Thompson, E. by K., Hill, E., Carlisle-Johnston, E., Dennie, D., & Fortin, É. (Eds.). (2023). Research Data Management in the Canadian Context. Western University, Western Libraries. https://doi.org/10.5206/ZRUV7849
2026-01-08 Lecture Introduction
Syllabus review, what is data, what is RDM, who and why.
https://libraries.dal.ca/services/research-data-management.html and https://dal.ca.libguides.com/rdm
Some grant/funding requires publishing the datasets used in research, and academic libraries are often the place where those live. We use dspace and borealis. Lots of academic librarian bibliometrics scholarly communication folks in this class. Everything that is researched in or about this land has to be done within the context of the Indigenous communities here. Ethics approval only applies to the specific study/data, you can't just secondary-use data because the ethics don't apply. Same for consent of study participants. Yes, you do have to go get consent from every member of that dataset for the secondary study.
Assignments changed order!! adventure jan15, personal organization 22nd, data lifecycle feb12, group presentation workshop march 5th, care analysis march 12, group rdm case study april 2. Updated brightspace and syllabus hopefully today. In-class participation hands on experiential learning. Have personal data ideas on hand to work with next week.
What is research data?
Data is objective pieces of information, algorithms even. Info is aggregated data. Knowledge is rules, principles derived from the information, internalized information used to make decisions. It's unclear when something moves from one classification to another. Contextual. Pyramids suck.
Research data specifically is data used for research. Raw material for research findings.
What is RDM
RDM: storage, access, preservation of data, policies and practices, planning, backing up etc. Research data lifecycle: Plan, create, process, analyze, disseminate, preserve, reuse, and back to the beginning. Use RDM in all of the phases, not just at the end.
Gotta publish your data, organize it well, include a guide to your variable names etc. (video) RDM plans and data required to be published and available based on funding and journal policies. Get involved in the beginning so you don't have to ask "What does the variable Sam2 mean?" at the end of the process. Be better researchers, but also be good members of communities.
Important for reproducibility, other researchers might need your data for other topics. Librarians often involved in writing and revising their organizational policies. Ethics approval includes things like anonymizing data that you release. Encryption requirements depend on context and policy. She wasn't permitted to use microsoft forms to collect data because of the lack of privacy etc. Journals also want ethics info including RDM plans at time of submission/publication; didn't consider the university's stamp of approval sufficient.
Librarians, researchers, repository managers, industry.
How to manage our own research?
Document your process, tools, variable names, software versions, etc. Record your decisions. How are you noting "there's no data" or "not applicable" with 99999 or 88888 or whatever. Null and blank are ambiguous. Readme files are useful for giving an overview of what's going on.
Chapter 1, intro and lifecycle
Columns for your variables, rows for your observations, one header row with variable names. Put your dictionary in a separate document. Put calculations/functions in a separate document. Don't use formatting as data (ie whether a cell's contents are red shouldn't mean anything) because formatting can get lost.
Reproducible: taking the data set, repeating the method of analysis, should get the same results.
Replicable: collect new data (from the same population?), repeat the method of analysis, get the same/similar results.
Traceable: raw data should be convertible to processed data using the documented methodology and should arrive at the same results (data manipulation verification). Document all modifications, who made them and why; chain of custody.
Tri-Agency Policy 3 Requirements:
- Institutions have to have an RDM strategy, submitted to the agencies and publicly available, that support researchers in that institution.
- Data management plans are required to be submitted for some funding.
- Data and code must be placed in a public trust/deposit, not necessarily publicly available (bc confidentiality). Only have to deposit what they actually use, not everything collected.
Data Management Plan (DMP): formal description of research data lifecycle process from collection to disposal. DMP Assistant is a web-based tool to aid researchers in writing a DMP. Living document, but has to be submitted before research starts? Agencies ask for how data is collected, formatted, protected, preserved; what new data is collected and what old data will be reused; whether and how data will be shared and deposited. DMP have little eveidence of their effectiveness, but if they're only measuring "benefit for researchers, institutions or funding bodies" (Smale et al 2020), maybe they're not measuring downstream benefits for replicability etc?
"Open formats are future-proof." ahem. Use open standards, sane file naming conventions. Documentation should include a notation for each change, who made it and why, where data came from what column names mean, etc. Use a README file that explains what the files are and how they interrelate. A codebook is a document that eg lists all of the questions and possible answers/formats in a survey and how the answers are represented in a spreadsheet of answers, how the survey sample was chosen, etc. Someone else in your field should be able to understand the decisions you made, reproduce your methods on your raw data to get your processed data, and reproduce your methods on the processed data to get the same result. If you use code to transform your data from raw to processed, traceability is as simple as running the code; be sure to include the code.
Store your data. Having an open format is useless if the data itself is gone. Store it on reliable media with a specific backup strategy and off-site replication. Remember that backups don't matter, restores do.
Preservation happens after research is complete. Put it in a trust/depository. Note what restrictions exist on its access, especially its reuse for future research. Include licensing terms, eg CC-BY-SA. Anonymization. Institutions, journals, and fields may have depositories. There are reasons why sharing data may not be appropriate, or may have some restrictions.
DMP Assistant intro video: https://www.youtube.com/watch?v=N_Mzq60D4nc
Chapter 4, history
Way too many organizations, reports, projects etc to cover. My takeaway is that everyone is doing their own thing.
Lecture 2026-01-15
People who have research experience found the choose-your-own-adventure easier. Not for beginners. I guess I'm not a beginner? But I also was having fun derailing things sometimes. It's possible to over-document (overwhelming, doesn't add value) but more often people under-document. Research is typically conducted by a team, so if you know RDM skills or are the liaison librarian you might have to take point on this for the team. Regional and national groups can provide support.
RDM includes stewardship of data. Researchers focus on where they get the most impact (citations), may disregard RDM in favor of things that matter more to them.
Store underlying data rather than calculated results you can derive later. Standardize variables and encoding, units, definitions of terms. Pulse vs HR vs spo2. Reproducibility is a good goal to guide decisions.
Things programmers get wrong about names, time, addresses etc all apply to this.
RDM can be efficiency gains, make the most of the funding. Earlier is better, coming at the end when your data is a mess there will be little we can do to help. Being able to reference the ethical stuff from an existing document can help during publication submission and peer review.
Portage 7-slide on RDM. Minimize context required to understand data.
The data produced during the research can be seen as an output, not just the paper. More opportunities for citation and impact.
Academic libraries are often responsible for managing article processing charge funds etc but they don't get money from the grants awarded they just eat the cost. But it gives us a good talking point for the value we add to the university.
Open access->open data.
The Corti 2020 p34-35 list of lifecycle activities would be a good checklist for writing a DMP.
Love data week, feb 9-13. CARL committees.
Lecture 2026-01-22
All of the technical requirements make sense to me and were in line with what I expected. But the portions about ethics have really made me think, and the personnel factors like succession planning were things I hadn't thought about before.
Some health librarians coming to do guest lectures/workshops later in the term.
Group discussions about the readings, assignments, etc to help us feel connected and to compare notes a little bit.
Review of a job ad for a data research librarian. Difference between experience (did something) and knowledge of (just have to know about it). Almost all of the things shown in the job ad are things we're going to be covering directly in this course. Having a well-crafted cover letter is extremely important for this kind of job. code4lib job board. Sometimes the contents of the job ad are the only things they're allowed to ask you about during the interview process/evaluation for the position. Usually a long list (10 ish) interview and then a short list interview of maybe 3 people.
Digital humanities east: Summer training program in digital humanities, certificate opportunity. $300 and the grant covers some of that maybe. Chance to network too.
OCAP applies only to First Nations, not to Metis or Inuit. And even then it needs to be handled per-community. Don't lean on OCAP or other standards as a pan-Indigenous solution. Similarities with the ethics TCPS2 work we're doing in Research Methods. The personnel requirements were new to me. The definitions of "commercial" can get tricky, so having CC-NC can be a problem for some uses we might not consider commercial. Triagency could compel people to share their data, private sponsors could compel its restriction. How do you handle public-private partnerships, or any multi-sponsor situation? Hopefully the sponsors all got together and came up with a unified set of requirements, but if the researcher is the one who is courting two sponsors then the research needs to be careful. All the legal stuff and negotiations etc is often handled by the university's legal department, and given all the complexity it's difficult for anyone outside the academe to do research.
Data management plans should be created in collaboration with all partners, including the researched population. Nothing about us without us. They are living documents, meaning they're continually revised as research demands.
DMP Assistant is pretty simple and straightforward, not unnecessarily complicated. Include collaboration options for researchers to write it together. Templates available from the institution, publicly, etc. Can log in through your institution or a separate account. Can link your ORCID.
Fixed syllabus dates to match brightspace dates for some assignments.
Data management for libraries chapter 2 starting a new service
Krier, L. (with Strasser, C. A., & Thomas, H.). (2014). Data management for libraries: A LITA guide (1st ed.). ALA TechSource.
Start with a beta and be prepared to fail and iterate. Be sure there's a need and that you have resources. Get stakeholders involved. Data management has a lot of pieces, so start with the ones that are easiest and have the most impact for patrons. Considered tiered services. Consider what to offer based on the stages of the data lifecycle creation through archiving. Consider training for library staff as well as the folks who will use the service. Be sure you have staff to allocate to these responsibilities.
Data management for libraries chapter 4 the data management interview
Krier, L. (with Strasser, C. A., & Thomas, H.). (2014). Data management for libraries: A LITA guide (1st ed.). ALA TechSource.
This is how a collaboration starts between the library and the patron. Lots of overlap with reference interview. Need to have a solid understanding of DM process and what's offered by the library to effectively host this interview. Build relationships with other departments like IT to coordinate efforts. If you're not sure what they need, you can start off by interviewing each department to see how they're currently managing their data. You can also interview the liaison and reference librarians that are already supporting research efforts, IT staff, department heads, grant process, lab heads and equipment maintainers, metadata standards by field.
https://datacurationprofiles.org/ is dead, https://docs.lib.purdue.edu/dcptoolkit/ seems to still work.
Variation in receptiveness by researchers, reiterate suggestion of tiered approach. Researchers often want to control how their data is disclosed and have varying opinions on repositories etc. Target individual researchers and use the data requirements from their grant intersected with the library's data management service offerings to help guide the conversation. A lot of this looks like drafting a DMP in an interactive format.
ARL/CARL joint task force on research data services
Association of Research Libraries & Canadian Association of Research Libraries. (2021). ARL/CARL joint task force on research data services: Final report. https://www.arl.org/wp-content/uploads/2021/07/2021.07.16-arl-carl-joint-task-force-on-research-data-services-final-report.pdf
Has recommendations that could be used as a roadmap for starting a data services program at a library.
- Conduct a cross-campus mapping of existing campus resources and researcher needs for RDS (what researchers are already doing)
- Define a library portfolio and strategy for RDS (SWOT analysis and strategic plan)
- Articulate library and institutional research data services and partnerships (who can we work with and what services are we already providing researchers)
- Formalize partnerships through development of a service catalog (so anyone can find them, 6 variables for each partner)
- Document services by elements of data management requirements (similar to above but by part of the data management lifecycle instead of by partner)
- Evaluate the program on a spectrum of maturity
- Define an institutional strategy for RDS (like 2, but for the whole institution?)
Each of those has a little explanation and links to people doing those things.
They also have notes on discipline-specific research data management, thoughts on priorities for automation of RDM, some notes about cost. Not a bad little document.
2026-01-29 Lecture
Starting with 10 minutes of random chatting. Look for jobs at Digital Research Alliance Canada. Comparison of scholarly communication job posting vs metadata specialist job. ORS office of research services. ROMEO funding application management, Dal Support for researchers. Don't apply directly to funding agencies, submit it to ROMEO and get approval from department and faculty and university. Somehow ROMEO doesn't cover human research ethics. All the IT problems (and policy problems) for inter-institution collaboration. General discussion about what you need to do as a researcher, what gets submitted where, IT resources, lots of other things. CHEB health library would be good to talk to. Head is swimming.
2026-02-05 Repositories and data cleaning and metadata
Prefer a topic/field-specific repository, then an institutional repository, then aa generalist repository.
So tools like dSpace are self-serve whereas the library catalog is only managed by library staff, kind of an axis of self served vs managed. And then open source and institution-manged vs closed source and paid hosting/support axis. Metadata vs data vs articles and other research outputs vs long-term preservation.
Archivmatica, Borealis, dSpace and be-press, Islandora.
Islandora as long-term preservation handles access controls, dSpace doesn't have good tools for that.
External platforms like disciplinary specific repositories.
What about Indigenous data repositories? Mukurtu as example. https://nunaliit.org/
2026-02-12 Love Data Week - Data Deposit 101, and Lecture
Data retention, accessibility, reproducibility and transparency, reuse. Vs FAIR: Findable, Accessible, Interoperable, Reusable.
Deposit != share. If there's no ethical way to make the data usable in the future, eg if it's a case study of a single individual with identifying information that can't be anonymized, would that be grounds for not depositing it? Or I guess more generally, are there times where deposit isn't appropriate? Conversely, are there reasons to deposit when you know you can't ever share? Sometimes worthwhile to deposit anyway with good metadata so people can at least know that someone collected this kind of data. Transparency and accountability too.
If Borealis is what you support the most, what repository do you support the least? AI scraping screwing up access for human researchers.
For Borealis, for scholars who are affiliated with multiple institutions, is there a way to link multiple institutions together under a single account somehow? Is there value in signing up with a separate email address in case you leave the institution in the future? Recommend to sign up via institution. Will continue to have access as alumnus.
Put the list of files in the README to make sure it's a complete copy, or README may be the only item available that could give clues to what's in the full dataset.
What is PIPEDA? The Personal Information Protection and Electronic Documents Act https://www.priv.gc.ca/en/privacy-topics/privacy-laws-in-canada/the-personal-information-protection-and-electronic-documents-act-pipeda/
Workshops: 20-30 minutes, 3-4 groups. File naming, life cycle, depositories, DMPs, digital humanities. DMP assistant, file / metadata, RDM for MI broadly, FAIR, CARE, Rights. It sounds like we're going to do a single four-part workshop conglomeration on March 5th covering DMP assistant, file / metadata, RDM for MI broadly, FAIR, CARE, Rights. And we're going to sign up for which part of that we want to work on via the Brightspace groups feature.
2026-02-26 CARE and OCAP
Settling/colonization requires the description and disappearing of the people indigenous people (tuck and yang, 2012). Our existing systems of power reinforce privilege and perpetuate imbalances, including in GLAM, including research. Critical analysis in research is required as much as anywhere else. Carroll Herczog Hudson russel Stall (2021) any data/info/knowledge in any format that impacts or relates to Indigenous peoples nations communities as collectives and individuals, conducting in Indigenous land, about their environment, etc.
OCAP ownership, control, acquisition, possession. FNIGC.ca First Nations information governance council? OCAP is First Nations, doesn't apply to Metis or Inuit. https://fnigc.ca/ocap-training/take-the-course/ $500. Nova Scotia has MAWKIM Mi'kmaw Data. Rowe 2021 Indigenous data can be used to unsettle existing colonial power structures.
CARE future use and future harms. UN Declaration of Rights of Indigenous Peoples article 31 applies too UNDRIP. Information as an economic resource, under capitalism that results in information being horded and exploited for the benefit of the few. Proponents of OpenData may not realize or appreciate the situations where openness of data is inappropriate and harmful. AI is an example of why open data/let everyone see everything all the time can be bad. Canada 2018 supposedly going to take TK and traditional cultural expressions seriously and align IP policy somehow, incommensurability (non-blendable). Canada late signer for UNDRIP because they have political and financial incentive to ignore Indigenous rights (same for US and Australia). Copyright/IP is based on English common law and represents a single cultural viewpoint of knowledge management and "ownership." There are ways to consider information that isn't based on economic benefit through exclusivity. A multicultural society can't have a unified IP law that applies to everything. "Distinctions-based approach" in tri-agency data policy means that you have to meet the distinct needs/rights/etc for each Indigenous people. "Turtle Island" is a single culture's creation story and isn't held by all Indigenous people.
GIDA global Indigenous data alliance. CARE: Collective benefit, Authority to control, Responsibility and Ethics. FAIR: Findable Accessible Interoperable Reusable. CARE is a global project, not specific to First Nations like OCAP is. https://www.youtube.com/watch?v=g7-RuXmLewk Local contexts video. A researcher documenting a cultural history gains copyright over it. Traditional Knowledge Labels ways of tagging info/artifacts in conflict/contrast to Western ownership/IP. Trying to recover information from wax cylinder recordings from 130 years ago as a reclaiming, trying to transcribe and understand the cultural context of the song and how would you use IK labels. And who ends up owning these? Digital infrastructure/metadata behind the LOC website for these songs, didn't have fields for Indigenous cultural context etc, so we have to create the metadata standards to allow entry for that information. Provenance vs providence? TK labels don't tell you what the resource is about, but its provenance, access rights/restrictions, etc. Netherlands 2 years ago law to return cultural artifacts to places they colonized but without provenance data they don't know where it came from. TK labels aren't pan-Indigenous, you'd need to verify with the TK label folks and the community the data belongs to. Provenance, access, permission. IEEE SA P2890 recommended practice for provenance of indigenous peoples' data.
https://openglam.org/principles/
Not just about the data from our own research, but as we support other people doing research.
Next up we're talking about the workshop assignment and the CARE assignment.
Workshops moved to April 2nd, CARE assignment posted with no change to due date (12th?), grades will scale linearly, no more case study assignment. Will post proposed new syllabus via teams for approval, workshop will be weighted more heavily to reflect that it's more work than expected. Use existing RDM teaching resources in the workshop as long as licensing allows it. Maybe make a moodle now that the timeline has changed?
By next week finalize the title of your workshop.
CARE assignment isn't intended to be super deep. Do a brief analysis of the two cases and a synthesis at the end. You don't have to cover very point listed on the assignment, it's there to give you a sense of what you're looking for. Borealis, Dal, FRBR, Dal libguide on data has repository list for places to look for cases to study. https://indigenousclimatemonitoring.ca/resources/map-of-indigenous-climate-monitoring-projects/ https://dal.ca.libguides.com/data/canada Identify missing cases. Include a link to your resources.
2026-03-12
Did I miss notes for a week? Looks like yes. It was an in-class exercise for depositing things in borealis.
Discipline differences and Indigenous data management.
Some disciplines lean more towards quantitative, some qualitative, some mixed. Usually when we think about RDM it's related to quantitative. Big datasets, statistical analyses. Proprietary sensor file formats or instruments. Data science and bibliometrics are quantitative.
We did an information world mapping exercise as a side quest. It's qualitative. I derailed things I think. Oops.
Pima Indian Diabetes Dataset. Treated as placeless, extractable, and controllable by others. No communication that this data was converted to ML usable form and put it into a data repository without knowledge/consent. Is ignorance of lack of consent a viable defense? Is using the Indigenous dataset a valid support for the idea that the research is benefiting Indigenous communities?
Kukutai and Taylor (2016) definition of Indigenous data sovereignty. They didn't come up with the idea.
Protection of the peoples and ensure cultural safety, and also self-determination and community decision making. Policy commits to it, but how do you actualize the principles? Need concrete steps/tools/workflows. Territorialization as a tool.
Danica Pawlick-Potts is building framework to analyze the deterritorialization conducted by data repositories and suggestions for how to fix things. Three case studies: Pima, Tsimshian archeological spacial archive, and Lunaris federated research data repository.
Data as a resource vs data as a state of information. Lumber is a state of wood, data is a state of information. Lumber is useful for building things, data is useful for understanding things? It's intended to be used. We can build things positively but not necessarily. Territoriality as a way of defining/defending access to resources. (Robert Sack's territoriality?)
State and corps assert control through server location, legal, protocols standards and infrastructure. Myth of data "placelessness." Parallel to "empty" land justifying seizure/enclosure. Space as a form of power. Repositories have obligation to expose the space element/territory.
How can we map the datasets to their Indigenous territories, both for spacial understanding and context but also for authority/control/sovereignty. Classification (who/what belongs to a territory, define boundaries); communication (make territory visible and understandable); enforcement (control access and behaviour within territorial boundaries).
A territory is by definition not a "neutral" space. Origin, location, responsibility. How does this account for the displacement from territory? Data will be territorialized somehow, and corp/state will always be the default.
Scalar-affordances: the implementation. User->perceives->decides->acts. Affordance is what is made possible, cf Organization of Information reading? Poirier Costelloe-Kuehn 2019 data culture heuristic (meta/macro/meso/micro, techno/data/nano); access being possible isn't the end of the consideration. Affordances can be canonical, relational (community, pens can be projectiles), hidden (requires expert knowledge), false (deceptive, doesn't actually do what's advertised/implied). Relational could be applied to Borealis's metadata options for notating Indigenous data sovereignty. Perception, intentionality, actualization (policy+infrastructure).
Metadata as boundary, creation of groups. Spatialy/land, language groups. It sounds like FRDR has the same problems as Borealis, geonames and notes field. Applies to discovery mechanisms/principles! Not just is it an option, is it a first-class access mechanism. Classification: what fields exist? Communication: search and discovery. Enforcement: access and permissions. Insiders vs outsiders access.
Technology by itself is not sufficient. Librarians shape these conversations with researchers and repositories.
2026-03-19 Qualitative data
Most RDM research/process is focused on quantitative data and large datasets. Qualitative gives voice to lived experience. Can do mixed methods to get some of each. Often has a social dimension.
More likely to identify the signal in the noise, but may struggle with figuring out what outsiders might need to have explained to them. Keeping a reflective journal as a researcher as a member of the studied community can become part of the research, and maybe otherwise.
https://www.tandfonline.com/doi/full/10.1080/03057925.2022.2116929#abstract
Difficulty in applying Western IP law to protect some types of Indigenous knowledges and data. "too old", "no identifiable author","collectively owned" no standing to make economic claims.
FNMI = First Nations, Metis, and Inuit.
Universities sue researchers when they leave the institution and take their data with them? Who owns things is apparently complicated. Indigenous materials frameworks/agreements. 7-generation agreement. Metis have OCAS, FN have OCAP, Inuit Principles (ITK Inuit Tapiriit Kansomething). GRATK treaty on genetic research and traditional knowledge. Kayla.lar-son@ubc.ca, Xwi7xwa.library@ubc.ca.
Books are considered depreciating assets and will eventually be deaccessioned due to that.
Pages that link here: