Improving Openness and Reproducibility of Scientific Research
Norms Counternorms Communality Universalism Disinterestedness Open sharing Universalism Evaluate research on own merit Disinterestedness Motivated by knowledge and discovery Organized skepticism Consider all new evidence, even against one’s prior work Quality Secrecy Closed Particularlism Evaluate research by reputation Self-interestedness Treat science as a competition Organized dogmatism Invest career promoting one’s own theories, findings Quantity Communality – open sharing with colleagues; Secrecy Universalism – research evaluated only on its merit; Particularism – research evaluated by reputation/past productivity Disinterestedness – scientists motivated by knowledge and discovery, not by personal gain; self-interestedness – treat science as a competition with other scientists Organized skepticism – consider all new evidence, theory, data, even if it contradicts one’s prior work/point-of-view; organized dogmatism – invest career in promoting one’s own most important findings, theories, innovations Quality – seek quality contributions; Quantity – seek high volume Merton, 1942
Norms Counternorms Communality Universalism Disinterestedness Open sharing Universalism Evaluate research on own merit Disinterestedness Motivated by knowledge and discovery Organized skepticism Consider all new evidence, even against one’s prior work Quality Secrecy Closed Particularlism Evaluate research by reputation Self-interestedness Treat science as a competition Organized dogmatism Invest career promoting one’s own theories, findings Quantity Communality – open sharing with colleagues; Secrecy Universalism – research evaluated only on its merit; Particularism – research evaluated by reputation/past productivity Disinterestedness – scientists motivated by knowledge and discovery, not by personal gain; self-interestedness – treat science as a competition with other scientists Organized skepticism – consider all new evidence, theory, data, even if it contradicts one’s prior work/point-of-view; organized dogmatism – invest career in promoting one’s own most important findings, theories, innovations Quality – seek quality contributions; Quantity – seek high volume Merton, 1942
Norms Counternorms Communality Universalism Disinterestedness Open sharing Universalism Evaluate research on own merit Disinterestedness Motivated by knowledge and discovery Organized skepticism Consider all new evidence, even against one’s prior work Quality Secrecy Closed Particularlism Evaluate research by reputation Self-interestedness Treat science as a competition Organized dogmatism Invest career promoting one’s own theories, findings Quantity Communality – open sharing with colleagues; Secrecy Universalism – research evaluated only on its merit; Particularism – research evaluated by reputation/past productivity Disinterestedness – scientists motivated by knowledge and discovery, not by personal gain; self-interestedness – treat science as a competition with other scientists Organized skepticism – consider all new evidence, theory, data, even if it contradicts one’s prior work/point-of-view; organized dogmatism – invest career in promoting one’s own most important findings, theories, innovations Quality – seek quality contributions; Quantity – seek high volume Merton, 1942
Norms Counternorms Communality Universalism Disinterestedness Open sharing Universalism Evaluate research on own merit Disinterestedness Motivated by knowledge and discovery Organized skepticism Consider all new evidence, even against one’s prior work Quality Secrecy Closed Particularlism Evaluate research by reputation Self-interestedness Treat science as a competition Organized dogmatism Invest career promoting one’s own theories, findings Quantity Communality – open sharing with colleagues; Secrecy Universalism – research evaluated only on its merit; Particularism – research evaluated by reputation/past productivity Disinterestedness – scientists motivated by knowledge and discovery, not by personal gain; self-interestedness – treat science as a competition with other scientists Organized skepticism – consider all new evidence, theory, data, even if it contradicts one’s prior work/point-of-view; organized dogmatism – invest career in promoting one’s own most important findings, theories, innovations Quality – seek quality contributions; Quantity – seek high volume Merton, 1942
Norms Counternorms Communality Universalism Disinterestedness Open sharing Universalism Evaluate research on own merit Disinterestedness Motivated by knowledge and discovery Organized skepticism Consider all new evidence, even against one’s prior work Quality Secrecy Closed Particularlism Evaluate research by reputation Self-interestedness Treat science as a competition Organized dogmatism Invest career promoting one’s own theories, findings Quantity Communality – open sharing with colleagues; Secrecy Universalism – research evaluated only on its merit; Particularism – research evaluated by reputation/past productivity Disinterestedness – scientists motivated by knowledge and discovery, not by personal gain; self-interestedness – treat science as a competition with other scientists Organized skepticism – consider all new evidence, theory, data, even if it contradicts one’s prior work/point-of-view; organized dogmatism – invest career in promoting one’s own most important findings, theories, innovations Quality – seek quality contributions; Quantity – seek high volume Merton, 1942
Anderson, Martinson, & DeVries, 2007
Anderson, Martinson, & DeVries, 2007
Anderson, Martinson, & DeVries, 2007
Incentives for individual success are focused on getting it published, not getting it right Nosek, Spies, & Motyl, 2012
Flexibility in analysis Problems Flexibility in analysis Selective reporting Ignoring nulls Lack of replication Examples from: Button et al – Neuroscience Ioannidis – why most results are false (Medicine) GWAS Biology Two possibilities are that the percentage of positive results is inflated because negative results are much less likely to be published, and that we are pursuing our analysis freedoms to produce positive results that are not really there. These would lead to an inflation of false-positive results in the published literature. Some evidence from bio-medical research suggests that this is occurring. Two different industrial laboratories attempted to replicate 40 or 50 basic science studies that showed positive evidence for markers for new cancer treatments or other issues in medicine. They did not select at random. Instead, they picked studies considered landmark findings. The success rates for replication were about 25% in one study and about 10% in the other. Further, some of the findings they could not replicate had spurred large literatures of hundreds of articles following up on the finding and its implications, but never having tested whether the evidence for the original finding was solid. This is a massive waste of resources. Across the sciences, evidence like this has spurred lots of discussion and proposed actions to improve research efficiency and avoid the massive waste of resources linked to erroneous results getting in and staying in the literature, and about the culture of scientific practices that is rewarding publishing, perhaps at the expense of knowledge building. There have been a variety of suggestions for what to do. For example, the Nature article on the right suggests that publishing standards should be increased for basic science research. [It is not in my interest to replicate – myself or others – to evaluate validity and improve precision in effect estimates (redundant). Replication is worth next to zero (Makel data on published replications; motivated to not call it replication; novelty is supreme – zero “error checking”; not in my interest to check my work, and not in your interest to check my work (let’s just each do our own thing and get rewarded for that) Irreproducible results will get in and stay in the literature (examples from bio-med). Prinz and Begley articles (make sure to summarize accurately) The Nature article by folks in bio-medicine is great. The solution they offer is a popular one in commentators from the other sciences -- raise publishing standards. Sterling, 1959; Cohen, 1962; Lykken, 1968; Tukey, 1969; Greenwald, 1975; Meehl, 1978; Rosenthal, 1979
Figure by FiveThirtyEight.com Silberzahn et al., 2015
A Garden of Forking Paths “Does X affect Y?” Exclude outliers? Control for year? Median or mean? A series of perfectly reasonable decisions coupled with motivated reasoning can quickly lead us to a subset of statistically significant results. In effect, our hypothesis changed without us even realizing it. Our confirmatory, hypothesis testing became exploratory, hypothesis generating without our permission! Jorge Luis Borges; Gelman and Loken
Franco, Malhotra, & Simonovits, 2015, SPPS We find that about 40% of studies fail to fully report all experimental conditions and about 70% of studies do not report all outcome variables included in the questionnaire. Reported effect sizes are about twice as large as unreported effect sizes and are about 3 times more likely to be statistically significant. N = 32 studies in psychology Unreported tests (147) Median p-value = .35 Median d = .13 % significant = 23% Reported tests (N = 122) Median p = .02 Median d = .29 % sig p<.05 = 63% Franco, Malhotra, & Simonovits, 2015, SPPS
Median effect size (d) = .29 % p < .05 = 63% Reported Tests (122) Median p-value = .02 Median effect size (d) = .29 % p < .05 = 63% We find that about 40% of studies fail to fully report all experimental conditions and about 70% of studies do not report all outcome variables included in the questionnaire. Reported effect sizes are about twice as large as unreported effect sizes and are about 3 times more likely to be statistically significant. N = 32 studies in psychology Unreported tests (147) Median p-value = .35 Median d = .13 % significant = 23% Reported tests (N = 122) Median p = .02 Median d = .29 % sig p<.05 = 63% Franco, Malhotra, & Simonovits, 2015, SPPS
Median effect size (d) = .29 % p < .05 = 63% Reported Tests (122) Median p-value = .02 Median effect size (d) = .29 % p < .05 = 63% Unreported Tests (147) Median p-value = .35 Median effect size (d) = .13 % p < .05 = 23% We find that about 40% of studies fail to fully report all experimental conditions and about 70% of studies do not report all outcome variables included in the questionnaire. Reported effect sizes are about twice as large as unreported effect sizes and are about 3 times more likely to be statistically significant. N = 32 studies in psychology Unreported tests (147) Median p-value = .35 Median d = .13 % significant = 23% Reported tests (N = 122) Median p = .02 Median d = .29 % sig p<.05 = 63% Franco, Malhotra, & Simonovits, 2015, SPPS
Estimating Reproducibility Increasing Depth Increasing Breadth
Reproducibility in other fields Developmental Psychology Michael Frank Ecology Emilio Bruna Health Sciences Computer Science Leslie McIntosh Cynthia Hudson-Vitale Christian Collberg Todd Proebsting
Solution? Appeal to intentions, values, and goals. “Hey You! Behave by your values! Be objective!”
Incentives for individual success are focused on getting it published, not getting it right Nosek, Spies, & Motyl, 2012
Rewards What is published? What is not? Replications Novel results Negative results Mixed evidence Novel results Positive results Clean results
Evidence to encourage change Technology to enable change Training to enact change Incentives to embrace change Improving scientific ecosystem
Infrastructure Metascience Community
Infrastructure
Technology to enable change Supporting these behavioral changes requires improving the full scientific ecosystem. At a conference like IDCC, there are many people in the room contributing important parts to this ecosystem. I hope you leave this talk seeing the potential for how we might be able to work together on connecting tools to provide for better transparency and reproducibility in the workflow.
Collaboration Documentation Archiving Content management and collaboration system Free service Connect, curate, search all aspects of the research project We don’t want to repeat ourselves Dataverse S3 Figshare Dropbox Service vs Application interface the service build new applicatoins this is right in line with SHARE-NS - openning, unlcoking this data allows for innovation From day one, we’ve been very excited about the SHARE partnerhship: community, expertise bringing to the table as well as shared interests mission Technical perspective, this project fit very much in line with what are building and more importantly how we are building it
Content management and collaboration system Free service Connect, curate, search all aspects of the research project We don’t want to repeat ourselves Dataverse S3 Figshare Dropbox Service vs Application interface the service build new applicatoins this is right in line with SHARE-NS - openning, unlcoking this data allows for innovation From day one, we’ve been very excited about the SHARE partnerhship: community, expertise bringing to the table as well as shared interests mission Technical perspective, this project fit very much in line with what are building and more importantly how we are building it
Version Control
Merges Public-Private Workflows
Incentives for Openness File downloads We can do small things like offer rich analytics to incentivize more open practices. Public projects gain immediate access to analytics showing visits over time, sources of traffic, geographic location of traffic, time of traffic, and download counts for files. This is a much faster reward for one’s effort than waiting months or longer until something is published and then even longer until it is cited.
File downloads We can do small things like offer rich analytics to incentivize more open practices. Public projects gain immediate access to analytics showing visits over time, sources of traffic, geographic location of traffic, time of traffic, and download counts for files. This is a much faster reward for one’s effort than waiting months or longer until something is published and then even longer until it is cited. Forks
Persistent Citable Identifiers
Registration
Connecting the workflow is critical to enabling change
Search and discover Publish report Write report Develop idea Interpret findings Design study Analyze Data Acquire materials There’s more to it than sharing of discrete objects. Think about using this as an opportunity to increase transparency by capturing the entire workflow, and to do so while connecting tools and services that make up the parts of the workflow, not requiring people to change all of their practices at once, and providing immediate efficiencies and value to the researcher AS they comply with requirements. Easy, right? Obviously not. Store data Collect data
Search and discover Publish report Write report Develop idea Interpret findings Design study Analyze Data Acquire materials There’s more to it than sharing of discrete objects. Think about using this as an opportunity to increase transparency by capturing the entire workflow, and to do so while connecting tools and services that make up the parts of the workflow, not requiring people to change all of their practices at once, and providing immediate efficiencies and value to the researcher AS they comply with requirements. Easy, right? Obviously not. Store data Collect data OpenSesame
Community
Training to enact change Once infrastructure is in place, we need to show researchers how to use it to improve their practices.
Free training on how to make research more reproducible Partner with others on training --- librarians are great partners in this ---- to teach researchers skills in how to deal with basic data management and how to improve their research workflows for personal and sharing purposes. Software Carpentry and Data Carpentry are other great examples of efforts in this area, and partnerships with those in libraries --- we’ve done some work with them and are exploring ways to do more. Free training on how to make research more reproducible http://cos.io/stats_consulting
Incentives to embrace change Supporting these behavioral changes requires improving the full scientific ecosystem. At a conference like IDCC, there are many people in the room contributing important parts to this ecosystem. I hope you leave this talk seeing the potential for how we might be able to work together on connecting tools to provide for better transparency and reproducibility in the workflow.
Transparency & Openness Promotion Guidelines Agnostic to discipline Low barrier to entry Modular
Transparency & Openness Promotion Guidelines Eight Standards Data Citation Design transparency Research materials Data Analytical methods Preregistered studies Preregistered analysis plans Registered Reports
Transparency & Openness Promotion Guidelines Eight Standards Data Citation Design transparency Research materials Data Analytical methods Preregistered studies Preregistered analysis plans Registered Reports Three Tiers Disclose Require Verify
Transparency & Openness Promotion Guidelines Eight Standards Data Citation Design transparency Research materials Data Analytical methods Preregistered studies Preregistered analysis plans Registered Reports Three Tiers Disclose Require Verify
Transparency & Openness Promotion Guidelines Eight Standards Data Citation Design transparency Research materials Data Analytical methods Preregistered studies Preregistered analysis plans Registered Reports Three Tiers Disclose Require Verify
Transparency & Openness Promotion Guidelines Signatories 539 Journal signatories 59 Organizational signatories Learn more at http://cos.io/top
Signals: Making Behaviors Visible Promotes Adoption Badges Open Data Open Materials Preregistration Psychological Science (Jan 2014)
40% 30% % Articles reporting that data was available 20% 10% 0%
100% 75% 50% 25% 0% % of Articles reporting that data was available On the y axis, we have % of articles reporting data available in an independent, open access location On the x axis, we have five categories: reportedly available, accessible, correct, usable, and complete data 25% 0% Reportedly Available Accessible Correct Data Usable Data Complete Data
100% 75% 50% 25% 0% % of Articles reporting that data was available In an ideal world, we’d see straight lines for articles published in all journals, and in psychological science before and after badges This would mean that all articles that reported available data had data that was accessible, correct, usable, and complete 25% 0% Reportedly Available Accessible Correct Data Usable Data Complete Data
100% 75% 50% 25% 0% % of Articles reporting that data was available In Psychological Science prior to badges and in other comparison conditions, only 39% or less (20%, 16%) of articles with reportedly available data in the comparison conditions had data that was accessible, correct, usable, and complete 25% 0% Reportedly Available Accessible Correct Data Usable Data Complete Data
100% 75% 50% 25% 0% % of Articles reporting that data was available Of articles in Psychological Science that earned badges, over 75% of articles with data reportedly available had data that was accessible, correct, usable, and complete While not perfect, reportedly available data were more likely to be persistent when badges were earned -- perhaps due to the accountability of receiving a badge on the publication. 25% 0% Reportedly Available Accessible Correct Data Usable Data Complete Data
The $1,000,000 Preregistration Challenge Another incentive for researchers to try out preregistration.
Exploratory research: Finds unexpected trends Pushes knowledge into new areas Results in a testable hypothesis
Confirmatory research: Puts a hypothesis to the test Does not allow data to influence the hypothesis Results are held to the highest standard of rigor
https://cos.io/prereg
Data collection methods Research questions Data collection methods Variables Statistical tests Outliers
Likelihood of Null Effects of Large NHLBI Clinical Trials Has Increased over Time Positive result rate dropped from 57% to 8% after preregistration became required for clinical trials. Kaplan and Irvin, 2015
Registered Reports Design Collect & Analyze Report Publish PEER REVIEW Review of intro and methods prior to data collection; published regardless of outcome Beauty vs. accuracy of reporting Publishing negative results Conducting replications Peer review focuses on quality of methods
Registered Reports Design Collect & Analyze Report Publish PEER REVIEW Review of intro and methods prior to data collection; published regardless of outcome Beauty vs. accuracy of reporting Publishing negative results Conducting replications Peer review focuses on quality of methods
Who Publishes Registered Reports? So.. who publishes these things? Here’s a partial (and growing!) list. You can view the complete list on the Registered Reports project page on the OSF. There’s even a table comparing features of RRs across journals. (just to name a few) See the full list and compare features: osf.io/8mpji
http://osf.io/8mpji eLife Chris chambers slide/ committee 16 journals incl. eLife Special issue http://osf.io/8mpji
https://osf.io/x5w7h Find this presentation at (contact information)