments, poor-quality rations, and gendered care burdens severely exacerbated their acute stress. Workers expressed their psychological suffering through culturally encoded idioms such as tension, chinta (worry), and khitkhite (irritability). Ultimately, the study highlights that workers' top priorities for mitigating distress include fair and sufficient wages, transparent payment systems, food security, and accessible healthcare. These findings emphasize that improving mental well-being in this vulnerable population requires systemic reforms in payment governance.

Using 3,154 tokens from American English, we test whether optionality in verb–particle placement increases speech-planning cost, measured as pre-verbal silence. Tokens were coded for object properties, idiomaticity and verb frequency. We find that pre-verbal silence does not differ between split (pick the book up) and joined (pick up the book) orders. While idiomaticity favours the joined order, it does not raise planning cost. Verb frequency shortens pauses only in fast speech, suggesting predictability acts lexically, not structurally. Choice symmetry does not lengthen pauses. We therefore fail to reject the null hypothesis: the two orders are equally easy to plan. This null result, from tests designed to detect a theoretically predicted effect, aligns with other evidence that syntactic choice imposes no production cost. We conclude that variation in verb–particle constructions (VPCs) is cost-free; distributional differences reflect object properties and idiomaticity, not derivational markedness.

We present and conduct a novel experiment on a multi-period beauty contest game. Leveraging the multi-period feature, we investigate how participants revise their forecasts in periods when new information—such as shocks or announcements—arrives and how they form their expectations in the absence of new information. We make two key contributions. First, we develop a new method based on forecast revisions to evaluate whether participants behave in a forward-looking manner. The experimental results show that participants do react to anticipated shocks: namely, the announcements of future shocks. Second, we identify a new strategic environment effect during periods without new information; only when the game exhibits strategic complementarity do participants use extrapolation and expect continuously rising prices. This finding suggests that expectation formation is endogenous to the economic environment; and policy design should thus take this endogeneity into account.

Although the shift from traditional to digital food environments has enhanced accessibility and convenience in food purchasing and delivery, emerging evidence suggests that current digital food environments tend to promote energy-dense foods rather than healthier options. Therefore, this commentary aims to highlight the challenges faced by healthcare professionals in shaping a healthier digital food environment and to offer practical recommendations for fostering such an environment in Malaysia. This commentary also emphasises that shaping a healthier digital food environment is not an individual endeavour but a collaborative effort among key stakeholders, including government ministries, platform operators, online food vendors and platform users.

Recent advances in Artificial Intelligence (AI) reveal a mismatch between the EU’s individual-centric data protection shell and automated systems’ ability to affect entire groups of people, sealed within that shell. AI routinely processes data about groups, while the GDPR still allocates rights and remedies to identifiable individuals. This design produces a structural collective redress gap where injunctions may be obtained without individual mandates (GDPR Art. 80(2), RAD Art. 8), but compensation remains tied to identifiability. Here, I show how the GDPR’s model (consent, data subjects’ rights, Art. 80) fits in and why it fails against AI’s capacity to de-anonymise and infer traits about persons whose data were never collected or volunteered. I then examine the AI Act 2024/1689. Although it repeatedly refers to “groups of persons,” trilogue negotiations removed the Parliament’s key collective remedy tools, leaving only ex‑ante risk and transparency duties. The example of online behavioural advertising or affinity profiling illustrates how AI‑designed groups fall outside the scope of meaningful GDPR or AI Act remedies. I argue for a right to group protection and propose four main revisions: a definition of groups, a mechanism for a group complaint and redress, group-level impact assessment requirements and group data protection by design.

Persistent funding shortfalls undermine protected areas (PAs) worldwide, yet few studies analyse these patterns across space and time. We examined funding deficits in 300 Brazilian federal PAs from 2014 to 2023 using spatial Durbin error models. Deficits were measured as the gap between evidence-based minimum management costs and actual spending. We analysed how PA age, size, management group, ecological region, population density and per capita GDP predict deficits, decomposing socioeconomic effects into direct and spillover components. In 2023, 72% of the PAs faced deficits totalling 958 million international dollars, despite a 30% investment increase over the decade. Larger PAs had greater shortfalls; older PAs had smaller ones. Amazon PAs averaged 79.2% deficits versus 27.6% in the Atlantic Forest. No significant difference emerged between management types. Higher population density predicted lower deficits, probably reflecting greater political visibility near urban centres. No direct local GDP effect was detected, but spillovers from neighbouring high-income regions suggest regional prosperity influences PA funding through spatial networks. Funding deteriorated in 2020–2021 amid fiscal contractions and policy shifts, then recovered in 2022–2023. These findings reveal deep structural inequities, particularly in the Amazon, highlighting the need for transparent national PA financing systems.

Early psychological deprivation is associated with increased risk for cardiometabolic disorders, and rapid catch-up growth following improved care environments may further influence this risk. Using data from the Bucharest Early Intervention Project – a randomized longitudinal study of children institutionalized in infancy assigned to high-quality foster care, continued institutional care, or raised by biological families – we examined cardiometabolic outcomes in early adulthood. Participants were 81 women and 58 men (mean age 22.6 years), predominantly Romanian. We assessed associations between care type and cardiometabolic traits including blood pressure, lipid and glucose metabolism, and metabolic syndrome. Results indicated that adults with histories of institutionalization displayed poorer lipid metabolism compared to non-institutionalized controls. Notably, individuals randomized to foster care showed increased BMI and waist circumference in adulthood relative to those who remained in institutional care. Longitudinal latent class analyses revealed four distinct BMI trajectories from infancy to early adulthood. The accelerated BMI trajectory, predominantly comprising foster care participants, was associated with the greatest prevalence of metabolic syndrome. These findings suggest that both lipid dysregulation and persistent increases in BMI represent ongoing concerns for cardiometabolic risk following early psychosocial deprivation. Those exposed to early adverse care may benefit from ongoing cardiometabolic monitoring and lifestyle interventions.

In 1970, Ted Williams – a medical missionary who had been running a small hospital at Kuluva in the West Nile region of Uganda for decades – was approached by the International Agency for Research on Cancer (IARC) to help establish a major study on the possible role of Epstein–Barr Virus in the aetiology of Burkitt’s lymphoma (BL). At the time, there was intense global interest in BL, as the cancer promised to be the first shown to have a viral aetiology. IARC centred its project at Kuluva Hospital and in the West Nile district because of Williams’ unusually detailed and accurate records. Williams was a meticulous record-keeper, who relied on various methods of data collection, from collecting and comparing to selecting and sorting. His paper-based archives and recording practices provide a rare window onto epidemiological knowledge production in East Africa in the decades before computing reshaped medical record-keeping. By tracing the ‘sociomaterial paper trail’ of Williams’s work, this paper examines how persona, place, and paper intersected in the making of medical knowledge, and how the researcher’s persona shapes the kinds of epidemiological data that are ultimately produced.