Replicability and reproducibility of predictive models for diagnosis of depression among young adults using Electronic Health Records.

BACKGROUND: Recent advances in machine learning combined with the growing availability of digitized health records offer new opportunities for improving early diagnosis of depression. An emerging body of research shows that Electronic Health Records can be used to accurately predict cases of depression on the basis of individual's primary care records. The successes of these studies are undeniable, but there is a growing concern that their results may not be replicable, which could cast doubt on their clinical usefulness. METHODS: To address this issue in the present paper, we set out to reproduce and replicate the work by Nichols et al. (2018), who trained predictive models of depression among young adults using Electronic Healthcare Records. Our contribution consists of three parts. First, we attempt to replicate the methodology used by the original authors, acquiring a more up-to-date set of primary health care records to the same specification and reproducing their data processing and analysis. Second, we test models presented in the original paper on our own data, thus providing out-of-sample prediction of the predictive models. Third, we extend past work by considering several novel machine-learning approaches in an attempt to improve the predictive accuracy achieved in the original work. RESULTS: In summary, our results demonstrate that the work of Nichols et al. is largely reproducible and replicable. This was the case both for the replication of the original model and the out-of-sample replication applying NRCBM coefficients to our new EHRs data. Although alternative predictive models did not improve model performance over standard logistic regression, our results indicate that stepwise variable selection is not stable even in the case of large data sets. CONCLUSION: We discuss the challenges associated with the research on mental health and Electronic Health Records, including the need to produce interpretable and robust models. We demonstrated some potential issues associated with the reliance on EHRs, including changes in the regulations and guidelines (such as the QOF guidelines in the UK) and reliance on visits to GP as a predictor of specific disorders.

Prediction and diagnosis of depression using machine learning with electronic health records data: a systematic review.

BACKGROUND: Depression is one of the most significant health conditions in personal, social, and economic impact. The aim of this review is to summarize existing literature in which machine learning methods have been used in combination with Electronic Health Records for prediction of depression. METHODS: Systematic literature searches were conducted within arXiv, PubMed, PsycINFO, Science Direct, SCOPUS and Web of Science electronic databases. Searches were restricted to information published after 2010 (from 1st January 2011 onwards) and were updated prior to the final synthesis of data (27th January 2022). RESULTS: Following the PRISMA process, the initial 744 studies were reduced to 19 eligible for detailed evaluation. Data extraction identified machine learning methods used, types of predictors used, the definition of depression, classification performance achieved, sample size, and benchmarks used. Area Under the Curve (AUC) values more than 0.9 were claimed, though the average was around 0.8. Regression methods proved as effective as more developed machine learning techniques. LIMITATIONS: The categorization, definition, and identification of the numbers of predictors used within models was sometimes difficult to establish, Studies were largely Western Educated Industrialised, Rich, Democratic (WEIRD) in demography. CONCLUSION: This review supports the potential use of machine learning techniques with Electronic Health Records for the prediction of depression. All the selected studies used clinically based, though sometimes broad, definitions of depression as their classification criteria. The reported performance of the studies was comparable to or even better than that found in primary care. There are concerns with generalizability and interpretability.

Fenneman et al.'s (2022) review of formal impulsivity models: Implications for theory and measures of impulsivity.

In Fenneman et al.'s (2022) review of theories and integrated impulsivity model, the authors distinguish between information impulsivity (i.e., acting without considering consequences) and temporal impulsivity (i.e., the tendency to pick sooner outcomes over later ones). The authors find that both types of impulsivity can be adaptive in different contexts. For example, when individuals experience scarcity of resources or when they are close to a minimum level of reserves (critical threshold). In this commentary, we extend their findings to a discussion about the measurement of impulsivity. We argue that a common method for measuring temporal impulsivity in which people make decisions between outcomes that are spaced out in time (intertemporal choice tasks), puts individuals in a specific context that is unlikely to generalize well to other situations. Furthermore, trait measures of impulsivity may only be modestly informative about future impulsive behavior because they largely abstract away from important context. To address these issues, we advocate for the development of dynamic measures of the two types of impulsivity. We argue that measuring temporal impulsivity in naturalistic contexts with varying environmental and state parameters could provide insights into whether individuals (i.e., humans and nonhuman animals) react to environmental changes adaptively, while trait measures of impulsivity more generally should collect and provide more contextual information. Dynamic measurement of different types of impulsivity will also allow for more discussion about adaptive impulsive responses in different contexts, which could help combat the stigmatization of various disorders associated with impulsivity. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Incomparability and Incommensurability in Choice: No Common Currency of Value?

Models of decision-making typically assume the existence of some common currency of value, such as utility, happiness, or inclusive fitness. This common currency is taken to allow comparison of options and to underpin everyday choice. Here we suggest instead that there is no universal value scale, that incommensurable values pervade everyday choice, and hence that most existing models of decision-making in both economics and psychology are fundamentally limited. We propose that choice objects can be compared only with reference to specific but nonuniversal "covering values." These covering values may reflect decision-makers' goals, motivations, or current states. A complete model of choice must accommodate the range of possible covering values. We show that abandoning the common-currency assumption in models of judgment and decision-making necessitates rank-based and "simple heuristics" models that contrast radically with conventional utility-based approaches. We note that if there is no universal value scale, then Arrow's impossibility theorem places severe bounds on the rationality of individual decision-making and hence that there is a deep link between the incommensurability of value, inconsistencies in human decision-making, and rank-based coding of value. More generally, incommensurability raises the question of whether it will ever be possible to develop single-quantity-maximizing models of decision-making.

Political Attitudes and Disease Threat: Regional Pathogen Stress Is Associated With Conservative Ideology Only for Older Individuals.

What environmental factors are associated with individual differences in political ideology, and do such associations change over time? We examine whether reductions in pathogen prevalence in U.S. states over the past 60 years are associated with reduced associations between parasite stress and conservatism. We report a positive association between infection levels and conservative ideology in the United States during the 1960s and 1970s. However, this correlation reduces from the 1980s onwards. These results suggest that the ecological influence of infectious diseases may be larger for older people who grew up (or whose parents grew up) during earlier time periods. We test this hypothesis by analyzing the political affiliation of 45,000 Facebook users, and find a positive association between self-reported political affiliation and regional pathogen stress for older (>40 years) but not younger individuals. It is concluded that the influence of environmental pathogen stress on ideology may have reduced over time.

Predicting implicit attitudes with natural language data.

Large-scale language datasets and advances in natural language processing offer opportunities for studying people's cognitions and behaviors. We show how representations derived from language can be combined with laboratory-based word norms to predict implicit attitudes for diverse concepts. Our approach achieves substantially higher correlations than existing methods. We also show that our approach is more predictive of implicit attitudes than are explicit attitudes, and that it captures variance in implicit attitudes that is largely unexplained by explicit attitudes. Overall, our results shed light on how implicit attitudes can be measured by combining standard psychological data with large-scale language data. In doing so, we pave the way for highly accurate computational modeling of what people think and feel about the world around them.

Rank-based alternatives to mean-based ensemble models of satisfaction with earnings: Comment on Putnam-Farr and Morewedge (2020).

How much satisfaction do we derive from a new salary or from receiving a bonus payment in an experiment? People do not judge monetary amounts in isolation but compare them to other amounts-judgments are context sensitive. A key question is, however, how context affects judgment. Across eight experiments, Putnam-Farr and Morewedge (2020) showed that people's self-reported satisfaction with a sum of money is predicted by the difference between that amount and the highest or lowest amount received by others. The authors found no evidence that people's judgments are sensitive to the ranked position of a monetary amount among other rewards. Putnam-Farr and Morewedge explained their results with reference to the ensemble representation literature, which shows that people can accurately estimate summary statistics, such as the maximum or mean, of stimulus distributions. In this commentary, we argue that their proposed interpretation is inconsistent with extensive theoretical and empirical research showing that judgments of stimuli reflect the relative ranked position of those stimuli within a comparison context. Building on this research, we show that the experimental results reported by Putnam-Farr and Morewedge can be explained on the assumption that people use contextual information to infer a distribution of monetary amounts and judge individual amounts by their relative ranked position within that inferred distribution. This inferred distribution theory accounts for empirical results reported in the original study while remaining consistent with the general and well-established principle of rank-based judgment. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Improving university students' mental health using multi-component and single-component sleep interventions: A systematic review and meta-analysis.

University is a time of significant transitions during a young adult's life, with delayed and shortened sleep and poor mental health a common occurrence. This systematic review and meta-analysis examined the effect of both multi-component and single-component sleep interventions on improving university students' sleep and mental health. Five databases (MEDLINE, PsycINFO, Embase, CINAHL and Cochrane Library) were searched for relevant literature published until April 2022. Treatment studies including university students aged 18-24 years, participating in a sleep intervention (multi-component, e.g., CBT-I, or single-component, e.g., sleep hygiene) were eligible. Comparator groups were either active, i.e., alternative intervention, or passive, i.e., waitlist control or treatment-as-usual, with study outcomes to include measures of sleep and mental health. Of 3435 references screened, 11 studies involving 5267 participants, with and without insomnia symptoms, were included for a narrative synthesis on intervention designs and methodology. Six studies eligible for meta-analyses showed a moderate effect of sleep interventions in reducing sleep disturbance (SMD = -0.548 [CI: -0.837, -0.258]) at post-treatment, alongside a small effect in improving anxiety (SMD = -0.226 [CI: -0.421, -0.031]) and depression (SMD = -0.295 [CI: -0.513, -0.077]). Meta-regression examining study and intervention characteristics identified subpopulation (experiencing insomnia or not) as a significant moderator for effects on sleep (p = 0.0003) and depression (p = 0.0063), with larger effects in studies with participants experiencing insomnia. Comparison group type (active or passive) was also a significant moderator (p = 0.0474), with larger effects on sleep in studies using passive comparison groups. Study type, delivery format, and intervention duration were not identified as significant moderators. At follow-ups, small but significant effects were sustained for anxiety and depression. Protecting and promoting sleep amongst university students may help safeguard and advance mental health.

Modelling pandemic behaviour using an economic multiplayer game.

During a pandemic, isolating oneself from the community limits viral transmission and helps avoid repeated societal lockdowns. This entails a social dilemma-either distance oneself from others for the benefit of the public good or free-ride and enjoy the benefits of freedom. It is not yet understood how the unfamiliar incentive structure and interpersonal context presented by a pandemic together modulate individuals' approach to this social dilemma. In this preregistered study, we take a game-theoretical approach and investigate people's decisions to self-isolate, using a novel iterated multiplayer game designed to capture the decision-making environment in the pandemic. To elucidate players' thinking, we use a variation of the strategy method and elicit beliefs about how much others will self-isolate. Players tend to respond to social norms with abidance, rather than transgression; they resist the temptation to freeride when others are self-isolating. However, they deal with exponential growth poorly, as they only self-isolate sufficiently when lockdowns are imminent. Further, increased collective risk can motivate more self-isolation, even though the link between self-isolation and lockdowns is stochastic. Players underreport the influence of others' choices on their own, and underestimate others' self-isolation. We discuss implications for public health, and communication to the public.

Effects of a hybrid digital cognitive-behavioural therapy for insomnia and emotion regulation in the workplace (SLEEP): study protocol for a randomised waitlist control trial.

INTRODUCTION: This trial tests the efficacy of implementing a hybrid digital cognitive-behavioural therapy for insomnia (dCBT-I) and emotion regulation (ER) in the workplace. The study protocol follows the SPIRIT (Standard Protocol Items: Recommendations for Intervention Trials) 2013 recommendations. METHODS AND ANALYSIS: This is a mixed methods evaluation with a two-arm randomised waitlist control design of a 6-week dCBT-I+ER intervention through self-guided online platform and four videoconferencing therapy sessions. A process evaluation will examine the fidelity of delivery and experiences of the intervention. The primary outcomes are the Insomnia Severity Index, the Patient Health Questionnaire-9 and the Generalised Anxiety Disorder-7. The secondary outcomes are job productivity, job satisfaction, well-being, quality of life, self-reported (sleep diary data) and objective (actigraphy) sleep parameters, and usage of online intervention platform. Assessments take place at baseline (T0), week 8 post-treatment (T1) and week 12 postrandomisation (T2). We will recruit 156 workers with sleep and ER problems ranging from subclinical to clinical levels not engaged in treatment at the time of the trial. ETHICS AND DISSEMINATION: Full approval was given by the University of Warwick Biomedical and Research Ethics Committee (BSREC 45/20-21). The current protocol version is 2.9_Dec21. Publication of results will inform the scientific, clinical and business communities through peer-reviewed articles, webinars, conferences and newsletters. TRIAL REGISTRATION NUMBER: ISRCTN13596153.

Pandemic and student mental health: mental health symptoms among university students and young adults after the first cycle of lockdown in the UK.

BACKGROUND: Early COVID-19 research suggests a detrimental impact of the initial lockdown on young people's mental health. AIMS: We investigated mental health among university students and young adults after the first UK lockdown and changes in symptoms over 6 months. METHOD: In total, 895 university students and 547 young adults not in higher education completed an online survey at T1 (July-September 2020). A subset of 201 university students also completed a 6 month follow-up survey at T2 (January-March 2021). Anxiety, depression, insomnia, substance misuse and suicide risk were assessed. RESULTS: At T1, approximately 40%, 25% and 33% of the participants reported moderate to severe anxiety and depression and substance misuse risk, clinically significant insomnia and suicidal risk. In participants reassessed at T2, reductions were observed in anxiety and depression but not in insomnia, substance misuse or suicidality. Student and non-student participants reported similar levels of mental health symptoms. Student status was not a significant marker of mental health symptoms, except for lower substance misuse risk.Cross-sectionally, greater symptoms across measures were consistently associated with younger age, pre-existing mental health conditions, being a carer, worse financial status, increased sleep irregularity and difficulty since lockdown. Longitudinally, T2 symptoms were consistently associated with worse financial status and increased difficulty sleeping at T1. However, these associations were attenuated when baseline mental health symptoms were adjusted for in the models. CONCLUSIONS: Mental health symptoms were prevalent in a large proportion of young people after the first UK lockdown. Risk factors identified may help characterise high-risk groups for enhanced support and inform interventions.

Content and Validity of Claims Made about Food Parenting Practices in United Kingdom Online News Articles.

The objective of this study was to qualitatively summarise the content of online news articles pertaining to food parenting practices and determine whether this content is substantiated by the scientific literature. News article data were identified and collected from United Kingdom online news published during 2010-2017 period using the News on the Web corpus. A coding framework was used to categorise the content of news articles to identify information related to food parenting practices. Then, claims made about food parenting practices were extracted from relevant news articles. Each claim was evaluated to determine the extent to which any claims were supported by the available scientific research evidence. The study identified ten claims across thirty-two relevant online news articles. Claims made across the news articles reported on the following food parenting practices: food restrictions, food-based threats and bribes, pressure to eat, use of food to control negative emotions, food availability, food preparation, and meal and snack routines. Eight out of the ten claims identified did not refer to scientific research evidence. News articles frequently lacked detail and information to explain to readers why and how the use of certain food parenting practices could have a lasting impact on children's health outcomes. Considering the influence that news media has on parents, the reporting of food parenting practices in news articles should aim to provide a balanced view of the published scientific evidence and recognise the difficulties and barriers that prevent the use of helpful and healthy food parenting practices. The study results in this paper could be used to aid and structure of the dissemination of food parenting practice research findings in the media, inform public health education to influence perceptions of unhelpful food parenting practices, and promote parental use of responsive food parenting practices.

Workplace inequality is associated with status-signaling expenditure.

Regional inequality is known to magnify sensitivity to social rank. This, in turn, is shown to increase people's propensity to acquire luxury goods as a means to elevate their perceived social status. Yet existing research has focused on broad, aggregated datasets, and little is known about how individual-level measures of income interact with inequality within peer groups to affect status signaling. Using detailed financial transaction data, we construct 32,008 workplace peer groups and explore the longitudinal spending and salary data associated with 683,677 individuals. These data reveal links between people's status spending, their absolute salary, salary rank within their workplace peer group, and the inequality of their workplace salary distribution. Status-signaling luxury spending is found to be greatest among those who have higher salaries, whose workplaces exhibit higher inequality, and who occupy a lower rank position within the workplace. We propose that low-rank individuals in unequal workplaces suffer status anxiety and, if they can afford it, spend to signal higher status.

Assessing physical access to healthy food across United Kingdom: A systematic review of measures and findings.

Background: Existing research suggests that physical access to food can affect diet quality and thus obesity rates. When defining retail food environment (RFE) quantitatively, there is a little agreement on how to measure "lack of healthy food" and what parameters to use, resulting in a heterogeneity of study designs and outcome measures. In turn, this leads to a conflicting evidence base being one of the many barriers to using evidence in policy-making. Aims: This systematic review aimed to identify and describe methods used to assess food accessibility in the United Kingdom (UK) to overcome heterogeneity by providing a classification of measures. Materials & Methods: The literature search included electronic and manual searches of peer-reviewed literature and was restricted to studies published in English between January 2010 and March 2021. A total of 9365 articles were assessed for eligibility, of which 44 articles were included in the review. All included studies were analysed with regards to their main characteristics (e.g., associations between variables of interest, setting, sample, design, etc.) and definition of RFE and its metrics. When defining these metrics, the present review distinguishes between a point of origin (centroid, address) from which distance was calculated, summary statistic of accessibility (proximity, buffer, Kernel), and definition of distance (Euclidean, network distance). Trends, gaps and limitations are identified and recommendations made for food accessibility research in UK. Results: Multiple theoretical and methodological constructs are currently used, mostly quantifying distance by means of Euclidean and ring-buffer distance, using both proximity- and density-based approaches, and ranging from absolute to relative measures. The association between RFE and diet and health in rural areas, as well as a spatiotemporal domain of food access, remains largely unaccounted. Discussion: Evidence suggests that the duration of exposure may bear a greater importance than the level of exposure and that density-based measures may better capture RFE when compared with proximity-based measures, however, using more complex measures not necessarily produce better results. To move the field forward, studies have called for a greater focus on causality, individual access and the use of various measures, neighbourhood definitions and potential confounders to capture different aspects and dimensions of the RFE, which requires using univariate measures of accessibility and considering the overall context in terms of varying types of neighbourhoods. Conclusion: In order to render ongoing heterogeneity in measuring RFE, researchers should prioritise measures that may provide a more accurate and realistic account of people's lives and follow an intuitive approach based on convergence of results until consensus could be reached on using some useful standards.

Impact of the "when the fun stops, stop" gambling message on online gambling behaviour: a randomised, online experimental study.

BACKGROUND: Safer gambling messages are a common freedom-preserving method of protecting individuals from gambling-related harm. Yet, there is little independent and rigorous evidence assessing the effectiveness of safer gambling messages. In our study, we aimed to test the effect of the historically most commonly-used UK safer gambling message on concurrent gambling behaviour of people who gamble in the UK. METHODS: In this study, three preregistered, incentivised, and randomised online experiments, testing the UK's "when the fun stops, stop" message, were carried out via the crowdsourcing platform Prolific. Adults based in the UK who had previously participated in the gambling activities relevant to each experiment were eligible to participate. Experiments 1 and 3 involved bets on real soccer events, and experiment 2 used a commercially available online roulette game. Safer gambling message presence was varied between participants in each experiment. In experiment 2, exposed participants could be shown either a yellow or a black-and-white version of the safer gambling message. Participants were provided with a monetary endowment with which they were allowed to bet. Any of this money not bet was afterwards paid to participants as a bonus, in addition to the payouts from any winning bets. In experiment 2 participants had the opportunity to re-wager any winnings from the roulette game. The primary outcome in experiment 1 was participants' decisions to accept (or reject) a series of football bets, which varied in their specificity (and payoffs), and the primary outcomes of experiments 2 and 3 were the proportion of available funds bet, which were defined as the total amount of money bet by a participant out of the total that could have been bet. FINDINGS: Participants for all three experiments were recruited between May 17, 2019, and Oct 17, 2020. Of the 506 participants in experiment 1, 41·3% of available bets were made by the 254 participants in the gambling message condition, which was not significantly different (p=0·15, odds ratio 1·22 [95% CI 0·93 to 1·61]) to the 37·8% of available bets made by the 252 participants in the control condition. In experiment 2, the only credible difference between conditions was that the 501 participants in the condition with the yellow version of the gambling message bet 3·64% (95% Bayesian credibility interval 0·00% to 7·27%) more of available funds left over than the 499 participants in the control condition. There were no credible differences between the bets made by the 500 participants in the black-and-white gambling message condition and the other conditions. In experiment 3, there were no credible differences between the 502 participants in the gambling message condition and the 501 participants in the control condition, with the largest effect being a 5·87% (95% Bayesian credibility interval -1·44% to 13·20%) increase in the probability of betting everything in the gambling message condition. INTERPRETATION: In our study, no evidence was found for a protective effect of the most common UK safer gambling message. Alternative interventions should be considered as part of an evidence-based public health approach to reducing gambling-related harm. FUNDING: University of Warwick, British Academy and Leverhume, Swiss National Science Foundation.

A feasibility study of the delivery of online brief cognitive-behavioral therapy (CBT-T) for eating disorder pathology in the workplace.

OBJECTIVE: CBT-T is a brief (10 sessions) version of cognitive behavioral therapy for non-underweight eating disorders. This report describes the protocol for a single center, single group, feasibility trial of online CBT-T in the workplace as an alternative to the health-service setting. By offering mental health services for eating disorders in the workplace, greater accessibility and increased help-seeking behaviors could be achieved. METHOD: Treatment will be delivered online over 10 weeks and offered to employees based on self-referral rather than meeting diagnostic criteria, making treatment available to employees with sub-threshold eating disorder symptoms. RESULTS: Assessments will be conducted at baseline, mid-treatment (week 4), posttreatment (week 10) and at follow-up (1 month and 3 months posttreatment). For the primary outcome, measures will include recruitment, attrition and attendance data using pre-set benchmarks to determine high, medium or low feasibility and acceptability. Qualitative participant experiences data will be analyzed using thematic analysis. Impact on work engagement and effect sizes will be determined from secondary outcome measures; the latter enabling sample size calculations for future study. DISCUSSION: These pilot data will provide insights to recruitment, acceptability, effectiveness and viability of a future fully powered clinical trial of online CBT-T in the workplace. PUBLIC SIGNIFICANCE STATEMENT: This study will present feasibility data from an eating disorders intervention (online CBT-T) using the workplace as an alternative to the healthcare setting to recruit and treat workers. Recruitment will be based on self-reported eating and weight concerns rather than diagnosis potentially enabling treatment to employees who have not previously sought help. The data will also provide insights to recruitment, acceptability, effectiveness, and future viability of CBT-T in the workplace.

Computational Methods for Predicting and Understanding Food Judgment.

People make subjective judgments about the healthiness of different foods every day, and these judgments in turn influence their food choices and health outcomes. Despite the importance of such judgments, there are few quantitative theories about their psychological underpinnings. This article introduces a novel computational approach that can approximate people's knowledge representations for thousands of common foods. We used these representations to predict how both lay decision-makers (the general population) and experts judge the healthiness of individual foods. We also applied our method to predict the impact of behavioral interventions, such as the provision of front-of-pack nutrient and calorie information. Across multiple studies with data from 846 adults, our models achieved very high accuracy rates (r2 = .65-.77) and significantly outperformed competing models based on factual nutritional content. These results illustrate how new computational methods applied to established psychological theory can be used to better predict, understand, and influence health behavior.

Psychosocial factors influencing the eating behaviours of older adults: A systematic review.

Our understanding of how eating behaviours change in later life have been dominated by the studies of physiological and biological influences on malnutrition. Insights from these studies were consequently used to develop interventions, which are predominantly aimed at rectifying nutritional deficiencies, as opposed to interventions that may enable older adults to eat well and enjoy their food-related life well into older age. The objective of the present review is to summarise the existing knowledge base on psychosocial influences on eating behaviours in later life. Following comprehensive searches, review, and appraisal, 53 articles were included (22 qualitative and 31 quantitative) to provide a greater understanding of the mechanisms underpinning the psychosocial factors influencing eating behaviours. Our analysis identified eight underpinning psychosocial factors that influences eating behaviours in later life; (1) health awareness & attitudes, (2) food decision making, (3) perceived dietary control, (4) mental health & mood, (5) food emotions & enjoyment, (6) eating arrangements, (7) social facilitation, and (8) social support. The importance and lasting influence of early food experiences were also identified as contributing to eating behaviours in later life. The review concludes with the call for further investigation into specific psychosocial factors that influence eating behaviour, calls for improvements in methodologies, and a summary of psychosocial barriers and enablers to eating well in later life.

Risk perception, illusory superiority and personal responsibility during COVID-19: An experimental study of attitudes to staying home.

Little is known about how different government communication strategies may systematically affect people's attitudes to staying home or going out during the COVID-19 pandemic, nor how people perceive and process the risk of viral transmission in different scenarios. In this study, we report results from two experiments that examine the degree to which people's attitudes regarding the permissibility of leaving one's home are (1) sensitive to different levels of risk of viral transmission in specific scenarios, (2) sensitive to communication framings that are either imperative or that emphasize personal responsibility, or (3) creating 'loopholes' for themselves, enabling a more permissive approach to their own compliance. We find that the level of risk influences attitudes to going out, and that participants report less permissive attitudes to going out when prompted with messages framed in imperative terms, rather than messages emphasizing personal responsibility; for self-loopholes, we find no evidence that participants' attitudes towards going out in specific scenarios are more permissive for themselves than for others. However, participants report they are more rigorous in staying home than others, which may cause moral licensing. Additionally, we find that age is negatively associated with permissive attitudes, and that male participants are more permissive to going out. Thus, during phases where it is important to promote staying home for all scenarios, including those perceived to be low-risk, imperative communication may be best suited to increase compliance.