Search for: All records

We provide systematic evidence on estimating household well-being from mobile phone data across four countries (Afghanistan, Côte d’Ivoire, Malawi, Togo). Using parallel, standardized machine learning experiments, we assess which welfare measures are most predictable and which data types most useful. Long-term poverty measures—wealth indices (Pearson’s ρ = 0.20–0.59) and multidimensional poverty (ρ = 0.29–0.57)—are predicted more accurately than consumption (ρ = 0.04–0.54); transient measures like food security are difficult to predict. Call and text behavior outperforms internet, mobile money, and airtime metadata. Nationally representative samples yield 20–70 percent higher accuracy than urban- or rural-only samples. 

Synopsis When instructors, curriculum developers, or education researchers implement an innovative lesson or develop a shareable lesson plan, they likely want to know if the lesson is effective at facilitating student learning. Published assessments, which have undergone a substantial development and optimization process, are commonly used to evaluate the impact of curricular and pedagogical innovations. However, there are some cases when using an existing concept assessment does not align with the goals of the person administering the assessment. Concept assessments usually cover a broad content domain spanning entire units, courses, or majors. If an instructor, curriculum developer, or researcher intends to assess learning associated with a discrete time period (e.g., a lesson), a concept assessment may be too broad and thus lack alignment validity. Consequently, they may decide to develop their own assessment. They can accomplish this by writing original questions, using question banks, using questions provided in open educational resources, or a combination of these approaches. Alternatively, they may pursue a time- and resource-intensive comprehensive assessment development process, typically involving several steps to accrue validity evidence. There are currently no guidelines for scaling the process for contexts in which existing concept assessments are misaligned and a comprehensive assessment development process is impractical. Here, we describe an adapted assessment development process to generate lesson-customized assessments. We started with five published biology lessons, covering a broad range of biology topics, and created a lesson-customized assessment for each. Our assessment development process consisted of identifying lesson learning objectives, drafting an initial assessment, soliciting expert feedback and student interviews, and conducting several cycles of review and revision. We provide suggestions to help with practicality and make the process more accessible. Though we developed customized assessments for published biology lessons, our approach is broadly applicable to other cases where customized assessments may be beneficial. 

Abstract Honeybees are master thermoregulators, capable of maintaining nest homeostasis across fluctuating ambient temperatures. When workers must cool their nest, they use multiple thermoregulatory behaviors (e.g., fanning, collecting water), but bearding, where hundreds to thousands of workers evacuate their nest and form a bivouac outside, is relatively unexplored. Here, we (1) describe natural bearding patterns, (2) experimentally manipulate colonies to determine what impacts beard size and timing, and (3) explore how workers dissipate back into their nest. We show that bearding occurs daily in hot weather, but the largest beards consistently happen in the evening/night, between the hours of 1800 and 2400. Beards are located around the nest entrance, but workers bias their position toward the shaded side of the nest box. As colony size increases, beard size and duration also increase, but the proportion of the colony bearding does not increase with colony size. Colonies with and without brood still cast beards; brood presence/absence did not impact beard size or duration. After noticing that beards tend to dissipate at sunrise, we experimentally showed that beards induced in the afternoon dissipate within 1–2 h, whereas beards induced in the evening remain overnight (10+ h). Bearding overnight, however, does carry risks for developing brood inside, as nest temperatures dropped below the optimal range, until the beard dissipated at sunrise. What cues workers use to depart the beard remain unknown, but experimentally illuminating colonies at night did not induce beards to dissipate. Our results suggest that bearding is an individual decision, not one that is coordinated across the colony. Still, these individual actions result in a dramatic collective response that colonies employ to reduce the temperature of their nest. Here, we show how and when colonies use bearding, despite its risks. 

Abstract Distributed acoustic sensing (DAS) is an emerging oceanographic technique in which an interrogator continuously records nanoscale strain of a fiber-optic cable, such as a telecommunication cable, with meter-scale measurement spacing over tens of kilometers. Empirical methods have recently been established for calculating pressure spectra to measure ocean surface gravity wave statistics from DAS strain. Here, we compile data from six submarine DAS experiments to provide a comparison between studies and establish recommendations for using DAS to measure ocean waves. Data were collected from Alaska, Hawaii, Massachusetts, North Carolina, and Oregon, United States, with different interrogators on different cable types in 0–60 m of water with 0–4 m of burial. Ground-truth measurements of ocean waves were provided by standard near-bed or sea surface instruments. The raw strain recorded in each experiment varied over four orders of magnitude, which could not be explained by water depth, wave conditions, or interrogator settings and suggests that cable characteristics and burial depth are important factors controlling strain magnitude and measurement quality. Strain spectra were converted to near-bed pressure spectra using a frequency-dependent, location-specific empirical correction factor, and DAS-derived pressure spectra were used to calculate wave statistics. The correction factors varied over 10 orders of magnitude between sites yet provided accurate calculations of wave height and period (root-mean-square error of 0.2–0.6 m forH_sand 0.2–1.6 s forT_eandT_p). The volume of data necessary for calibration is discussed. This meta-analysis highlights future oceanographic applications of DAS. Significance StatementDistributed acoustic sensing (DAS) is an emerging technology for measuring ocean waves on seafloor fiber-optic cables, such as telecom cables. The advantage of DAS is that it can record thousands of measurements per second at meter-scale spacing over tens of kilometers. We compare six datasets to characterize DAS-derived strain for measuring ocean waves. The differences in strain magnitude observed between datasets were not explained by water depth, wave height or period, or instrument settings. It is likely that cable composition and depth of burial control the magnitude of the recorded strain. Despite these differences, each dataset was empirically calibrated to produce accurate measurements of wave statistics. DAS is a promising new oceanographic technology, and new applications should be explored.