Access to data is a critical feature of an efficient, progressive and ultimately self-correcting scientific ecosystem. But the extent to which in-principle benefits of data sharing are realized in practice is unclear. Crucially, it is largely unknown whether published findings can be reproduced by repeating reported analyses upon shared data (‘analytic reproducibility’). To investigate this, we conducted an observational evaluation of a mandatory open data policy introduced at the journal Cognition. Interrupted time-series analyses indicated a substantial post-policy increase in data available statements (104/417, 25% pre-policy to 136/174, 78% post-policy), although not all data appeared reusable (23/104, 22% pre-policy to 85/136, 62%, post-policy). For 35 of the articles determined to have reusable data, we attempted to reproduce 1324 target values. Ultimately, 64 values could not be reproduced within a 10% margin of error. For 22 articles all target values were reproduced, but 11 of these required author assistance. For 13 articles at least one value could not be reproduced despite author assistance. Importantly, there were no clear indications that original conclusions were seriously impacted. Mandatory open data policies can increase the frequency and quality of data sharing. However, suboptimal data curation, unclear analysis specification and reporting errors can impede analytic reproducibility, undermining the utility of data sharing and the credibility of scientific findings.

28Access to data is a critical feature of an efficient, progressive, and ultimately self-correcting 29 scientific ecosystem. But the extent to which in-principle benefits of data sharing are realized 30 in practice is unclear. Crucially, it is largely unknown whether published findings can be 31 reproduced by repeating reported analyses upon shared data ("analytic reproducibility"). To by poor access to research data [7][8][9][10][11][12][13][14]. Furthermore, even when data are shared, inadequate 57 documentation and formatting can render them unusable [10]. Thus, whilst data sharing has 58 many benefits in principle, the extent to which they are being realized in practice is unclear. 59Crucially, whether data access enables independent verification of analytic reproducibility is 60 largely unknown. 61Any investigation of data sharing utility faces an immediate impediment: research data 62 are typically not available. The policies of journals and professional societies, such as the 63The American Psychological Association, often fall short of imposing mandatory data 64 sharing requirements on researchers, and merely recommend that data be "available upon 65 request", if they make any recommendation at all [7,15]. In the absence of stringent 66 community norms or regulations, scientific claims are regularly published without public 67 release of the research data upon which they are based [7,9,14]. Post-publication efforts to 68 obtain data directly from authors frequently go unanswered, or are refused [11][12][13] 87Even when data are available, and in-principle reusable, the extent to which they In the present investigation, we sought to examine the state of data availability, 99 reusability, and analytic reproducibility within a sub-field of psychology. We capitalized on should be shared in a form that enables reuse and analytic reproducibility [24]. 140We are not aware of any pertinent co-interventions occurring during the assessment 2 Note that by "submitted" we technically mean that the article was formally logged at the journal following author submission. Cognition refers to this as the "received date" in article headers. 197In order to estimate the causal effect of the policy independent of any contemporary however, when time is the variable determining the point of discontinuity, the more commonly used terminology is "interrupted time series", which we employ here. Also note that we did not pre-specify the details of this analysis, such as the exact model specification. 4 We visually inspected the density of DAS inclusion over time and find no evidence that articles were preferentially submitted just before the deadline. 5 This specification appeared to model the secular trend adequately because the resulting residuals showed no evidence of autocorrelation. We attempted to fit alternative model specifications that would directly model the probability of DAS inclusion (i.e., a linear probability model) or its risk ratio (i.e., a log-linear model), either of which would obviate conversio...

No abstract

The problem of automatic identification of physical activities performed by human subjects is referred to as Human Activity Recognition (HAR). There exist several techniques to measure motion characteristics during these physical activities, such as Inertial Measurement Units (IMUs). IMUs have a cornerstone position in this context, and are characterized by usage flexibility, low cost, and reduced privacy impact. With the use of inertial sensors, it is possible to sample some measures such as acceleration and angular velocity of a body, and use them to learn models that are capable of correctly classifying activities to their corresponding classes. In this paper, we propose to use Convolutional Neural Networks (CNNs) to classify human activities. Our models use raw data obtained from a set of inertial sensors. We explore several combinations of activities and sensors, showing how motion signals can be adapted to be fed into CNNs by using different network architectures. We also compare the performance of different groups of sensors, investigating the classification potential of single, double and triple sensor systems. The experimental results obtained on a dataset of 16 lowerlimb activities, collected from a group of participants with the use of five different sensors, are very promising.