JavaScript isn't enabled in your browser, so this file can't be opened. Enable and reload.

1 of 25

Data Publishing and Preservation

Michael Shensky

Head of Research Data Services

m.shensky@austin.utexas.edu

Meryl Brodsky

Liaison Librarian

meryl.brodsky@austin.utexas.edu

2 of 25

Welcome

This is the last of five Data & Donuts workshops in the Fall 2022 term
Click the links below to register for additional workshops

September 2: Data Management with SQL and Relational Databases
September 16: Introduction to Python for Data Management
September 30: Introduction to R and R Studio
October 14: Python for Data Retrieval and Visualization
October 28: Data Preservation and Sharing

Recorded workshops and lecture slides will be posted online at:

https://guides.lib.utexas.edu/data-and-donuts

3 of 25

Workshop Logistics

All workshops this fall will be conducted as Zoom webinars
You can register for upcoming workshops by visiting https://guides.lib.utexas.edu/data-and-donuts
Feel free to ask questions and add comments in the chat
Workshop instruction will run from 12pm to 1pm and there will be time for questions until 1:15pm
Workshops this fall will be recorded and links to view the recordings will be posted at https://guides.lib.utexas.edu/data-and-donuts/information-about-past-workshops

4 of 25

Other Fall 2022 Research Data Services Announcements

UT GIS Day 2022

UT GIS Day will take place on Wednesday 11/16 from 9am to 6pm
We will have a series of event dedicated to showcasing how geographic information

systems & related technologies are being used by the campus community

Events will include:

Maya interactive lidar event
Panel discussion
GIS career event
Lightning talks
Drone demo
PCL Map Room tour
And more…

Proposals are now being accepted for presentations and maps/posters!

Visit the event website using the QR code or link below to submit your proposal!

Visit the UT GIS Day site at

https://guides.lib.utexas.edu/gis

for full event details!

5 of 25

Goals for This Workshop

Gain an understanding of why data publication is beneficial (and/or required!)
Learn about good data management practices in preparation for data publishing and preservation
Discuss data publication best practices including: file naming, metadata, and more
Build familiarity with strategies for protecting sensitive information
Explore different data repositories
Log in to the Texas Data Repository and Texas ScholarWorks

6 of 25

Why Preserve & Publish your Research Data?

7 of 25

Why Preserve & Publish your Research Data?

Ensures your data will not be lost
Required by many funding agencies and some journal publishers
Open Science can support researchers in other fields, or countries (improve equity)
A second research product will increase your research impact

Meryl

As you can see from the video clip, which is incredibly annoying, but memorable – some people don’t have a clue about how to share their data, even if it’s required and even if it could help another researcher.

In the case of this researcher, he doesn’t even really know where his data is. By publishing your data, you will always know where it is, and you don’t have to worry about changing jobs or devices, because the data will be live in a repository.

It is increasingly a requirement of government funded research, that you make your data available. The idea is that your research was funded by taxpayer monies so that the public as well as other researchers should have access to your data.

In addition, it could really help folks who aren’t at research institutions or who might not have access to the kinds of labs that we have at UT. I have helped a researcher at UT find data for an article she was writing in which she was able to use data collected by someone else. This wasn’t key to her doctoral work, but was more like a small project to prepare for her larger one.

Last, the data can be viewed as a second research product (where your thesis or an article might be the first). This can increase your research impact.

8 of 25

Data Preservation & Publication

within the Research Data Lifecycle

Typically comes towards the end of the research data lifecycle
Requires that the initial stages of the lifecycle be carried out successfully
If done correctly, data preservation & publication helps foster further research

Plan

Find/

Create

Process

Analyze

Preserve

Publish

Reuse

Meryl

This is the research data lifecycle, you can see it starts with planning. Then you can either find existing data or create your own by conducting some type of research.

You then must process the data (to clean it) and analyze it to come up with your research findings.

Then you can preserve it and publish it so that others may reuse it.

If you go through each of the steps thoughtfully, you’ll be able to have data that won’t be difficult to publish.

However, we all know that things don’t always go according to plan, and that creating data, and analyzing it is often an iterative process. Hence, it’s a good idea to keep the research data lifecycle in mind, so that you know what the steps ahead of you are, even if it takes a while to get there.

Today we’re focusing primarily on preserving and publishing data.

9 of 25

Good Data Management Practices

Write a data management plan, designate a point person in charge of data decisions
Create a data dictionary to define the variables
Create a file naming protocol and document it
Document workflow and use scripted workflows where possible
Keep raw data raw
Backup data regularly to different servers or computers
Use open standard file formats instead of proprietary file formats when possible

Meryl - First you have to think about how to manage your data throughout the research data lifecycle. All of this will occur in the planning stage.

A Data Management Plan is a written document that describes your data and what you will do with it during and after your research project. Many funding agencies require that you create a Data Management Plan when you request funding. The elements of a data management plan include: who is responsible for the data, the data types, the metadata you will use, plans for access and sharing, as well as re-use and distribution, preservation and long-term storage.

You should consider creating a data dictionary to describe how the variables are defined and what will be collected. We’ll talk more about this in a bit.

Creating a file name protocol is important and we’ll also see an example of that.

You should document your workflow and use scripted flows when you can. You can imagine that this would help new team members to become familiar with the process.

One best practice is to keep “raw data raw.” It’s inevitable that mistakes will occur, so always work with a copy of your data, and leave the original as it is.

Back up your data regularly, in multiple places. UT Box is a great place for interim storage, if your short-term storage is on a department server. You could also consider backing up to an external hard drive, just to be on the safe side.

Also, we recommend saving your data in open file formats such as CSV rather than Excel, so that people who re-use your data are able to use a program that they prefer and they don’t have to rely on software in order to access your data.

10 of 25

Good Data Management Practices

11 of 25

Data Dictionary & Variables

Please define your variables so that you don’t end up in the Sam2 situation we saw in the video.

Defining your variables may seem redundant since you are working on this experiment and you know what everything means right now. However, you can make things easier on your future self, your team, and any users of this data, by defining the variables ahead of time.

Providing a data dictionary is one way to do it. In the data dictionary, you would define the name of the variable, what type of variable it is (text or numeric), what the measurement or format is… i.e., pounds versus kilograms as shown in the weight column above or how many decimal points, or yes or no, or how much as shown in the smoking field above. You can see that people collected the data in different ways on this spreadsheet.

A data dictionary is part of your documentation.

Another way to handle it is to design your data capture document, so that the only choices data intake researchers can make, fits the variable you have. This way you’ll spend less time cleaning your data, when it’s time to do the analysis.

12 of 25

File Name Protocol - Be consistent, document rules

Example: AtherRat_012_056_mb_0423_raw.csv
AtherRat = Experiment Name
012 = Experiment Number
056 = Sample Number
mb = Stain Used, Methylene Blue
0423 = 2 digit coordinates of image, 4 across, 23 down
Raw = data stage

13 of 25

Readme File = Inventory of data files + project

General Information

Sharing/Access Information

Data & File Overview

Methodology Information

Data-Specific Information

Source: Readme Template. https://utexas.box.com/v/readmetemplatefordata2022

Meryl

In mentioning documentation, I also want to point out the utility of a readme file. Think of the readme as the inventory of everything that goes in the package with your data files to explain its contents.

A ReadMe file can also help you to get organized.

GENERAL INFORMATION

Title of Dataset:

Author/Principal Investigator/co-Investigator <provide at least two contacts>

Contact Information: Name, ORCID, Institution, Address, E-mail

Date of data collection: <provide single date, range, or approximate date; suggested format YYYY-MM-DD>

Geographic location of data collection: <provide latitude, longitude, or city/region, State, Country>

Information about funding sources that supported the collection of the data:

SHARING/ACCESS INFORMATION

Licenses/restrictions placed on the data:

Links to publications that cite or use the data:

Links to other publicly accessible locations of the data:

Links/relationships to ancillary data sets:

Was data derived from another source? If yes, list source(s):

Recommended citation for this dataset:

DATA & FILE OVERVIEW - this is basically a file directory

File Directory : <list all files (or folders, as appropriate for dataset organization) contained in the dataset, with a brief description> here’s where you’ll put inputs, outputs, raw data, processed data, source code

Relationship between files, if important:

Additional related data collected that was not included in the current data package:

Are there multiple versions of the dataset?

If yes, name of file(s) that was updated:

Why was the file updated?

When was the file updated?

METHODOLOGICAL INFORMATION

Description of methods used for collection/generation of data: <include links or references to publications or other documentation containing experimental design or protocols used in data collection>

Methods for processing the data: <describe how the submitted data were generated from the raw or collected data

Instrument- or software-specific information needed to interpret the data: <include full name and version of software, and any necessary packages or libraries needed to run scripts>

Standards and calibration information, if appropriate:

Environmental/experimental conditions:

Describe any quality-assurance procedures performed on the data:

People involved with sample collection, processing, analysis and/or submission:

DATA-SPECIFIC INFORMATION FOR: [FILENAME] <repeat this section for each dataset, folder or file, as appropriate>

Number of variables:

Number of cases/rows:

Variable List: <list variable name(s), description(s), unit(s) and value labels as appropriate for each>

Missing data codes: <list code/symbol and definition>

Specialized formats or other abbreviations used:

14 of 25

Good Data Publication Practices

Findable = Use metadata for datasets
Accessible = Publish in an appropriate repository
Interoperable = Do not compress & use non-proprietary formats
Reusable = Data should be well described, so it can be replicated or combined with other data

Meryl

Image: SangyaPundir, CC BY-SA 4.0via Wikimedia Commons

https://upload.wikimedia.org/wikipedia/commons/a/aa/FAIR_data_principles.jpg

Fair is an acronym for Findable, Accessible, Interoperable and Reusable. The FAIR framework is a way to publish your data, that will help others to find and re-use it.

Findable

The first step in (re)using data is to find them. Metadata makes data easy to find for both humans and computers. You may be wondering what metadata is. It’s information about this dataset. It typically includes the title of the project, the institution, the funding agency, the principal investigator and/co-PI. It may also include data licensing or use restrictions as well as subject headings that describe what your data is about. So, if you were interviewing elementary school teachers about their experiences in dealing with stress in the classroom, some of your keywords might be elementary schools, teachers, stress, interviews and education. Librarians can assist you with this part. The idea is to describe your dataset, so that others will be able to find it using a keyword search.

Accessible

Publishing in appropriate repository means using a subject specific repository, of which there are many and we will take a look at these later. If there are no subject specific repositories, you can publish in a more general one such as the Texas Data Repository. In some cases, funding agencies require the use particular repositories. The NIH is specific about its repositories because they would like health-related scientists to share their data in easy to find repositories. It is precisely this kind of access to data that was able to accelerate research into COVID vaccines. Some journal publishers also require you to publish the data your article is based on in a freely available repository. Here, they are mainly looking for a DOI to connect to. You can use the Texas Data Repository for this purpose.

Interoperable

The data usually need to be integrated with other data so use open file formats such as txt or csv rather than Microsoft or other proprietary saved documents, such as the one from the video from the company that went bankrupt. In the past I worked with a faculty member who needed to be able to find a computer than ran an old operating system in order to read a CD that had some historic data that he needed to use. This situation is more common than you might think.

Reusable

The ultimate goal of FAIR is to optimise the reuse of data. To achieve this, the process and software should be well documented so that they can be replicated and/or combined in different ways.

15 of 25

Publishing & Protecting Sensitive Data

Anonymize
Generalize
Encryption
Aggregation
Destruction

You can still publish sensitive data, but it must be protected.

You must protect:

Personally identifiable information (PII), like names, home addresses, and social security numbers
Protected health information (PHI), including medical records and insurance information
Non-personal commercial sensitive information, such as supplier information or intellectual property (IP)

In addition, some other people or entities may need special protection such as: locations of endangered species, people who engage in illegal activities or those who are critical of some governments.

If you’ve gone through IRB training then you know what kinds of data needs to be protected. Hopefully you’ve been working with it in a password protected environment, such as UT Box, so that only the members of your research team can access it.

Here are a few techniques to hide or protect sensitive data when you want to publish it.

Anonymization is technique that removes or modifies personally identifiable information; it results in anonymized data that cannot be associated with any one individual. You can do this through redaction (that’s like placing a black mark over or placing xx’s in fields that might be identifying or sensitive. Or you can use false identifiers (such as different names).

Generalization removes a portion of the data or replace some part of it with a common value. For example, you could use generalization to replace segments of all area codes or phone numbers with the same sequence of numbers.

Encryption is when you have to have a certain key to decode the data You could make it so that authorized users have access to a key or password that makes it possible for them to retrieve the data.

Aggregation is when data is grouped together to talk about things like average incomes in a particular zip code, rather than household income.

Destruction is more often paired with a time frame. For instance, five years after publishing this data, all the original records will be destroyed.

Now I am going to turn it over to Michael, and he will talk about the Texas Data Repository.

16 of 25

Texas Data Repository (TDR)

TDR is a multi-institutional data repository managed by the Texas Digital Library and built on open source Dataverse software
UT Austin has its own “Dataverse” within TDR
All UT Austin affiliates can log in with their UT EID
If you are leaving UT Austin (graduating, etc.) contact TDR support so that you can maintain access to your account

Try logging in

https://dataverse.tdl.org/

17 of 25

Why use the Texas Data Repository?

Comply with funding requirements
Ensure reliable, managed access for data
Make your research data citable and increase scholarly impact since all published datasets receive a DOI
Collaborate with research team members who can also use TDR
Access local data publication support through the UT Libraries
Great default choice for a data repository if you don’t have time or interest to search through all possible options

18 of 25

Adding Data to the Texas Data Repository

Create a dataverse under which all of your related projects can be grouped
Create datasets within that dataverse to store data about a particular project or study
Create metadata for your dataverse or dataset
Upload individual data files to your dataset
Determine data licensing (default is CC0) and any restrictions you want to impose on data access

19 of 25

Managing Data in the Texas Data Repository

You can save draft datasets in TDR if you are not quite ready to publish
Publish new “versions” if data changes
You can ask for a review of your data prior to publication if you would like
Try to avoid deaccessioning data if at all possible

This will lead to a “tombstone”in TDR

20 of 25

Data Limits in the Texas Data Repository

Datasets should be less than 10GB in size
Individual data files can be up to 4GB, but files larger than 2GB may experience upload issues
For projects with 10GB+ of data, consider breaking the data up into multiple datasets

21 of 25

Detailed Info about the Texas Data Repository

Vist the TDR user guide at https://texasdigitallibrary.atlassian.net/wiki/spaces/TDRUD/overview for detailed instructions about all aspects of utilizing the repository

Works best in Google Chrome

22 of 25

Texas Data Repository

TDR was built using Dataverse software, you can use the very powerful Dataverse API to interact with TDR for operations like

Searching
Downloading
Creating dataverses or datasets
Examining dataverse content statistics
And more!

Some API functions require you to generate your own API key
Try using the API with Python scripts that utilize the requests module

https://guides.dataverse.org/en/latest/api/

23 of 25

Other Data Repository Options

Re3data.org is a website you can use to search for online data repositories
This site can be helpful when you are looking for data to download or looking for a place to publish your data
Practice using re3data.org to find a data repository that might be relevant for your research

https://www.re3data.org/

24 of 25

Texas Scholar Works (TSW)

TSW is a repository for depositing scholarly publications associated with your data
Posters, conference presentations, articles, theses, dissertations, and more
Over 100,000 publications already deposited
All publications receive a DOI and Handle

https://repositories.lib.utexas.edu/

Try logging in

25 of 25

Wrap Up

Michael Shensky

Head of Research Data Services

m.shensky@austin.utexas.edu

Questions? Comments?

Upcoming Events

UT GIS Day will take place on Wednesday 11/16 from 9am to 6pm

Meryl Brodsky

Liaison Librarian

meryl.brodsky@austin.utexas.edu