Culture Change: How Studying Monarchs Inspires Transparent and Reproducible Science

This seminar is part of a series to provide societies and their journals with information and resources to help their communities be more knowledgeable and prepared to share data (and software) in a way that is relevant and meaningful for each discipline.  This is a 12-month series.

Culture Change: How Studying Monarchs Inspires Transparent and Reproducible Science

   3 September 2021, 10am ET (1400 UTC)

Speakers:

• Elise F Zipkin, Director, Ecology, Evolution, and Behavior Program, Michigan State University; Associate Editor, Journal of Animal Ecology (bio)
• Kayla Davis, Ph.D. Candidate, Ecology, Evolution, and Behavior Program, Department of Integrative Biology, Michigan State University (bio)

Moderator:

• Shelley Stall, Senior Director of Data Leadership, American Geophysical Union 

Description: This seminar will be an engaging conversation with Dr. Elise Zipkin and her students on their tenacious approach to implementing data and software stewardship and sharing within their lab. Dr. Zipkin and researcher Kayla Davis will share their experience, the value the team sees in adopting these practices, their challenges, and how this lab practice prepares them for sharing more broadly.

Seminar Recording: https://youtu.be/JwuSx4pHT2k

Resources referenced during the presentation:

AGU Data Leadership: https://www.agu.org/Learn-About-AGU/About-AGU/Data-Leadership 

AGU Data Blog and Resources:  https://data.agu.org/resources/

Earth Science Information Partnership (ESIP): https://www.esipfed.org

Coalition for Publishing Data in the Earth and Space Sciences : https://copdess.org

Research Data Alliance, specifically the Organizational Assembly: https://www.rd-alliance.org/oa-members

Zipkin Quantitative Quality Lab: https://ezipkin.github.io

Zipkin Lab policies including project completion guidelines: https://github.com/zipkinlab/Policies

Observations made during discussion:

• It's a work in progress. The lab does its best with what we know at the time to make the models valuable and useable by others.
• Code and data must be made available in a consistent way at the time the research is published. The exact practice of each researcher to get to this requirement is flexible.  Some share the progress of their work. Some keep that work private until it's time for quality checks. The final version is made available on the Zipkin Lab GitHub (and preserved on a repository like Zenodo).
• Code needs to be well-commented to be understood.
• The pre-processing code for preparing the data for use by the model needs to be included and well documented. 
• Just sharing a chunk of code is likely not useful.  We need to make sure all components are included that allow for the full workflow (pre-processing, model, post processing) to be included in the package.   
• Good version control is critical in knowing exactly what code is used for your research.  That final version must be preserved and link to your publication.  This saves awkwardness and embarrassment when asked for your code at a later time.
• There is an educational component to our research. We need to make sure when we share the elements (data, software, notebook) that supports our research and that it is clear how the pieces fit together. 
• Do everything you can to have high-quality code/process/research, but also have a plan for how you handle mistakes. Humans are not perfect, and unintentional mistakes can and will happen.
• Commenting code is important for the code developer to help organize their code, and also to remind them of what happens in each section. This is useful when there has been a gap of time since you last reviewed the code. 
• Look for good examples of code to use as a best practice. This helps you improve your technique.
• We need an easy-to-use resource for current funding requirements around data and software specific to grants and data management plans.
• GitHub repos need readme files that can be kept accurate and identify any errors found later.
• Good documentation and version control takes time. 
• Intentional review of lab policies help improve/update the practice. 
• The activity of preparing a data management plan for a grant is a great way to refresh current policy and learn about current expectations.
• It would be helpful if funders, institutions, and other organizations valued data and software research products that are prepared for review and use by others.