Probability-based Sampling Research Articles

Active learning in automated text classification: a case study exploring bias in predicted model performance metrics

Machine learning has emerged as a cost-effective innovation to support systematic literature reviews in human health risk assessments and other contexts. Supervised machine learning approaches rely on a training dataset, a relatively small set of documents with human-annotated labels indicating their topic, to build models that automatically classify a larger set of unclassified documents. “Active” machine learning has been proposed as an approach that limits the cost of creating a training dataset by interactively and sequentially focussing on training only the most informative documents. We simulate active learning using a dataset of approximately 7000 abstracts from the scientific literature related to the chemical arsenic. The dataset was previously annotated by subject matter experts with regard to relevance to two topics relating to toxicology and risk assessment. We examine the performance of alternative sampling approaches to sequentially expanding the training dataset, specifically looking at uncertainty-based sampling and probability-based sampling. We discover that while such active learning methods can potentially reduce training dataset size compared to random sampling, predictions of model performance in active learning are likely to suffer from statistical bias that negates the method’s potential benefits. We discuss approaches and the extent to which the bias resulting from skewed sampling can be compensated. We propose a useful role for active learning in contexts in which the accuracy of model performance metrics is not critical and/or where it is beneficial to rapidly create a class-balanced training dataset.

Prevalence and Severity of Food Allergies Among US Adults

Food allergy is a costly, potentially life-threatening condition. Although studies have examined the prevalence of childhood food allergy, little is known about prevalence, severity, or health care utilization related to food allergies among US adults. To provide nationally representative estimates of the distribution, severity, and factors associated with adult food allergies. In this cross-sectional survey study of US adults, surveys were administered via the internet and telephone from October 9, 2015, to September 18, 2016. Participants were first recruited from NORC at the University of Chicago's probability-based AmeriSpeak panel, and additional participants were recruited from the non-probability-based Survey Sampling International (SSI) panel. Demographic and allergic participant characteristics. Self-reported food allergies were the main outcome and were considered convincing if reported symptoms to specific allergens were consistent with IgE-mediated reactions. Diagnosis history to specific allergens and food allergy-related health care use were also primary outcomes. Estimates were based on this nationally representative sample using small-area estimation and iterative proportional fitting methods. To increase precision, AmeriSpeak data were augmented by calibration-weighted, non-probability-based responses from SSI. Surveys were completed by 40 443 adults (mean [SD] age, 46.6 [20.2] years), with a survey completion rate of 51.2% observed among AmeriSpeak panelists (n = 7210) and 5.5% among SSI panelists (n = 33 233). Estimated convincing food allergy prevalence among US adults was 10.8% (95% CI, 10.4%-11.1%), although 19.0% (95% CI, 18.5%-19.5%) of adults self-reported a food allergy. The most common allergies were shellfish (2.9%; 95% CI, 2.7%-3.1%), milk (1.9%; 95% CI, 1.8%-2.1%), peanut (1.8%; 95% CI, 1.7%-1.9%), tree nut (1.2%; 95% CI, 1.1%-1.3%), and fin fish (0.9%; 95% CI, 0.8%-1.0%). Among food-allergic adults, 51.1% (95% CI, 49.3%-52.9%) experienced a severe food allergy reaction, 45.3% (95% CI, 43.6%-47.1%) were allergic to multiple foods, and 48.0% (95% CI, 46.2%-49.7%) developed food allergies as an adult. Regarding health care utilization, 24.0% (95% CI, 22.6%-25.4%) reported a current epinephrine prescription, and 38.3% (95% CI, 36.7%-40.0%) reported at least 1 food allergy-related lifetime emergency department visit. These data suggest that at least 10.8% (>26 million) of US adults are food allergic, whereas nearly 19% of adults believe that they have a food allergy. Consequently, these findings suggest that it is crucial that adults with suspected food allergy receive appropriate confirmatory testing and counseling to ensure food is not unnecessarily avoided and quality of life is not unduly impaired.

The potential for citizen science to produce reliable and useful information in ecology

We examined features of citizen science that influence data quality, inferential power, and usefulness in ecology. As background context for our examination, we considered topics such as ecological sampling (probability based, purposive, opportunistic), linkage between sampling technique and statistical inference (design based, model based), and scientific paradigms (confirmatory, exploratory). We distinguished several types of citizen science investigations, from intensive research with rigorous protocols targeting clearly articulated questions to mass‐participation internet‐based projects with opportunistic data collection lacking sampling design, and examined overarching objectives, design, analysis, volunteer training, and performance. We identified key features that influence data quality: project objectives, design and analysis, and volunteer training and performance. Projects with good designs, trained volunteers, and professional oversight can meet statistical criteria to produce high‐quality data with strong inferential power and therefore are well suited for ecological research objectives. Projects with opportunistic data collection, little or no sampling design, and minimal volunteer training are better suited for general objectives related to public education or data exploration because reliable statistical estimation can be difficult or impossible. In some cases, statistically robust analytical methods, external data, or both may increase the inferential power of certain opportunistically collected data. Ecological management, especially by government agencies, frequently requires data suitable for reliable inference. With standardized protocols, state‐of‐the‐art analytical methods, and well‐supervised programs, citizen science can make valuable contributions to conservation by increasing the scope of species monitoring efforts. Data quality can be improved by adhering to basic principles of data collection and analysis, designing studies to provide the data quality required, and including suitable statistical expertise, thereby strengthening the science aspect of citizen science and enhancing acceptance by the scientific community and decision makers.

Active Machine Learning in Systematic Literature Reviews: Bias, Fixes, and Appropriate Use

Text analytics has emerged as a cost-effective innovation used to support systematic literature reviews in human health risk assessments. Text analytics approaches allow some studies to be removed from consideration without undergoing manual review. Supervised machine learning relies on a training dataset to build models that go on to automatically classify a larger set of unclassified documents. However, manually reviewing references to create the training dataset can be resource intensive. “Active” machine learning is a potential solution that uses an algorithm to focus on the most informative documents, reducing the number of references that must be manually reviewed. We simulate active machine learning by using a set of approximately 7,000 abstracts from the scientific literature that was previously classified by subject matter experts with regard to relevance to epidemiology. We examine the performance of alternative sampling approaches to sequentially expanding the training dataset, specifically looking at uncertainty-based sampling and probability-based sampling. We discover that while such active learning methods can potentially reduce training dataset size compared to random sampling, active machine learning-based predictions of model performance potentially suffer from bias that negates its potential benefits. We discuss approaches and the extent to which the bias resulting from skewed sampling can be compensated. We compare the results from active machine learning to those based on a semi-supervised machine learning method called supervised clustering and show how the latter, with only a small training dataset, can outperform the former in terms of (i) accuracy of model predictions and (ii) the fraction of documents eliminated from review. Finally, we propose a useful role for active learning in contexts where accuracy metrics are not critical and/or where it is necessary to rapidly retrieve a subset of relevant literature.

Community organizations and mental health after the 2013 Boston Marathon bombings

ObjectiveDisasters are place-based traumatic events, yet contemporary understandings of disaster recovery often do not consider the role of community organizations. We examine organization type and proximity as they relate to post-disaster mental health in a longitudinal study following the 2013 Boston Marathon bombings. MethodResidents of metropolitan Boston (N = 846) were recruited via a probability-based sampling strategy within weeks of the bombings and were surveyed several times over a two-year period. Residents of metropolitan New York (N = 941) were recruited and surveyed at the same time and used for comparison due to similarities in community demographics, geography, and disaster histories. We identified six different organization types nearby resident: safety-based organizations, religious organizations, educational organizations, child- and family-promoting organizations, health-based organizations, and voluntary community organizations. With possible environmental detriments (crowds and noise) or benefits of organizations amplified in areas closest to the resident, the concentration of these local organization types was examined at different distance-based boundaries. Contextual data for both communities came from the U.S. Census, Google Places API, and Guidestar. ResultsFor Boston metropolitan area residents, having more safety-based organizations within a half-mile to one-mile area in the aftermath of the bombings was associated with poorer functioning six to seven months later and greater psychological distress two years later. However, the presence of more safety-based organizations in the one to three mile area was associated with decreased psychological distress two years later. More health-based and voluntary community organizations in the half-mile to one-mile area were also associated with fewer fears and worries about future adversity two years post-bombing. Exposure to the bombings and other community traumas moderated this relationship among Boston area participants. ConclusionResults suggest that local community organizations are not merely buildings or structures but ecological sources of support to those in need after a disaster.

How job resources and personal resources influence work engagement and burnout

PurposeThe purpose of this paper is to present a model of the relationships between personal resources (Psychological Capital (PsyCap)) and satisfaction with job resources, and their effect on work engagement and burnout.Design/methodology/approachData were collected from a convenience sample of 407 full-time employees from various public and private sector organisations, using a questionnaire consisting of PsyCap (PCQ-24), the Utrecht Work Engagement Scales, the Maslach Burnout Inventory, and a questionnaire measuring job resources (Parker and Hyett, 2011). The data were analysed using variance-based structural equation modelling (SmartPLS 3).FindingsThe influence of employees’ satisfaction with job resources on both dimensions of burnout (emotional exhaustion and cynicism) was negative and statistically significant. Satisfaction with job resources had a statistically significant positive influence on both dimensions of work engagement (vigour and dedication). PsyCap had a statistically significant positive influence on satisfaction with job resources. Satisfaction with job resources partially mediated the influence of PsyCap on emotional exhaustion and cynicism, and partially on vigour and dedication.Research limitations/implicationsAs this was an exploratory study, it used a convenience sample and a variance-based approach to structural equation modelling (SmartPLS). It is suggested that future researchers replicate the model in different contexts to corroborate the proposed relationships using larger samples, probability-based sampling and a covariance-based approach to structural equation modelling.Practical implicationsManagement must realise that employees’ satisfaction with job resources plays a central role in their work engagement and burnout levels. Workplace practices that reflect respect and care for the employee and the development of employees’ personal resources (i.e. PsyCap) will improve work engagement and reduce burnout.Originality/valueThis paper fills a gap in the literature by explaining how personal resources (PsyCap) and job resources (the organisation’s perceived respect for the employee and employer care) influence work engagement and burnout via mediation paths.

Combining Survey and Non-survey Data for Improved Sub-area Prediction Using a Multi-level Model

Combining information from different sources is an important practical problem in survey sampling. Using a hierarchical area-level model, we establish a framework to integrate auxiliary information to improve state-level area estimates. The best predictors are obtained by the conditional expectations of latent variables given observations, and an estimate of the mean squared prediction error is discussed. Sponsored by the National Agricultural Statistics Service of the US Department of Agriculture, the proposed model is applied to the planted crop acreage estimation problem by combining information from three sources, including the June Area Survey obtained by a probability-based sampling of lands, administrative data about the planted acreage and the cropland data layer, which is a commodity-specific classification product derived from remote sensing data. The proposed model combines the available information at a sub-state level called the agricultural statistics district and aggregates to improve state-level estimates of planted acreages for different crops. Supplementary materials accompanying this paper appear on-line.

Alternative methods for selecting web survey samples

Probability-based sampling is the gold standard for general population surveys. However, when interested in more specific populations (e.g., consumers of a particular brand), a lot of research uses data from non-probability-based online panels. This article investigates different ways to select a sample in an opt-in panel: without previous information, using profiling information, or using passive data from a tracker installed on the panelists’ devices. Moreover, it investigates the effect of sending the survey closer to the “moment-of-truth,” which is expected to reduce memory limitations in recall questions. Using additional information (profiling or passive) to select the sample leads to clear improvements in terms of levels of participation and fieldwork efficiency, but not in terms of data quality (measured by the proportion of don’t know answers and the length of answers to open narrative questions) or accuracy (measured by comparing the answers to 14 questions to an external source of information). Doing the survey closer to the “moment-of-truth” further improves the fieldwork efficiency; however, there are still many challenges to implement true “in-the-moment” surveys. We also observed differences across the different samples in respondents’ socio-demographic characteristics and in the survey evaluation.

The Examination of the Social Integration Perceptions of Undergraduate Students

This study was aimed to determine the social integration perceptions of undergraduate students and to examine them in terms of certain variables. It was a descriptive study with survey methodology. The data were obtained using the Social Integration Scale. The study group consisted of 545 undergraduate students during the fall semester of 2016-2017 academic year. The research sample was determined using probability-based sampling techniques. SPSS 20.0 software package was used in data analysis. Mann Whitney U and Kruskal Wallis H-Tests were conducted to research data. According to the research results, it was revealed that undergraduate students had higher social integration perceptions in general. The research findings indicated that undergraduate students’ perceptions of integration with the social environment differ significantly according to their departments, ages and income; their local integration perceptions according to the geographical region they reside and the occupation of mother; their educational integration perceptions according to their departments, ages and income; their normative-moral integration perceptions according to their departments and the educational degrees of parents.

Disaster-related communication preferences of homeless and nonhomeless VA patients.

To identify the communication preferences of homeless (H) and nonhomeless (NH) Veterans Affairs (VA) patients for receiving information about the impact of natural disasters on VA healthcare services. Probability-based sampling design stratified on age categories, urban/rural, coastal/noncoastal, and homelessness. Northeast United States. Eligible NH and H participants included those who lived in and had at least one healthcare visit to a VA medical center/clinic in the region in the previous 24 months. Homeless participants included those receiving VA homeless services or having ICD9-CM V60.0 (Lack of Housing). 2,264/6,088 NH and 383/2,000 H completed the survey. Cross-sectional, mixed-mode survey administered August-November 2015. Helpfulness of ten communication modes for receiving information about the impact of natural disasters on VA healthcare services. Despite sociodemographic and health status differences, the top five communication modes rated extremely/very helpful were the same for both groups: telephone (76 percent H; 81 percent NH), TV (63 percent H; 60 percent NH), text (62 percent H; 50 percent NH), e-mail (56 percent H; 58 percent NH), radio (54 percent H; 47 percent NH). Newspaper and online modes (VA/other Web sites, Twitter, Facebook, MyHealtheVet) were rated substantially lower. H and NH patients prefer similar communication modes for receiving updates about the impact of natural disasters on VA healthcare services. Findings suggest that a multimodal communication strategy that incorporates phone, TV, text, e-mail, and radio will help Veterans Affairs (VA) reach its diverse patient population during natural disasters. Dissemination of messages via online modes, while rated less helpful, may augment the number of patients reached.

Prevalence, treatment patterns, and risk factors of hypertension and pre-hypertension among Bangladeshi adults

In Bangladesh, morbidity and mortality due to non-communicable diseases (NCDs) has increased over the last few decades. Hypertension is an important risk factor for NCDs, specifically cardiovascular disease. The objective of this study was to assess prevalence and risk factors for hypertension and pre-hypertension among adults in Bangladesh. Data for this analysis were collected during the national NCD Risk Factor Survey of Bangladesh conducted in 2010 from a representative sample of men and women, aged 25 years or above. The survey adopted a multistage, geographically clustered, probability-based sampling approach. WHO STEPS questionnaire was used to collect data on demographics, behavioral risk factors, and physical measurements. Overall, 20% of the study population were hypertensive at study measurement. The prevalence of hypertension increased with age and body mass index (BMI). Twelve percent of the population were previously diagnosed with hypertension. Among these individuals, nearly half were not taking any medications to control their hypertension. Additionally, the prevalence of pre-hypertension was 43%, with higher levels among males, older age groups, and those with higher education, higher wealth index and high BMI. Predictors of hypertension, included older age, high BMI, and diabetes comorbidity. Based on this study, we estimate that 1 out of 5 Bangladeshi adults have hypertension. The risk of hypertension increases with older age and high BMI. Additionally, prevalence of pre-hypertension is high in Bangladesh in both rural and urban areas. Findings from this study can be used to inform public health programming to control the spread of NCDs in Bangladesh.

Community Policies and Programs to Prevent Obesity and Child Adiposity

Evidence regarding impact of community policies and programs (CPPs) to prevent child obesity is limited, and which combinations of strategies and components are most important is not understood. The Healthy Communities Study was an observational study to assess relationships of characteristics and intensity of CPPs with adiposity, diet, and physical activity in children, taking advantage of variation across the U.S. in community actions to prevent child obesity. The study examined the association of CPPs to prevent child obesity with measured BMI and waist circumference, hypothesizing that communities with more-comprehensive CPPs would have children with lower adiposity. The study included 130 communities selected by probability-based sampling or because of known CPPs targeting child obesity. Data were collected at home visits on 5,138 children during 2013-2015. CPPs were scored for multiple attributes to create a CPP intensity score. A CPP target behavior score reflected the number of distinct target behaviors addressed. Scores were standardized with the smallest observed score across communities being 0 and the largest 1. Multilevel regression analysis in 2016 adjusted for community, household, and individual characteristics. Higher CPP target behavior score was significantly associated with lower BMI and waist circumference in a dose-response relationship, with magnitude for the past 3 years of CPPs of 0.843 (p=0.013) for BMI and 1.783 cm (p=0.020) for waist circumference. This study provides plausible evidence that comprehensive CPPs targeting a greater number of distinct physical activity and nutrition behaviors were associated with lower child adiposity.

Impact of an HIV Prevention Intervention on HIV Risk Behavior and Sexually Transmitted Infection among Female Sex Workers in Tamil Nadu, India.

BackgroundIn 2003 a large-scale HIV prevention program was launched for key populations in six high HIV prevalence states of India. This paper assesses the effect of exposure to the intervention on condom use with commercial clients and experiences of sexually transmitted infection (STI) among female sex workers (FSWs) in Tamilnadu, a southern Indian state, between 2006 and 2009.MethodsData were drawn from two rounds of cross-sectional surveys conducted in 2006 (N = 2010) and 2009 (N = 2500) among FSWs in Tamilnadu, recruited through probability-based sampling. A series of multivariate logistic regression models were constructed to examine the association between exposure to the intervention and change over time with condom use and self-reported STI. All the analyses were performed using STATA 11.1.ResultsOverall, 48% of FSWs in 2006 and 90% in 2009 reported exposure to the intervention. Compared to 2006, there was a considerable increase in the proportion of FSWs reporting consistent condom use with regular and occasional clients at follow-up (2009). Further, the increase in consistent condom use over time with occasional (adjusted OR = 3.53, 95% CI: 2.47 - 5.05) and regular clients (adjusted OR = 4.97, 95% CI: 3.43 - 7.16) was significantly higher among FSWs exposed to the intervention than those not exposed. Additionally, a significant decline was observed in self-reported STI overtime among FSWs exposed to the intervention compared to their counterparts (adjusted OR = 0.39, 95% CI: 0.26 - 0.59).ConclusionThe HIV prevention program in Tamilnadu resulted in increased consistent condom use and a decrease in self-reported STI among FSWs exposed to intervention. These findings suggest that HIV prevention programs should aim to saturate coverage among key populations to sustain the gains achieved.

Support for Food and Beverage Worksite Wellness Strategies and Sugar-Sweetened Beverage Intake Among Employed U.S. Adults.

Sugar-sweetened beverage (SSB) consumption is high among U.S. adults and is associated with obesity. Given that more than 100 million Americans consume food or beverages at work daily, the worksite may be a venue for interventions to reduce SSB consumption. However, the level of support for these interventions is unknown. We examined associations between workday SSB intake and employees' support for worksite wellness strategies (WWSs). We conducted a cross-sectional study using data from Web-based annual surveys that gather information on health-related attitudes and behaviors. Study setting was the United States. A total of 1924 employed adults (≥18 years) selected using probability-based sampling. The self-reported independent variable was workday SSB intake (0, <1 or ≥1 times per day), and dependent variables were employees' support (yes/no) for the following WWSs: (1) accessible free water, (2) affordable healthy food/drink, (3) available healthy options, and (4) less available SSB. Multivariable logistic regression was used to control for sociodemographic variables, employee size, and availability of cafeteria/vending machine. About half of employees supported accessible free water (54%), affordable healthy food/drink (49%), and available healthy options (46%), but only 28% supported less available SSB. Compared with non-SSB consumers, daily SSB consumers were significantly less supportive of accessible free water (adjusted odds ratio, .67; p < .05) or less available SSB (odds ratio, .49; p < .05). Almost half of employees supported increasing healthy options within worksites, although daily workday SSB consumers were less supportive of certain strategies. Lack of support could be a potential barrier to the successful implementation of certain worksite interventions.

Methodology for Evaluating the Quality of Ecosystem Maps: A Case Study in the Andes

Uncertainty in thematic maps has been tested mainly in maps with discrete or fuzzy classifications based on spectral data. However, many ecosystem maps in tropical countries consist of discrete polygons containing information on various ecosystem properties such as vegetation cover, soil, climate, geomorphology and biodiversity. The combination of these properties into one class leads to error. We propose a probability-based sampling design with two domains, multiple stages, and stratification with selection of primary sampling units (PSUs) proportional to the richness of strata present. Validation is undertaken through field visits and fine resolution remote sensing data. A pilot site in the center of the Colombian Andes was chosen to validate a government official ecosystem map. Twenty primary sampling units (PSUs) of 10 × 15 km were selected, and the final numbers of final sampling units (FSUs) were 76 for the terrestrial domain and 46 for the aquatic domain. Our results showed a confidence level of 95%, with the accuracy in the terrestrial domain varying between 51.8% and 64.3% and in the aquatic domain varying between 75% and 92%. Governments need to account for uncertainty since they rely on the quality of these maps to make decisions and guide policies.

Multilevel Provider-Based Sampling for Recruitment of Pregnant Women and Mother-Newborn Dyads.

In 2010, the National Children's Study launched 3 alternative recruitment methods to test possible improvements in efficiency compared with traditional household-based recruitment and participant enrollment. In 2012, a fourth method, provider-based sampling (PBS), tested a probability-based sampling of prenatal provider locations supplemented by a second cohort of neonates born at a convenience sample of maternity hospitals. From a sampling frame of 472 prenatal care provider locations and 59 maternity hospitals, 49 provider and 7 hospital locations within or just outside 3 counties participated in study recruitment. During first prenatal care visits or immediately postdelivery at these locations, face-to-face contact was used to screen and recruit eligible women. Of 1450 screened women, 1270 were eligible. Consent rates at prenatal provider locations (62%-74% by county) were similar to those at birth locations (64%-77% by county). During 6 field months, 3 study centers enrolled a total prenatal cohort of 530 women (the majority in the first trimester) and during 2 months enrolled a birth cohort of an additional 320 mother-newborn dyads. As personnel became experienced in the field, the time required to enroll a woman in the prenatal cohort declined from up to 200 hours to 50 to 100 hours per woman recruited. We demonstrated that PBS was feasible and operationally efficient in recruiting a representative cohort of newborns from 3 diverse US counties. Our findings suggest that PBS is a practical approach to recruit large pregnancy and birth cohorts across the United States.

Effect of Physical Violence on Sexually Transmitted Infections and Treatment Seeking Behaviour among Female Sex Workers in Thane District, Maharashtra, India.

BackgroundViolence against sex workers can heighten their vulnerability to HIV and other sexually transmitted infections (STIs). Evidence suggests the risk of acquiring STI/HIV infections among female sex workers (FSWs) who have experienced violence to be almost three-times higher than FSWs, who have not experienced violence. Moreover, an experience of physical and sexual violence makes it difficult for them to negotiate safer sex with their partners and often act as a barrier to utilization of prevention services.MethodsThis study utilizes data from 2785 FSWs aged 18 years and above who participated in a cross-sectional behavioural study conducted during 2013–14 in Thane district, Maharashtra. A probability-based two-stage cluster sampling method was used for data collection. This study assesses the effect of physical violence on self-reported STI symptoms (any STI and multiple STIs) and treatment seeking for the last STI symptom using propensity score matching method.ResultsAbout 18% of sampled FSWs reported physical violence at the time of the survey. The likelihood of experiencing such violence was significantly higher among FSWs who solicited clients at public places, engaged in other economic activities apart from sex work, had savings, and reported high client volume per week. FSWs experiencing violence were also inconsistent condom users while engaging in sex with regular partners and clients. The average adjusted effect of violence clearly depicted an increase in the risk of any STI (11%, p<0.05) and multiple STIs (8%, p<0.10) and reduction in treatment seeking (10%, p<0.05).ConclusionsThis study demonstrates a significant effect of physical violence on reporting of any STI symptom and treatment seeking. Findings call for the immediate inclusion of strategies aimed to address violence related challenges in HIV prevention program currently being provided at Thane district. Such strategies would further help in enhancing the access to tailored STI prevention and care services among FSWs in the district.

Features in Grocery Stores that Motivate Shoppers to Buy Healthier Foods, ConsumerStyles 2014.

We examined nine features in grocery stores shoppers reported motivated them to purchase more healthful foods in the past month. Features were compiled from common supermarket practices for each of the 4Ps of marketing: pricing, placement, promotion, and product. We examined percentages of the features overall and by shopping frequency using Chi square tests from a 2014 cross sectional web-based health attitudes and behaviors survey, ConsumerStyles. The survey was fielded from June to July in 2014. Participants were part of a market research consumer panel that were randomly recruited by probability-based sampling using address-based sampling methods to achieve a sample representative of the U.S. Data from 4242 adults ages 18 and older were analyzed. About 44% of respondents indicated at least one feature motivated them to purchase more healthful foods. Top choices included in-store coupons or specials (20.1%), availability of convenient, ready-to-eat more healthful foods (18.8%), product labels or advertising on packages (15.2%), and labels or signs on shelves that highlighted more healthful options (14.6%). Frequent shoppers reported being motivated to purchase more healthful foods by in-store tastings/recipe demonstrations and coupons/specials more often than infrequent shoppers. Enhancing the visibility and appeal of more healthful food items in grocery stores may help improve dietary choices in some populations but additional research is needed to identify the most effective strategies for interventions.

Mapping and estimating forest area and aboveground biomass in miombo woodlands in Tanzania using data from airborne laser scanning, TanDEM-X, RapidEye, and global forest maps: A comparison of estimated precision

Abstract Field surveys are often a primary source of data for aboveground biomass (AGB) and forest area estimates — two fundamental parameters in forest resource assessments and for measurement, reporting, and verification (MRV) under the United Nations Collaborative Program on Reducing Emissions from Deforestation and Forest Degradation in Developing Countries (REDD +). However, plot-based estimates of such parameters are often not sufficiently precise for their intended purposes, and especially so in developing and tropical countries in which implementation of extensive sample surveys can be cost-prohibitive or infeasible due to inaccessibility. Remotely sensed data can improve the precision of estimates and thereby reduce the need for field samples. To guide investment decision in MRV systems, comparative analyses of the contribution of different types of remotely sensed data to improve precision of estimates are required. The aim of the current study was to quantify the contribution of data from (1) airborne laser scanning (ALS), (2) interferometric synthetic aperture radar (InSAR) derived from TanDEM-X, (3) RapidEye optical imagery, and global forest map products derived from (4) Landsat and (5) ALOS PALSAR L-band radar imagery to improve precision of AGB and forest area estimates beyond the precision that could be obtained by a pure field-based survey in miombo woodlands of Tanzania. Miombo woodlands is one among the most wide-spread vegetation types in eastern, central, and southern Africa, occupying about 9% of the entire African land area. A 365.6 km2 region in Liwale district in Tanzania served as area of interest for this study. Eighty-eight ground plots distributed on 11 clusters of eight plots each according to a probability-based single-stage cluster sampling design served as field data for regression model calibration used for mapping and estimation of AGB and forest area. Model-assisted estimators were used in the estimation. The relative efficiency (RE) of the ALS-assisted estimates of mean AGB per hectare (variance of the field-based estimate relative to the variance of the ALS-assisted estimate) was 3.6. Relative efficiency translates directly to the factor by which the sample size used for the ALS-assisted estimate would have to be multiplied to arrive at the same precision for a pure field-based estimate. RE values for InSAR and RapidEye were 2.8 and 3.3, while the global Landsat and PALSAR map products contributed only marginally to improve precision (RE = 1.3–1.4). For forest area estimation, ALS-assisted estimates showed an RE of 3.7–4.6, while InSAR, RapidEye, and global Landsat and PALSAR maps resulted in RE values of 1.0–1.3, 2.0–2.1, 1.4–1.8, and 1.7, respectively.

Critical review of the United Kingdom's "gold standard" survey of public attitudes to science.

Since 2000, the UK government has funded surveys aimed at understanding the UK public's attitudes toward science, scientists, and science policy. Known as the Public Attitudes to Science series, these surveys and their predecessors have long been used in UK science communication policy, practice, and scholarship as a source of authoritative knowledge about science-related attitudes and behaviors. Given their importance and the significant public funding investment they represent, detailed academic scrutiny of the studies is needed. In this essay, we critically review the most recently published Public Attitudes to Science survey (2014), assessing the robustness of its methods and claims. The review casts doubt on the quality of key elements of the Public Attitudes to Science 2014 survey data and analysis while highlighting the importance of robust quantitative social research methodology. Our analysis comparing the main sample and booster sample for young people demonstrates that quota sampling cannot be assumed equivalent to probability-based sampling techniques.