Agreement of research results repeated. Reproducibility, replicability, repeatability, robustness, generalizability, organization, documentation, automation, dissemination, guidance, definitions, and more.
Central location housing curriculum materials and planning tools for trainers of the COS Introduction to Open and Reproducible Research workshop.
Several fields of science are experiencing a "replication crisis" that has negatively impacted their credibility. Assessing the validity of a contribution via replicability of its experimental evidence and reproducibility of its analyses requires access to relevant study materials, data, and code. Failing to share them limits the ability to scrutinize or build-upon the research, ultimately hindering scientific progress.Understanding how the diverse research artifacts in HCI impact sharing can help produce informed recommendations for individual researchers and policy-makers in HCI. Therefore, we surveyed authors of CHI 2018–2019 papers, asking if they share their papers' research materials and data, how they share them, and why they do not. The results (N = 460/1356, 34% response rate) show that sharing is uncommon, partly due to misunderstandings about the purpose of sharing and reliable hosting. We conclude with recommendations for fostering open research practices.This paper and all data and materials are freely available at https://osf.io/csy8q
Background: Reproducible research is a foundational component for scientific advancements, yet little is known regarding the extent of reproducible research within the dermatology literature. Objective: This study aimed to determine the quality and transparency of the literature in dermatology journals by evaluating for the presence of 8 indicators of reproducible and transparent research practices. Methods: By implementing a cross-sectional study design, we conducted an advanced search of publications in dermatology journals from the National Library of Medicine catalog. Our search included articles published between January 1, 2014, and December 31, 2018. After generating a list of eligible dermatology publications, we then searched for full text PDF versions by using Open Access Button, Google Scholar, and PubMed. Publications were analyzed for 8 indicators of reproducibility and transparency—availability of materials, data, analysis scripts, protocol, preregistration, conflict of interest statement, funding statement, and open access—using a pilot-tested Google Form. Results: After exclusion, 127 studies with empirical data were included in our analysis. Certain indicators were more poorly reported than others. We found that most publications (113, 88.9%) did not provide unmodified, raw data used to make computations, 124 (97.6%) failed to make the complete protocol available, and 126 (99.2%) did not include step-by-step analysis scripts. Conclusions: Our sample of studies published in dermatology journals do not appear to include sufficient detail to be accurately and successfully reproduced in their entirety. Solutions to increase the quality, reproducibility, and transparency of dermatology research are warranted. More robust reporting of key methodological details, open data sharing, and stricter standards journals impose on authors regarding disclosure of study materials might help to better the climate of reproducible research in dermatology. [JMIR Dermatol 2019;2(1):e16078]
The UKRN primer series is designed to introduce a broad audience to important topics in open and reproducible scholarship. Each primer includes an overview of the topic in the introductory “What?” section, reasons for undertaking these practices in the “Why?” section, followed by a longer “How?” section that provides guidance on how to do that open research behaviour practically. Throughout the primers there are embedded explanatory weblinks, and at the end of each is a collated list of links to useful further resources.
Software Carpentry lesson on how to use the shell to navigate the filesystem and write simple loops and scripts. The Unix shell has been around longer than most of its users have been alive. It has survived so long because it’s a power tool that allows people to do complex things with just a few keystrokes. More importantly, it helps them combine existing programs in new ways and automate repetitive tasks so they aren’t typing the same things over and over again. Use of the shell is fundamental to using a wide range of other powerful tools and computing resources (including “high-performance computing†supercomputers). These lessons will start you on a path towards using these resources effectively.
This lesson is part of the Software Carpentry workshops that teach how to use version control with Git. Wolfman and Dracula have been hired by Universal Missions (a space services spinoff from Euphoric State University) to investigate if it is possible to send their next planetary lander to Mars. They want to be able to work on the plans at the same time, but they have run into problems doing this in the past. If they take turns, each one will spend a lot of time waiting for the other to finish, but if they work on their own copies and email changes back and forth things will be lost, overwritten, or duplicated. A colleague suggests using version control to manage their work. Version control is better than mailing files back and forth: Nothing that is committed to version control is ever lost, unless you work really, really hard at it. Since all old versions of files are saved, it’s always possible to go back in time to see exactly who wrote what on a particular day, or what version of a program was used to generate a particular set of results. As we have this record of who made what changes when, we know who to ask if we have questions later on, and, if needed, revert to a previous version, much like the “undo†feature in an editor. When several people collaborate in the same project, it’s possible to accidentally overlook or overwrite someone’s changes. The version control system automatically notifies users whenever there’s a conflict between one person’s work and another’s. Teams are not the only ones to benefit from version control: lone researchers can benefit immensely. Keeping a record of what was changed, when, and why is extremely useful for all researchers if they ever need to come back to the project later on (e.g., a year later, when memory has faded). Version control is the lab notebook of the digital world: it’s what professionals use to keep track of what they’ve done and to collaborate with other people. Every large software development project relies on it, and most programmers use it for their small jobs as well. And it isn’t just for software: books, papers, small data sets, and anything that changes over time or needs to be shared can and should be stored in a version control system.
This webinar will introduce the concept of version control and the version control features that are built into the Open Science Framework (OSF; https://osf.io). The OSF is a free, open source web application built to help researchers manage their workflows. The OSF is part collaboration tool, part version control software, and part data archive. The OSF connects to popular tools researchers already use, like Dropbox, Box, Github and Mendeley, to streamline workflows and increase efficiency. This webinar will discuss how keeping track of the different file versions is important for efficient reproducible research practices, how version control works on the OSF, and how researchers can view and download previous versions of files.
In this review, the author discusses several of the weak spots in contemporary science, including scientific misconduct, the problems of post hoc hypothesizing (HARKing), outcome switching, theoretical bloopers in formulating research questions and hypotheses, selective reading of the literature, selective citing of previous results, improper blinding and other design failures, p-hacking or researchers’ tendency to analyze data in many different ways to find positive (typically significant) results, errors and biases in the reporting of results, and publication bias. The author presents some empirical results highlighting problems that lower the trustworthiness of reported results in scientific literatures, including that of animal welfare studies. Some of the underlying causes of these biases are discussed based on the notion that researchers are only human and hence are not immune to confirmation bias, hindsight bias, and minor ethical transgressions. The author discusses solutions in the form of enhanced transparency, sharing of data and materials, (post-publication) peer review, pre-registration, registered reports, improved training, reporting guidelines, replication, dealing with publication bias, alternative inferential techniques, power, and other statistical tools.
More researchers are preregistering their studies as a way to combat publication bias and improve the credibility of research findings. Preregistration is at its core designed to distinguish between confirmatory and exploratory results. Both are important to the progress of science, but when they are conflated, problems arise. In this webinar, we discuss the What, Why, and How of preregistration and what it means for the future of science. Visit cos.io/prereg for additional resources.
Replications are inevitably different from the original studies. How do we decide whether something is a replication? The answer shifts the conception of replication from a boring, uncreative, housekeeping activity to an exciting, generative, vital contributor to research progress.
This video is the first in a series of videos related to the basics of power analyses. All materials shown in the video, as well as content from the other videos in the power analysis series can be found here: https://osf.io/a4xhr/
A general introduction to open scholarship.
Badges are a great way to signal that a journal values transparent research practices. Readers see the papers that have underlying data or methods available, colleagues see that norms are changing within a community and have ample opportunities to emulate better practices, and authors get recognition for taking a step into new techniques. In this webinar, Professor Stephen Lindsay of University of Victoria discusses the workflow of a badging program, eligibility for badge issuance, and the pitfalls to avoid in launching a badging program. Visit cos.io/badges to learn more.
Reproducibility is unquestionably at the heart of science. Scientists face numerous challenges in this context, not least the lack of concepts, tools, and workflows for reproducible research in today's curricula.This short course introduces established and powerful tools that enable reproducibility of computational geoscientific research, statistical analyses, and visualisation of results using R (http://www.r-project.org/) in two lessons:1. Reproducible Research with R MarkdownOpen Data, Open Source, Open Reviews and Open Science are important aspects of science today. In the first lesson, basic motivations and concepts for reproducible research touching on these topics are briefly introduced. During a hands-on session the course participants write R Markdown (http://rmarkdown.rstudio.com/) documents, which include text and code and can be compiled to static documents (e.g. HTML, PDF).R Markdown is equally well suited for day-to-day digital notebooks as it is for scientific publications when using publisher templates.2. GitLab and DockerIn the second lesson, the R Markdown files are published and enriched on an online collaboration platform. Participants learn how to save and version documents using GitLab (http://gitlab.com/) and compile them using Docker containers (https://docker.com/). These containers capture the full computational environment and can be transported, executed, examined, shared and archived. Furthermore, GitLab's collaboration features are explored as an environment for Open Science.Prerequisites: Participants should install required software (R, RStudio, a current browser) and register on GitLab (https://gitlab.com) before the course.This short course is especially relevant for early career scientists (ECS).Participants are welcome to bring their own data and R scripts to work with during the course.All material by the conveners will be shared publicly via OSF (https://osf.io/qd9nf/).
Irreproducibility of preclinical biomedical research has gained recent attention. It is suggested that requiring authors to complete a checklist at the time of manuscript submission would improve the quality and transparency of scientific reporting, and ultimately enhance reproducibility. Whether a checklist enhances quality and transparency in reporting preclinical animal studies, however, has not been empirically studied. Here we searched two highly cited life science journals, one that requires a checklist at submission (Nature) and one that does not (Cell), to identify in vivo animal studies. After screening 943 articles, a total of 80 articles were identified in 2013 (pre-checklist) and 2015 (post-checklist), and included for the detailed evaluation of reporting methodological and analytical information. We compared the quality of reporting preclinical animal studies between the two journals, accounting for differences between journals and changes over time in reporting. We find that reporting of randomization, blinding, and sample-size estimation significantly improved when comparing Nature to Cell from 2013 to 2015, likely due to implementation of a checklist. Specifically, improvement in reporting of the three methodological information was at least three times greater when a mandatory checklist was implemented than when it was not. Reporting the sex of animals and the number of independent experiments performed also improved from 2013 to 2015, likely from factors not related to a checklist. Our study demonstrates that completing a checklist at manuscript submission is associated with improved reporting of key methodological information in preclinical animal studies.
Efforts to make research results open and reproducible are increasingly reflected by journal policies encouraging or mandating authors to provide data availability statements. As a consequence of this, there has been a strong uptake of data availability statements in recent literature. Nevertheless, it is still unclear what proportion of these statements actually contain well-formed links to data, for example via a URL or permanent identifier, and if there is an added value in providing them. We consider 531,889 journal articles published by PLOS and BMC which are part of the PubMed Open Access collection, categorize their data availability statements according to their content and analyze the citation advantage of different statement categories via regression. We find that, following mandated publisher policies, data availability statements have become common by now, yet statements containing a link to a repository are still just a fraction of the total. We also find that articles with these statements, in particular, can have up to 25.36% higher citation impact on average: an encouraging result for all publishers and authors who make the effort of sharing their data. All our data and code are made available in order to reproduce and extend our results.
The widespread use of ‘statistical significance’ as a license for making a claim of a scientific finding leads to considerable distortion of the scientific process (according to the American Statistical Association). We review why degrading p-values into ‘significant’ and ‘nonsignificant’ contributes to making studies irreproducible, or to making them seem irreproducible. A major problem is that we tend to take small p-values at face value, but mistrust results with larger p-values. In either case, p-values tell little about reliability of research, because they are hardly replicable even if an alternative hypothesis is true. Also significance (p ≤ 0.05) is hardly replicable: at a good statistical power of 80%, two studies will be ‘conflicting’, meaning that one is significant and the other is not, in one third of the cases if there is a true effect. A replication can therefore not be interpreted as having failed only because it is nonsignificant. Many apparent replication failures may thus reflect faulty judgment based on significance thresholds rather than a crisis of unreplicable research. Reliable conclusions on replicability and practical importance of a finding can only be drawn using cumulative evidence from multiple independent studies. However, applying significance thresholds makes cumulative knowledge unreliable. One reason is that with anything but ideal statistical power, significant effect sizes will be biased upwards. Interpreting inflated significant results while ignoring nonsignificant results will thus lead to wrong conclusions. But current incentives to hunt for significance lead to selective reporting and to publication bias against nonsignificant findings. Data dredging, p-hacking, and publication bias should be addressed by removing fixed significance thresholds. Consistent with the recommendations of the late Ronald Fisher, p-values should be interpreted as graded measures of the strength of evidence against the null hypothesis. Also larger p-values offer some evidence against the null hypothesis, and they cannot be interpreted as supporting the null hypothesis, falsely concluding that ‘there is no effect’. Information on possible true effect sizes that are compatible with the data must be obtained from the point estimate, e.g., from a sample average, and from the interval estimate, such as a confidence interval. We review how confusion about interpretation of larger p-values can be traced back to historical disputes among the founders of modern statistics. We further discuss potential arguments against removing significance thresholds, for example that decision rules should rather be more stringent, that sample sizes could decrease, or that p-values should better be completely abandoned. We conclude that whatever method of statistical inference we use, dichotomous threshold thinking must give way to non-automated informed judgment.
Improving the reliability and efficiency of scientific research will increase the credibility of the published scientific literature and accelerate discovery. Here we argue for the adoption of measures to optimize key elements of the scientific process: methods, reporting and dissemination, reproducibility, evaluation and incentives. There is some evidence from both simulations and empirical studies supporting the likely effectiveness of these measures, but their broad adoption by researchers, institutions, funders and journals will require iterative evaluation and improvement. We discuss the goals of these measures, and how they can be implemented, in the hope that this will facilitate action toward improving the transparency, reproducibility and efficiency of scientific research.
Poor research design and data analysis encourage false-positive findings. Such poor methods persist despite perennial calls for improvement, suggesting that they result from something more than just misunderstanding. The persistence of poor methods results partly from incentives that favour them, leading to the natural selection of bad science. This dynamic requires no conscious strategizing—no deliberate cheating nor loafing—by scientists, only that publication is a principal factor for career advancement. Some normative methods of analysis have almost certainly been selected to further publication instead of discovery. In order to improve the culture of science, a shift must be made away from correcting misunderstandings and towards rewarding understanding. We support this argument with empirical evidence and computational modelling. We first present a 60-year meta-analysis of statistical power in the behavioural sciences and show that power has not improved despite repeated demonstrations of the necessity of increasing power. To demonstrate the logical consequences of structural incentives, we then present a dynamic model of scientific communities in which competing laboratories investigate novel or previously published hypotheses using culturally transmitted research methods. As in the real world, successful labs produce more ‘progeny,’ such that their methods are more often copied and their students are more likely to start labs of their own. Selection for high output leads to poorer methods and increasingly high false discovery rates. We additionally show that replication slows but does not stop the process of methodological deterioration. Improving the quality of research requires change at the institutional level.
It is widely believed that research that builds upon previously published findings has reproduced the original work. However, it is rare for researchers to perform or publish direct replications of existing results. The Reproducibility Project: Cancer Biology is an open investigation of reproducibility in preclinical cancer biology research. We have identified 50 high impact cancer biology articles published in the period 2010-2012, and plan to replicate a subset of experimental results from each article. A Registered Report detailing the proposed experimental designs and protocols for each subset of experiments will be peer reviewed and published prior to data collection. The results of these experiments will then be published in a Replication Study. The resulting open methodology and dataset will provide evidence about the reproducibility of high-impact results, and an opportunity to identify predictors of reproducibility.