In order to explain later parts of the presentation, I need to break down the title and use that as the basis for the order I will talk about things.
Science has many philosophies
Focus of talk will be on: Verifying claims
Science has many philosophies
Focus of talk will be on: Verifying claims
How? By independently checking findings are:
What does philosophy have to do with this? It's important to bring this up because it is the basis to understand why the tools I'll talk about are important.
Verification... How claim was made is more important than the claim itself. Anyone can make a claim, but a philosophy of science is that any claim needs be to proven.
Reproducible: Verify that exact process leading to findings can be produced again and should be first step. (reproduce: literally to produce again)
Replication: Verify that process can replicate (of findings, without which finding probably doesn't reflect reality) (replicate: to make a replica, make a copy of)
... terms are often confused with each other [1]
... terms are often confused with each other [1]
... terms are often confused with each other [1]
I'll get into research on reproducibility later.
Study design, methods, and analysis need:
Detailed description
Transparent and clear
Easy to access to any researcher
Record of activity
Description: This includes how exactly analysis was done, ie. based on analysis code
Record: (for fraud identification, logs, trace-ability of what was done)
Inspectable: (linked to accessibility, but also common language, simple to read, logical, well-reasoned)
Already know replication is a major problem
Don't know extent of reproducibility
Already know replication is a major problem
Don't know extent of reproducibility
OSC project: Open Science Collaboration Project
Except maybe bioinformatics, where about 60% of studies do.
There are lots of reasons for this, likely due to:
Research studies need to be more reproducible and replicable [1]
Research output needs to include more than just the manuscript (e.g. data, code) [1]
There are several efforts at addressing the replicability problem, but currently there isn't much focus on the reproducibility problem.
As I mentioned earlier, reproducibility is about taking the same data and the same analysis plan with the code and having someone else reproduce the exact same findings as the original authors.
There are various degrees of this and we in health research have challenges with the data side of the issue, but sharing the
Not same as keeping backups like when e.g. using Dropbox
Most common form of 'version control' is through naming files, but is very manual and error-prone.
An extremely common and very manual form of version control that many of you may recognize is by naming your files differently depending on the date of edits or version that was sent around to e.g. co-authors.
However, this is a highly informal system and very prone to error and mistakes.
Tracks and saves changes done to files into a history of changes.
Tracks and saves changes done to files into a history of changes.
Think about these questions:
Do you have many files of different versions of a manuscript do you have, including getting feedback from co-authors, etc?
Have you ever wanted to write out and test some ideas in your code or manuscript and need to make a new file so that the original is not modified?
Have you ever deleted text in a manuscript or code and wish you hadn't?
Tracks and saves changes done to files into a history of changes.
Think about these questions:
Do you have many files of different versions of a manuscript do you have, including getting feedback from co-authors, etc?
Have you ever wanted to write out and test some ideas in your code or manuscript and need to make a new file so that the original is not modified?
Have you ever deleted text in a manuscript or code and wish you hadn't?
=> Answered yes to any? Git helps with these situations.
Version control like Git fulfills:
Transparency
Record-keeping
Self-containment
Accessibility
Version control like Git fulfills:
Transparency
Record-keeping
Self-containment
Accessibility
And key aspect:
Self contained in that all files related to a scientific claim are kept together and discoverable together.
Accessible: many ways to find and share research project
There are numerous other tools that improve or contribute to reproducibility that work through version control software like Git. Without it, these other tools don't work or are quite difficult.
Work more efficiently on your own
Better collaboration
Better project management
Greater visibility
Greater potential for gaining collaborators
Defend against fraud or auditing purposes
Evidence of contributions and work
Solo research project: Code for a paper from PhD
Team-based and national/local teaching project: r-cubed
Keyboard shortcuts
↑, ←, Pg Up, k | Go to previous slide |
↓, →, Pg Dn, Space, j | Go to next slide |
Home | Go to first slide |
End | Go to last slide |
Number + Return | Go to specific slide |
b / m / f | Toggle blackout / mirrored / fullscreen mode |
c | Clone slideshow |
p | Toggle presenter mode |
t | Restart the presentation timer |
?, h | Toggle this help |
Esc | Back to slideshow |