Title (of Document) An Introduction to Data Management

Creator (author) Alejandra Sarmiento Soler, Mara Ort, Juliane Steckel

Contributions Jens Nieschulze

Project BEFmate, GFBio

Date 22/02/2016
Date of publication
access Date of cited URLS 12.01.2016

Version 4
Filename Reader_GFBio_BefMate_20160222
Internal storage location \PowerFolder\GFBio Training
Material\BEFmate_WoSho_2014\Reader

Subject (key words) Data management, data life cycle

Description (abstract) Handbook on data management for researchers.
Follows ten steps of the Data Life Cycle (propose,
collect, assure, describe, submit, preserve, discover,
integrate, analyse, publish). Provides information as
well as practical tips and further resources. Informs
about GFBio tools and services.

Type Text
Format MS Word 2010
Resource Identifier
Language English
Licence CC BY-NC-SA 4.0

This work is licensed under the Creative Commons

Attribution-NonCommercial-ShareAlike 4.0
International License. To view a copy of this license,
visit http://creativecommons.org/licenses/by-nc-
sa/4.0/.
An Introduction to
Data Management
Content

1. About this Reader .......................................................................................................... 1

2. What is Data Management? ........................................................................................ 2
3. Data Life Cycle ................................................................................................................ 7
3.1 Propose .............................................................................................................................10
3.2 Collect ...............................................................................................................................12
3.3 Assure................................................................................................................................17
3.4 Describe ............................................................................................................................20
3.5 Submit ...............................................................................................................................23
3.6 Preserve ............................................................................................................................26
3.7 Discover ............................................................................................................................29
3.8 Integrate ...........................................................................................................................31
3.9 Analyse..............................................................................................................................33
3.10 Publish...............................................................................................................................36
4. Data Management with BExIS.................................................................................. 40
5. Data Management at a glance: Summary ............................................................. 42
6. More Data Management: Recommended further reading ................................. 43
7. Glossary ........................................................................................................................ 44
8. References ................................................................................................................... 47
 1. About this Reader

Data are the fundament of science. In the last years, awareness for the management of
data increases more and more and data management is actively performed and
integrated in the research process. This reader aims at further raising awareness for
data management in the research community and introduce activities related to data
management. The structure of the reader follows the concept of the Data Life Cycle
with these steps: propose, collect, assure, describe, submit, preserve, discover,
integrate, analyse, and publish. After briefly describing each step and its role, the
corresponding data management activities are presented, including best practice
examples, tips and further resources.
The disciplinary focus of this “Introduction to Data Management” lies on biology and
environmental sciences and quantitative data. Nevertheless, many aspects apply
universally to quantitative and qualitative data, not depending on discipline. This
reader is for everybody who wants to deepen his or her knowledge on data
management. A primer on this topic consisting of factsheets about each step of the
Data Life Cycle can be found on the GFBio Homepage (http://www.gfbio.org/data-
life-cycle). In this reader, the topics are discussed with more detail.

The German Federation for the Curation of Biological Data (GFBio)

This reader is published as training material by GFBio. GFBio aims at establishing a
sustainable, service oriented, national data infrastructure facilitating data sharing for
biological and environmental research. It acts as a single point of contact for
collection, archiving, curation, integration and publication of data. GFBio offers
advanced tools to support data-driven research, access to data and archiving data.
Data can be discovered by a faceted search. The services support and facilitate several
activities of data management. Links to tools and services offered by GFBio are
presented in the respective chapters.
GFBio provides further training materials on its website. On a regularly basis
workshops are offered about data management and Digital Curation. Visit
http://www.gfbio.org/ for more information.

Acknowledgement
The content of this reader builds up upon the materials provided by the Digital
Curation Centre (http://www.dcc.ac.uk/) and DataONE (https://www.dataone.org/).

1
 2. What is Data Management?

Data management concerns the dealing with data in the scientific context. Often, more
importance is given to results, analysis and derived conclusion than to the data
themselves. However, data are a product of the science enterprise and are more and
more understood as a valuable research output themselves (DataONE 2012b; Ludwig
and Enke 2013; Data Service 2012-2015a). Research data are considered all
information collected, observed or created for purposes of analysis and validation of
original research results. Data can be quantitative or qualitative and comprises also
photos, objects or audio files, resulting from as different sources as field experiments,
model outputs or satellite data. In the following, the focus lies on the management of
quantitative digital data. One reason why data management is important is that the
value of research data is sometimes not yet visible nowadays, which can lead to
neglecting proper data management:
In many sciences experiments or observations cannot be repeated making at least
part of the data so valuable that it needs to be stored for a long time. In many
cases the value of data can only be realized after many years by new generations
(RDA Europe 2014a).

One factor making data management even more important is the growing amount of
digital data available:
The amount of data collected is growing exponentially nowadays. New
environmental observing systems […] will provide access to data collected by
aerial, ground-based and underwater sensor networks encompassing tens of
thousands of sensors that, when combined, will generate terabytes to petabytes of
data annually (Michener and Jones 2012).

In some fields and disciplines, data-intensive research is opening up innovative

research possibilities (Mantra et al. 2014). If well-managed, these data can be used in
order to answer (new) research questions (Corti et al. 2011).
Ideally, data management is accepted as integral part of the research idea and is
already considered early in the research proposal. It includes the collection phase, the
processing and analysis of data, the documentation and preservation. Different data
management activities are associated with each step, ensuring a reliable and
accessible data fundament for the researchers work as well as facilitating sharing and
publishing of data. Well-managed data will further facilitate 1) re-use by oneself or
others over time, 2) to replicate or validate research results (think of the good
scientific practice obligation) 3) processing of so-called wide databases (integration of
many small files of varying syntax) and so-called deep data bases (handling of BIG
data).

Importance and benefits of data management

An example from Mantra Online Course (Mantra et al. 2014) illustrates what role data
management can play and how it may support your research:
You have completed your postgraduate study with flying colours and published a
couple of papers to disseminate your research results. Your papers have been

2
 cited widely in the research literature by others who have built upon your
findings. However, three years later a researcher has accused you of having
falsified the data.

 Do you think you would be able to prove that you had done the work as
described? If so, how?
 What would you need to prove that you have not falsified the data?

The documentation of data analysis and transformation as well as the storing of data
and research results are integral part of data management and could help you to prove
your work. Without data management, not only a solid basis ensuring replicability for
your research results may be missing, but your data can also be subject to data loss
more easily. This may happen due to technical problems (hardware failure), due to
software obsolescence, due to missing information (data cannot be understood in the
future) or due to not storing data in an appropriate way (data will never be found
again). In Figure 1, the loss of information content of data is related over time to the
career of a researcher. It illustrates that often much information is tied to specific
persons. If they leave the project or retire, their knowledge is not available anymore.
And people do of course also forget details over time. So it doesn’t have to be such a
drastic case as in the opening example to clarify that data management may help and
facilitate research.

Figure 1: Loss of meta information over time. Michener 2006

Figure 2 displays some benefits of research data management planning. A Data

Management Plan provides guidelines and procedures for data management
encouraging systematic documentation and description. Data management planning
enhances the security of data. It safeguards against data loss as storage, backups and
archiving are planned. Compliance to funder or publisher requirements of the
collected data is ensured. DFG proposals require for example a specification about the
research data generated in a project (DFG 2014). Quality of research in general is
enhanced as data management ensures that research data and records are accurate,
consistent, complete, authentic and reliable. It also allows for reproducibility of
results. As a side effect data management planning streamlines data handling and can

3
thus create efficiency gains for the whole research project. Data management also
facilitates the handling of big amounts of data. Access and restrictions of use can be
documented in Data Management Plans and metadata. Access to data is possible
when data are shared and made available. This enables collaboration, prevents
duplication and can increase citations for the data creator (DataONE Community
2014; DCC 2008a; UK Data Service 2014a; University of Western Australia 2015).

Compliance

Security Quality
Benefits of
Data
Management

Access Efficiency

Figure 2: Benefits of Research Data Management Planning. Own design after University of Western Australia 2015

Incorporating data management as a routine part of the research process can save time
and resources in the long run. In the beginning, some time is needed to prepare a Data
Management Plan and to get used to new practices and activities. This is rewarded by
extra funding for your data management, increased citations, and less work organising
and understanding data later on (DataONE 2012a).
The costs of data management can be either calculated by total costs of all activities
related to the Data Life Cycle (introduced in Chapter 3). As it is often hard to cost
data management practices, as many activities are part of standard research activities
and data analysis, the costs of data management can also be calculated by focusing on
expenses which are additional to standard research procedures (Corti et al. 2011).
Some costs and benefits of data management can be measured quantitatively, in terms
of people’s time or costs of physical resources like hardware or software (see Figure
3). Others have qualitative character or are impossible to measure at all in advance
(e.g. possible new scientific findings; Houghton 2011).

4
Figure 3: Costs and benefits of data management.

Sharing
The growing awareness for the importance of data results in the conclusion that
research data should be made accessible. The DFG states in its “Guidelines on the
Handling of Research Data in Biodiversity Research” (2015) that data management
should assure a re-use of data also for purposes other than those they were collected
for. Furthermore, they emphasize access to data:
Enabling free public access to data deriving from DFG-funded research should
be the norm.

Sharing data can bring advantages for individual researchers as well as for the
scientific community in general. There are three dimensions of sharing data. First,
data can be shared among researchers within the project team. Therefore data is
submitted to a shared drive. Second, data can be shared with researchers outside the
core research team, e.g. when there is cross-institutional cooperation. In this case, data
is submitted to collaboratively used drive, e.g. BExIS. Third, data can be made
publicly available. This is referred to as publishing data. Data centres like GFBio
preserve data and make them discoverable. A study showed that the publication of
data may increase citations (Piwowar and Vision 2013).

Sharing data facilitates the collaboration within and outside research projects and
establishes links to the next generation of researchers because data are discoverable
and understandable. It also allows to approach research questions which were not
thought of when the research started. Another advantage is the prevention of
unnecessary duplication of data collection. Furthermore, a key factor for science is
replicability, so researchers can collect data and analyse them in order to produce
similar results or assess previous work in the light of new approaches (e.g. voice
recording of bat signals and the determination of species) (UK Data Service 2012-
2015b). However, if that information is not available or poorly documented and
difficult to understand, re-use or replication is difficult (Heidorn 2008). Besides the

5
voluntary sharing of data, many journals already request the submission of data
underpinning a paper.

Public Library of Science (PLOS) (http://journals.plos.org/plosone/s/data-availability)

PLOS journals require authors to make all data underlying the findings described
in their manuscript fully available without restriction, with rare exception.

Nature (http://www.nature.com/authors/policies/availability.html)

An inherent principle of publication is that others should be able to replicate and

build upon the authors' published claims. Therefore, a condition of publication in
a Nature journal is that authors are required to make materials, data and
associated protocols promptly available to readers without undue qualifications.

It has to be acknowledged that there are of course also a number of reasons why
researchers do not wish or are not able to share research data (see also 3.10 Publish).
Not all of these reasons may be overcome (Mantra et al. 2014). These barriers include
finances, confidentiality of data and ownership issues. Corti et al. (2011) discuss some
of these barriers and show possible solutions (see Chapter 3.10 Publish).

6
 3. Data Life Cycle

The different activities concerning data management can be structured in the so called
Data Life Cycle (Figure 4). The Data Life Cycle is a conceptual tool which helps to
understand 10 different steps that data management follows from data generation to
knowledge creation. The Data Life Cycle incorporates planning and collection of
data, quality assurance, metadata creation, submission, preservation, discovery,
integration, analysis and publication. Many steps of the Data Life Cycle are not only
performed once, but multiple times or continually over the life cycle. The order of the
Data Life Cycle is also adapted to the needs of a research project. It can be
approached from different perspectives, such as data producer and data re-user. For
example, a data re-user does not collect data, and not every data producer integrates
data from other researchers.
There are practices and steps within the ideal Data Life Cycle where research has yet
to discover its potential, and which are worth adopting as routines (which ones have
you thought of?). Some practices and steps are normally carried out by specialists
called (digital) curators working at data repositories. Curation is managing digital
items in a storage to ensure long-term preservation. One step further is to make them
discoverable and accessible as soon as it is possible.
This reader provides best practices for every step of the Data Life Cycle. These
practices and activities are a suggestion. Every researcher should check what is
suitable and makes sense for her or his specific project and adapt the practices
accordingly.

For a short overview over the different steps of the Data Life Cycle, the fact sheets on
the GFBio Homepage are recommended (http://www.gfbio.org/data-life-cycle).

7
 Propose

Publish Collect

Analyse Assure

Integrate Describe

Discover Submit

Preserve

Figure 4: Data Life Cycle after GFBio

The Data Life Cycle can be understood as a part of the Research Life Cycle. The
Research Life Cycle (Figure 5) is a model for the steps followed in order to create
scientific knowledge. The Research Life Cycle, depicted in orange, starts with the
research idea and comprises the establishment of cooperation with research partners,
the composition of a research proposal, the granting of funding, the Data Life Cycle
(except the step “propose”, which is already included in the research proposal) and
finally the publication of research results. The generated scientific knowledge and
information serves as starting point for new research problems.

8
Figure 5: Research Life Cycle after GFBio

9
 3.1 Propose

Data management ideally begins at the planning and proposal phase of the research
project. This is the best moment to establish a Data Management Plan to provide a
framework that supports researchers and their data throughout the course of research
and to provide guidelines for everyone to work with (Mantra et al. 2014). Some
funders require a Data Management Plan. And it’s also possible to get funding (e.g.
from the DFG for archiving costs) for data management activities.
Table 1: Examples for components of a Data Management Plan. Michener and Jones 2012, supplemented

Component Description and examples

Information about Types of data that will be produced (e.g. experimental, observational, raw
data and data format or derived, physical collections, models, images, etc.)
Volume of data
When, where and how the data will be acquired (e.g. methods,
instruments)
How the data will be processed (e.g. software, algorithms and workflows)
File formats (e.g. csv, tab-delimited or naming conventions)
Quality assurance and control procedures used
Other sources of data (e.g. origins, relationship to one’s data and data
integration plans)
Approaches for managing data in the near-term (e.g. version control,
backing up, security and protection, and responsible party)
Metadata content Metadata that are needed
and format How metadata will be created or captured (e.g. lab notebooks, auto-
generated by instruments, or manually created)
Format or standard that will be used for the metadata
Policies for access, Requirements for sharing (e.g. by research sponsor or host institution)
sharing and re-use Details of data sharing (e.g. when and how one can gain access to the
data)
Ethical and privacy issues associated with data sharing (e.g. human subject
confidentiality or endangered species locations)
Intellectual property and copyright issues
Intended future uses for data
Recommendations for how the data can be cited
Long-term storage Identification of data that will be preserved
and technical data Repository or data centre where the data will be preserved
management Data transformations and formats needed (e.g. data centre requirements
and community standards)
Identification of responsible parties
Budget Anticipated costs (e.g. data preparation and documentation, hardware and
software costs, personnel costs and archive costs)
How costs will be paid (e.g. institutional support or budget line items)
Responsibilities Who is responsible for what?

Establishing a Data Management Plan at proposal stage in the Data and Research Life
Cycle facilitates a structured work with data and saves time later on. A Data
Management Plan means to plan all activities and things that should be considered
concerning the data foundation of the research project. It clarifies resources needed in
terms of money for long term preservation, or skills and software as well as
responsibilities and roles of stakeholders, project members and lab staff (e.g. will the
computing centre be involved?). A Data Management Plan is a living document that is

10
to be maintained and kept up-to-date, e.g. if staff changes. It is important to base the
plan on available resources and support to ensure that implementation is feasible.
Table 1 gives a first insight in what can be included in a Data Management Plan. A
very detailed list elaborated by the WissGrid-Project is available online. From such
lists, the suitable aspects can be chosen accordingly to the needs and characteristics of
the project.

Resources
WissGrid: Detailed checklist for creating Data Management Plans (in German).
http://www.wissgrid.de/publikationen/deliverables/wp3/WissGrid-oeffentlicher-
Entwurf-Checkliste-Forschungsdaten-Management.pdf

Online tools for creating Data Management Plans:

https://dmptool.org/
https://dmponline.dcc.ac.uk

Digital Curation Centre: Advices on DMP writing.

http://www.youtube.com/watch?v=7OJtiA53-Fk

Digital Curation Centre: How to include costs in Data Management Plans.

https://www.youtube.com/watch?v=nKeVPpupsYI&feature=c4-
overview&list=UULTOHF6qQrYhEvQzbu03tTg%00

UK Data Service: Data Management Costing Tool. http://www.data-

archive.ac.uk/media/247429/costingtool.pdf

11
 3.2 Collect

Ecological data can be collected in many different ways. Collection includes various
procedures such as manual recordings of observations in the laboratory or field on
hand-written data sheets as well as automated collection by data loggers, satellites or
airborne platforms (Michener and Jones 2012). Data created in digital form is “born
digital”, manually collected data are digitised later on. When collecting data, it is
helpful to think of subsequent steps of the Data Life Cycle: what is going to happen
with the data? In this way, data collection can be organised in a way that supports
following activities and saves time later on. Here are some tips for data collection
which are particularly important when several people are involved with data
collection and entry:

• Decide what data will be created and how - this should be communicated to
the whole research team.
• Be clear about methods.
• Use collecting protocols.
• Develop procedures for consistency and data quality.

There are many activities directly related to data collection. Apart from data entry and
file naming (which are discussed below), the choice of an appropriate software format
for the collected data has to be taken into consideration (see also 3.5 Submit). Many
different programmes are used for data collection, ranging from spreadsheets and
statistical software to relational database management systems and geographic
information systems (Michener and Jones 2012). Every software and format has
advantages and disadvantages, depending on what kind of analyses are planned,
software availability and costs, the hardware which is used to capture the data and
discipline specific standards and customs (UK Data Service 2014b). Also, the file
format for working with the data can differ from the formats used for storing and
long-term preservation.

Data entry
Spreadsheet software (like Excel) is very common for working with quantitative data,
especially during data collection and entry. Its advantages are that many people
already know working with it and first steps are quite simple. However, special care
should be taken when working with spreadsheets as they can be quite error-prone.
Excel can be a good choice for data entry, but use a syntax that allows information to
be stored without loss in csv-files (so that they can be easily accessed with other
programs e.g. for analysis). If spreadsheets are used, systematic and accurate work
from the beginning on facilitates the process of data exchange to other programmes
like R. Especially for analysis and data transformation, it is recommended to use
scripted environments like R. But you wouldn’t like to enter data in R.

12
For entering data, it is recommended to use codes like ASCII, UTF-8 or ISO 8859-1
(Latin1). These codes contain characters which can be read by most programmes
without any problems (in contrast to codes using e.g. umlauts). If a file is opened
using a wrong encoding, something like this can happen:

Bärbel Bürgenßen (2015): Encoding Problems in the City of Mörgäl.

Bärbel Bürgenßen (2015): Encoding Problems in the City of Mörgäl.

When you receive data where umlauts are incorrectly displayed, use e.g. the simple
notepad editor to change the encoding.

Spreadsheet structure
Some basics for spreadsheet file structure:

• Variable names without spacing, name variables consistently

• Use codes (ASCII, UTF-8), avoid umlauts
• Only one type of information in each cell (atomize values)
• Record full dates, standardize formats (recommended YYYY-MM-DD, allows
sorting)

Figure 6: Example for poor data entry. DataONE 2012d

Figure 6 shows an example for poor data entry in Excel. Three different trapping
periods of a project were entered in one table. There are many inconsistencies
between data collection events:
• Location of date information, inconsistent date format
• Column names
• Order of columns
• Different site spellings, capitalization, spaces in site names
• Mean1 value is in weight column
• Text and numbers in same column
A table structured in that way is hard to filter and to analyse. Data should be
structured as consistent as possible. Even if there are any errors, they can be fixed
much easier (via scripting) than if the data entry was extremely unstructured.

13
In Figure 7, a corrected version is shown. The entries are consistent now: only
numbers or dates or text was entered. Consistent names, codes and formats (date) are
used in each column. And data are all in one table, which is much easier for a
statistical programme to work with than multiple small tables which each require
human intervention. This record also underlines the importance of additional
information about the data (metadata). It is not apparent from the table what
measurement unit is used for weight, or what the species abbreviations mean. This
information can be given on a separate sheet or in the metadata documentation of the
dataset (see 3.4 Describe).

Figure 7: Example for good data entry. DataONE 2012d

Figure 8 shows more problematic data records. In the left column (red), the data
structure is inconsistent. If you use underscores or hyphens, stick to one mark and be
consistent in using it. In the right column (green) there are several values in one cell
as well as several units per variable were used. Only data of one unit should be
entered, additional information like the aspect can be entered in a separate column.

Figure 8: Example for problematic data records.

14
 Practical tips for data entry in spreadsheets
When entering data manually in spreadsheets, the data validation feature of Excel can
help to prevent data entry errors and detect erroneous values. The data validation
feature can be used to define what can be entered in a cell (e.g. characters, positive
values) and warns the user if any other content is entered. Also a valid range between
a minimum and a maximum value can be defined. The feature is accessed via Data >
Data Tools > Data Validation. Double data entry or controlling the data entry by
another person is another way to reduce data entry errors.

A challenge for exporting data or further analysing it is when there is more than one
dataset on one sheet. This is sometimes helpful for data entry, but makes subsequent
work difficult. Not only in this case it makes sense to separate data entry from the
dataset. This can be done by preparing a data entry mask e.g. in MS-Excel (Figure 9).
The data which is entered will be automatically saved in a data file.

Figure 9: Example for data entry mask in Excel. Dennis Heimann.

File Naming
As well as the file formats, also the file names matter. Naming files is important for
organising data on a lab’s network drive or personal hard disk and for identifying it
later. In data repositories the corresponding information is available in the metadata.
File names should be unique and use ASCII characters and avoid spaces. The latter
ensures that the file can be read by different operation systems and programs.

15
The amount and type of information in a file title varies, depending on the type and
amount of data and the projects requirements. The content of the file should be
reflected in the title. In Table 2 example file namings are displayed. The first name
has very little information, whereas the third title is already very long. Much of that
information could be documented in the metadata (project, place, time of collection,
time of processing, subject).
Table 2: Example file naming.

Title Information content

Water samples ???
Rhine_water_samples_20140901_V1.0 Rhine (where) water_samples (what)
20140901 (when) V1.0 (version status)
Ecoproject_2011_2014_Water_quality_ Additional information can be documented
Rhine_water_samples_Cologne_ in metadata
20140901_V1.0

Including a version number in the file name is a good idea to identify the most recent
file, be able to return to older versions and indicate that changes and transformations
have been executed on the data. In Table 3, an example for a versioning system used
by BExIS is given. Changes in metadata are indicated by the third digit, smaller
changes to the dataset by the second digit and major alterations by the first digit.

Table 3: File versioning in BExIS. BExIS How To: Version numbers in BExIS.

Resources
MANTRA Video: Jeff Haywood talks about the importance of good file management
in research. https://www.youtube.com/watch?v=i2jcOJOFUZg
Software Carpentry: Lecture on data management.
https://www.youtube.com/watch?v=3MEJ38BO6Mo&html5=True
New York University Health Sciences Library: How to avoid a Data Management
Nightmare.
https://www.youtube.com/watch?v=nNBiCcBlwRA00_SomeDataManagement_1_14
0902

UK Data Service: Formatting and organising research data.

http://ukdataservice.ac.uk/media/440281/formattingorganising.pdf

16
 3.3 Assure

Assure refers to quality control and assurance. The latter encompasses all those
activities which ensure the reliability of data. High quality data are a key element for
research and impact replicability of results. Quality checks should be performed
during collection, data entry and analysis and answer the following questions:

• Are the data complete?

• Are the data correct?
• Is the format consistent throughout the data set?
• If it contains errors, which errors?
• Are there missing values?

Quality assurance is already applied prior to data collection by defining standards for
formats, codes, units and metadata. Also the assignment of responsibility for data
quality is part of quality assurance (Michener and Jones 2012). In the validation
process it should be checked whether data are incomplete, unreasonable or inaccurate.
This can already be included in the data entry process (see 3.2 Collect). Statistical and
graphical summaries (e.g. max/min, average, range) help to check for impossible
values and outliers. After validation, the dataset is cleaned. This means to check
outliers, correct and fix errors.

Basic quality control

Quality control practices are specific to the type of data being collected, but some
generalities exist (DataONE n.y. Best Practices: Ensure basic quality control):

Data collected by instruments: Values recorded by instruments should be checked

to ensure they are within the sensible range of the instrument and the property being
measured. Example: Concentrations cannot be < 0, and wind speed cannot exceed the
maximum speed that the anemometer can record.

Analytical results: Values measured in the laboratory should be checked to ensure

that they are within the detection limit of the analytical method and are valid for what
is being measured. If values are below the detection limit, they should be properly
coded and qualified. Any complementary data used to assess data quality should be
described and stored. Example: data used to compare instrument readings against
known standards.

Observations (such as bird counts or plant cover): Range checks and comparisons
with historic maxima will help identify anomalous values that require further
investigation. Comparing current and past measurements help identify highly unlikely
events. For example, it is unlikely that the girth of a tree will decrease from one year
to the next.

Dates and times: Ensure that dates and times are valid. Time zones should be clearly
indicated (UTC or local).

17
Data Types: Values should be consistent with the data type (integer, character, date,
time) of the column in which they are entered. Example: 12-20-2000A should not be
entered in a column of dates. Use consistent data types in your data files. A database,
for instance, will prevent entry of a string into a column identified as holding integer
data.

Geographic coordinates: Map coordinates to detect errors.

(DataONE n.y. Best Practices: Ensure basic quality control)

Outliers
Outliers are unexpected values. They may not be the result of actual observations, but
rather the result of errors in data collection, data recording, or other parts of the Data
Life Cycle. To identify outliers, statistical tests can be used (Dixon’s test, Grubbs test,
Tietjen-Moore test). Another possibility is the visualization of data (Box plots, scatter
plots when there is an expected pattern, such as a daily cycle). A third way for
detecting outliers is the comparison to related observations. Difference plots for co-
located data streams can show unreasonable variation between data sources (Example:
Difference plots from weather stations in close proximity or from redundant sensors
can be constructed). Comparisons of two parameters that should covary can indicate
data contamination (Example: Declining soil moisture and increasing temperature are
likely to result in decreasing evapotranspiration) (DataONE n.y. Best Practices:
Identify outliers).
No outliers should be removed without careful consideration and verification that they
are not representing true phenomena. Although outliers may be valid observations it is
important to identify and examine their validity. Outliers may represent data
contamination, a violation of the assumptions of the study, or failure of the
instrumentation (DataONE n.y. Best Practices: Identify outliers).

Documentation of quality
Document in the metadata what quality checks were performed and what was their
result. The data quality can be communicated by flagging values, coding or labelling
of the dataset and in metadata documentation. For example, entire data sets can be
labelled 0 for unexamined, -1 for potential problems and 1 for "good data". Some
research communities have developed their own standard protocols (DataONE n.y.
Best Practices: Mark data with quality control flags). In Table 4, an example for
quality codes used by the University of Oregon for data about solar radiation is
shown. Similar to codes is the practice of file versioning. Version numbers indicate
that a data set was subject to any kind of transformation (see 3.2 Collect).

18
Table 4: Example for quality codes. http://solardat.uoregon.edu/QualityControlFlags.html

First digit Second digit

1 - Observed data
1 Raw data
2 Processed data
3 Possible problems in data
9 – Missing or bad data
9 Missing or bad data

Quality assurance also detects missing values. Missing values are common in
environmental data. They can occur if they were not collected, not analysed or got
lost. Impossible values are values which lie outside the range for a parameter (e.g. a
negative pH). The codes in Table 5 can be used to indicate missing values in a data
field. Especially “none” and “unknown” should be used meaningfully. “None” can be
a proper value (e.g. a “0” for precipitation) whereas “unknown” means that a value is
missing. The chosen code should be used consistently throughout the data set and be
documented in the metadata.
Table 5: Codes for missing values. DataONE (n.y.) Best Practices: Identify Missing Values

Code Use for

-999.99 Use only if this is an impossible value within the data set
Not applicable Character fields
Unknown
None
Pending assignment When additional information is compiled to complement value

Resources
Open Refine (tool to work with messy data). http://openrefine.org/

R (software environment for statistic computing and graphics). http://www.r-

project.org/

Matlab (high-level language and interactive environment).

http://uk.mathworks.com/products/matlab/

Finding the Antarctic Ozone Hole: This is an illustrative example how outliers or
values rated to be erroneous prevented NASA detecting the Ozone Hole.
http://www.statsci.org/data/general/ozonehol.html

Data Validation in Excel. https://support.office.com/en-za/article/Apply-data-

validation-to-cells-c743a24a-bc48-41f1-bd92-95b6aeeb73c9

19
 3.4 Describe

Additional information about data is called metadata. Metadata describe all aspects of
data (e.g. who, why, what, when and where) that would allow one to understand the
physical format, content and context of the data, as well as possibly how to acquire,
use and cite the data (Michener and Jones 2012). Sometimes it can be unclear if a
value is considered as metadata or as a record. For instance, for one research approach
the locational information of an observation or experiment is metadata, whereas in
another approach this is “primary” data directly underpinning research results.

Metadata can describe the data

• Title of data set
• how, where, by whom and when the data were collected: methods, dates,
persons, places
• data content and format
• the data itself: units, parameters, abbreviations and acronyms
• quality of data, version of data set
• information about the project

Metadata can describe the digital object

• Technical information about the file: is special software needed? What format
do the files have?
• Administrative information: restriction of re-use, how to cite

Metadata accompanying a data set should be written for somebody re-using them 20
years later: what information is necessary to use the data? Data and documentation
should be prepared in a way that is understandable for someone unfamiliar with your
project and methods. Descriptive and clear writing helps to understand metadata
easily. Do not assume that acronyms or abbreviations are eternally comprehensible.
Jargon, technical terms, acronyms and abbreviations should be explained. When data
are made available for sharing, metadata should give the user enough information to
use the dataset without contacting the author. Figure 10 illustrates the importance of
metadata: what is meant by RW or HW? What do the values indicate?

Figure 10: Variable names are hard to understand without metadata.

Metadata is ideally created as soon as possible to minimize the loss of information.

Metadata can be provided in simple text documents or spreadsheets where a key/value
approach is recommended. In spreadsheets two columns can be used, in a text
document, the information can be separated by a colon (e.g. author: John Doe;

20
taxonomic reference list: WoRMS). This schema is human and machine readable.
Especially when it is intended to submit data to a repository, it is helpful and often
necessary to use a metadata standard. Data without associated description are
impossible to archive, because not enough is known about the data to ascertain what
they are, or whether they are worth curation attention (DCC 2004-2015).

Tips for writing Metadata

• Keywords: It is recommended to use thesauri.
• Where: Fully qualify geographic locations. Think of using bounding box
coordinates defining an area by two longitudes and two latitudes for multiple
locations; name the corresponding coordinate system.
• Methodology: Document what you did (a published article may not give
enough information but a reference to it is a good starter).
• Data Quality: Clearly state data limitations and quality (e.g. data set
omissions, completeness of data).

Metadata Standards
Metadata can be made available in a simple text form, accompanying the data set.
Another possibility is to use a metadata standard. Metadata standards usually structure
the information using XML (Extensible Markup Language) and are both human and
machine readable. XML can still be viewed with any text editor but allows for easy
verification of syntax and partly content. There exist different standards for different
purposes, depending on the discipline. It may be tricky to find the right standard.
Librarians might be able to give advice. Some institutions also suggest or prescribe
metadata standards in their data policies. In Table 6, some example metadata
standards are listed.
Remember: a computer will read the metadata. This means you have to be very
careful using symbols. Do not use symbols that could be misinterpreted (! @ # % { } |
- / \ < > ~). Tabs, indents, line feeds or carriage returns could also be misinterpreted.
When copying and pasting from other sources, use a text editor to eliminate hidden
characters.
Table 6: Examples of metadata standards.

Metadata Standard Special focus URL

Ecological EML Ecological data http://knb.ecoinformatics.org/eml

Metadata _metadata_guide.html
Language
Darwin Code DWC Museum specimen http://rs.tdwg.org/dwc/index.htm
Dublin Core DC web resources, http://dublincore.org/documents/
Element Set publications dces/
ISO 19115 Geospatial data and http://www.fgdc.gov/metadata/ge
ISO 19139 services ospatial-metadata-standards#
fgdcendorsedisostandards

21
Access to Biological ABCD Biological collection http://www.tdwg.org/standards/1
Collection Data data 15/
Metadata Encoding METS descriptive, http://www.loc.gov/standards/met
and Transmission administrative, and s/
Standard structural metadata
for objects within a
digital library
DataCite Metadata Publication of data http://schema.datacite.org/
Schema http://schema.datacite.org/meta/k
ernel-3/example/datacite-example-
full-v3.1.xml

Resources
Morpho (Programme to enter metadata, stored in files conform to EML).
https://knb.ecoinformatics.org/#tools/morpho

ISA-tab (Tool for creating metadata). http://isatab.sourceforge.net/format.html

Digital Curation Centre: online catalogue of disciplinary metadata standards.

http://www.dcc.ac.uk/resources/metadata-standards.

22
 3.5 Submit

Submission is the transfer of data to a curated environment. This is usually an archive,

a data centre, a repository, or a collection. Submission to a curated environment
ensures safe long term storage and makes data discoverable for other researchers (in
the team or outside). Please note: discoverable does not necessarily imply
immediately accessible, as many repositories allow for limited (commonly 6-12
months) embargo times. During submission phase, researchers can decide on how the
data can be accessed. In any case, if submitted to an archive or repository, the data
will be preserved and curated. Access may vary from immediately available for re-
use, available with restrictions or an embargo or not accessible for others. See Table 7
for different restriction possibilities on data. If restriction applies make sure that
contact details are up to date. The access to some kind of data may be restricted due to
its content to protect individuals or interests (confidential information about persons,
ethical issues, environmental protection, endangered species, intellectual property,
laws, security) (Downs 2013). Contact you library or the targeted repository for
possible solutions for anonymisation or secured off-line storage.
Table 7: Restrictions for data. Downs 2013

Access and use Dissemination and copies

Limit who may access data or how they are May not be authorized to re-distribute or
used (authorized users) copy data
Authorize use only for specific purposes, Limit distribution to a specific location or
such as education service
Limit whether data may be used in new Limit distribution to a specific time period,
products or services possibly in the future
Modifications or derivations of data may be May apply to any products created from
prohibited data

If data are submitted to a data centre, the files receive a persistent identifier. These
identifiers are independent of an URL, which might change over time. A common
persistent identifier is the Digital Object Identifier (DOI). It consists of numbers and
letters. The DOI can be assigned to an internet location, the preferred is
http://dx.doi.org/ + DOI = URL (ESRC 2012). Figure 11 shows an example URL
containing a DOI. The registration agencies are in charge of administrating DOI
registers and resolving DOIs to actual URLs. For research data the agency is
“DataCiteConsortium”, which is composed of libraries and data centres around the
world. TIB German National Library of Science and Technology in Hanover
(Germany) is the leading organization for the DataCite Consortium.

Figure 11: Elements of a Digital Object Identifier (DOI). ESRC 2012

23
 How to submit data
• Use appropriate repositories and data catalogues (see links below).
• License the data so it is clear how they can be re-used (see 3.10 Publish).
• Make sure it’s clear how to cite the data (see link below).

What to keep?
Often, it’s not necessary or possible to keep everything. Decision criteria should be
documented.

• Which data set or digital resources (e.g. the script/source code employed in the
analysis) do you want to keep?
• Which elements or characteristics of those data sets or digital resources do you
want to keep?
• What do you need to keep to support you research findings?
• What has to be kept (e.g. data underlying publications)?
• What can’t be recreated (e.g. environmental recordings)?
• What is potentially useful to others, to whom, what should they be able to do
with the data?
• What has scientific, cultural or historical value?
• What must be destroyed due to legislation?

(DCC 2010, 2014)

File formats
It is worth thinking about software and file formats when preparing data for long term
storage. Proprietary software may become obsolete, or newer versions won’t open
older files anymore. Additionally, open formats and standards facilitate exchange and
can often be handled with the programme of choice. Also, ASCII formats should
usually be preferred over binary formats. For instance, R changed its internal binary
storage format and it is likely that reading/importing such data will incur additional
work at best and render re-use impossible at worst. In Table 8, recommended file
formats are listed for different types of data. Some formats are suitable or even
necessary to work with during collection or analysis phase, but are not recommended
for storing and archiving data (e.g. Excel). Recommended formats either do not need
special software or are fully and openly described. In any case, you should keep the
raw data in a format that can be easily processed.

24
Table 8: Recommended file formats for storing and sharing data. UK Data Service 2012-2015c

Type Recommended Avoid for data sharing

CSV (Comma separated values), TSV Excel
Tabular data
(Tab separated values)
Plain text, HTML, RTF Word
Text PDF/A only if layout matters
Container: MP4, Ogg Quicktime
Media
Codec: Theora, Dirac, FLAC H264
Images TIFF, JPEG2000, PNG GIF, JPG

Structured data XML, RDF RDBMS

Resources
Repositories:
PANGAEA (cooperation with GFBio): http://www.pangaea.de/
Dryad: http://datadryad.org/
Figshare: http://figshare.com/
Symbiota (biology): http://symbiota.org/docs/
VertNet (biodiversity): http://www.vertnet.org
Morphbank (biological images): http://www.morphbank.net/
iDigBio (Integrated Digitized Biocollections, specimen): https://www.idigbio.org

Re3data: Registry of research data repositories. http://www.re3data.org/

Five steps to decide what data to keep (DCC): http://www.dcc.ac.uk/resources/how-

guides/five-steps-decide-what-data-keep

How to cite Data sets (DCC):http://www.dcc.ac.uk/resources/how-guides/cite-

datasets

Data Deposition Decision Tree (University of Oxford):

http://researchdata.ox.ac.uk/files/2014/01/Data-deposit-decision-tree.pdf

Nature (http://www.nature.com/authors/policies/availability.html):
Unstructured repositories like figshare and Dryad are suitable alternatives if no structured
public repositories exist.

When repositories do not exist for a particular data type, authors can deposit and share data
via figshare or Dryad, two general-purpose scientific data repositories.

PLOS (http://journals.plos.org/plosone/s/data-availability):
PLOS requires that authors comply with field-specific standards for preparation and
recording of data and select repositories appropriate to their field, for example deposition of
microarray data in ArrayExpress or GEO; deposition of gene sequences in GenBank, EMBL
or DDBJ; and deposition of ecological data in Dryad. Authors are encouraged to select
repositories that meet accepted criteria as trustworthy digital repositories.

25
 3.6 Preserve

Digital data are fragile. Hardware fails, software becomes obsolete, files are subject to
bit rot which produces bit errors. Digital preservation as one part of digital curation
encompasses a set of actions which ensure long-term usability by maintaining the
accessibility, integrity and longevity of data:

• Longevity: extend the lifespan of data for current and future user requirements
• Integrity: ensure the authenticity and reliability of data (no undocumented
manipulations)
• Accessibility: store data in formats which ensure their future use

Preservation protects against data loss and obsolescence. Migration and emulation are
principal preservation actions (see below). These actions are carried out by
specialized curators. Researchers and project staff can also carry out some measures
by themselves. At an early stage, researchers should ask themselves what they want
others to be able to do with their data as this will impact many aspects of preservation
(DCC 2009). Data can be stored or migrated to a suitable format, avoiding proprietary
file formats (see 3.5 Submit). Also, a suitable medium should be chosen with care.
The lifespan of digital storage media have to be considered. A burned CD may only
last some years and USB-sticks get lost easily. Backups and storage as short-term
preservation is the duty of the research staff (see below). Also responsibilities have to
be taken into account. Who looks after stored data, who is responsible for transferring
data to a curated environment? The creation of metadata and documentation of
preservation actions helps to understand and use data later on. Significant properties
of data recorded in the metadata help the people carrying out preservation actions.
The actual archiving and long-term preservation actions are performed by specialists
like professional curators (DCC 2008d).
Be critical when reviewing “best practices”. They might works for specific scenarios
but not for you. For support, consult and work with experts in your field. Data centres
are there to offer solutions, help and professional curation of data.

Backup
Backup of data is not the same as preservation. Backups are short-term recovery
solutions whereas preservation includes measures taken for long term storage and
archiving. For backups, there exist ideally at least 3 copies of a file, on at least 2
different media, with at least 1 offsite. Managed services like university drives are
always a better choice than external hard drives or USB-sticks, but make sure you
know the conditions as e.g. normal network drives may keep backup copies of a file
for only 6 months and if overwritten with messy data and looked at after 7 months
you may run into trouble. The IT-Team of your institution might give support and
advice.

26
 Digital preservation methods in data curation
Preservation actions aim to ensure that data are in the best possible shape to be stored
or archived by a repository. The proceeding steps of the Data Life Cycle (3.3 Assure,
3.4 Describe) are crucial for this.

Migration

Migration refers to the transformation of data or other digital material from one
format or technology to another (software or hardware). The ability to retrieve,
display and use the contents is maintained by this action (DCC 2008b). Files can be
migrated within one software product to a newer version or to other file formats when
obsolescence occurs. Every migration changes data a little bit. If migrations are
carried out multiple times, data can be subject to major changes. Chris Rusbridge
describes a commonly- experienced migration result:

So if I migrate from those PowerPoint 4 files to today’s PowerPoint, and then

from today’s to tomorrow’s PowerPoint, and then from tomorrow’s to the next
great thing, I will introduce cumulative errors whose impact I will only be able to
assess at some horribly cringe-making moment, like in the middle of a
presentation using a host’s machine. So the best way to do migration is to start
from the original file and migrate to today’s version (DCC 2008b).

Because functionality is lost and integrity comprised as a result of migration, care

must be taken to strict quality checking procedures, for example, to compare the
original bit stream and the migrated bit stream. To check data integrity and detect bit
errors, checksums can be used. A checksum algorithm calculates a value which can be
compared to the value of the original file. A widely used algorithm is MD5.

Refreshment
Media refreshment refers to copying data to a new medium after a fixed time (e.g. 5
years) to preserve the bit stream and consider the life span of digital storage media.

Emulation
Emulation refers to the development of software that can mimic (obsolete) systems on
current and future generation of computers. By emulating applications, operating
systems and hardware architecture, older files and programmes can still be used and
software can be kept alive (e.g. using a 16 bit application like Harvard Graphics on a
DOS System) (DCC 2008c).

Resources
University of Cambridge: Preservation.
http://www.lib.cam.ac.uk/dataman/pages/preservation.html

27
UK Data Service: Store your data. http://ukdataservice.ac.uk/manage-
data/store.aspx

28
 3.7 Discover

Discovering data means to search and find data collected by other researchers. This
data can be used for different purposes like long-time analysis, modelling or
comparative studies. Data and metadata are made accessible through submission on
any kind of shared environment. The pre-requisite for discovering data is that the
authors are willing to share their data with the research community. Michener and
Jones (2012) name the problem:
Many valuable and relevant data products are not readily available as they are
stored on laptops and computers in the offices of individual scientists, projects
and institutions.

So, only submitted and published data can be discovered. Specialized data archives,
centres and repositories offer search functions to discover relevant data. Another way
to discover data is via journals, when data are appended to articles. Often it is
necessary to register at a repository or archive to get access to the data. A persistent
identifier is important to locate the data, e.g. if the URL changes. The persistent
identifier is also an important element for citing data which were discovered and used.
GFBio offers a faceted search tool that searches the associated data centres (Figure
12). In the field at the top, any word can be entered. When the results of the search are
displayed, a search facet appears on the left. Here, you can specify your search by
choosing authors, publication years, geographic regions and data centres. The search
results are displayed geographically by clicking on the trolley symbol.

Figure 12: GFBio search screenshot. http://www.gfbio.org/archives/-data-centers

29
 Checklist

 Are data accessible, are there any use restrictions?

 Do appropriate metadata exist for citing, understanding and evaluating the
data?
 Is the provenance of the data visible?

If you want to make your own data available and discoverable, it is indispensable to
create metadata to communicate significant properties of the data. Consider if there
should be any access or use restrictions (DCC 2008d).

Resources
GFBio search function: http://www.gfbio.org/data-portal

PANGAEA (cooperation with GFBio): http://www.pangaea.de/

knb (Knowledge Network for Biocomplexity): https://knb.ecoinformatics.org/#about

Thredds Data Server (data server which provides metadata and data access for
scientific data sets): http://www.unidata.ucar.edu/software/thredds/current/tds/

ModisLand (land cover satellite data): http://modis-land.gsfc.nasa.gov

Dryad (especially natural and environmental sciences):

http://datadryad.org/discover?query=&submit=Search#

Global Spatial Data Infrastructure (Links to geographic data portals):

http://www.gsdi.org/ElectronicGateways (international)

NOAA US National Oceanic and Atmospheric Administration Data Center:

http://www.ngdc.noaa.gov/ngdcinfo/onlineaccess.html

NASA Global Change Master Directory: http://gcmd.nasa.gov/index.html

Symbiota (biology): http://symbiota.org/docs/

VertNet (biodiversity): http://www.vertnet.org

Morphbank (biological images): http://www.morphbank.net/

iDigBio (Integrated Digitized Biocollections, specimen): https://www.idigbio.org

BISE Biodiversity Information System for Europe:

http://biodiversity.europa.eu/data

GBIF Global Biodiversity Information Facility: http://www.gbif.org/

Re3data: Registry of research data repositories. http://www.re3data.org/

30
 3.8 Integrate

Integration is the merging of multiple datasets from different sources, like your
recently collected data with former data from other owners, resulting in a new, bigger
dataset. You might want to integrate a dataset that you discovered and that fits to your
own data in order to verify your results, as a starting point for an integrative study or
just to test a new hypothesis for a follow-up study. Large-scale ecological studies
require the integration of data from different studies and disciplines (e.g. population
studies, hydrology and meteorology; Michener and Jones 2012). Integrating data from
different sources is
labor intensive and time consuming, because it requires understanding
methodological differences, transforming data into a common representation, and
manually converting and recoding data to compatible semantics before analysis
can begin. Data integration for crosscutting studies is generally a manual
process and can consume the majority of time involved in conducting
collaborative research […] (Michener and Jones 2012).

For an effective integration it is important that in the early steps of the Data Life
Cycle, good data management practices were applied. That means for example that
metadata about the dataset were created, like information about the quality of the data,
measurement methods or abbreviations. This allows to assess fitness of data for re-
use. When other authors’ data are re-used, it is fundamental to provide credit to the
data creators through a robust data citation mechanism.

Data can be integrated manually or automatically. It is important to document the

integration workflow, all the steps used for cleaning, analysing and visualizing the
data and new data products so that the output is comprehensible.
There are common problems when integrating data that complicate database
compilation. These include for example the lack of a consistent coordinate reference
system, taxonomic naming inconsistencies, and the semantics of variable names
(Rüegg et al. 2014). If integrating datasets, ensure that formats and parameters are
compatible (datum, resolution, metric units) and assess the quality of the data.

Document the integration of multiple datasets

Ideally, mechanisms to systematically capture the integration process are adopted, e.g.
in an executable form such as a script or workflow. Another possibility is to document
the process, scripts, or queries in the metadata or any other accompanying document.
Also describe the relationship between datasets from different sources.
When you use data from other sources than your own collection, it is important to
indicate this. Citing allows for tracing the chain of use of datasets and data elements,
credit the creators of the data as well as the possibility that if errors or new
information about the original datasets or data elements comes to light, that any
impact on your new datasets and interpretation of such could be traced (DataONE n.y.
Best Practices: Document the integration of multiple data sets).

31
 Data Citation
Data have to be cited like bibliographic resources have to be cited in publications. But
there are some differences between data and publications (Mayernik 2013):
• Data sets are extremely variable in structure, representation schemes, and
access mechanisms, unlike traditional scholarly publications.
• Data sets can be dynamic (changing or growing) whereas scholarly articles do
not change once published.
• Data sets are often hierarchical composites of data and metadata files,
software, and other related documents.
• Authorship is a problematic notion in relation to data sets, particularly in
distributed and collaborative work.
However, data which are integrated from other sources can be cited. Figure 13
displays an example citation with all properties which should be included in the
citation of a data set (ESRC 2012; Mayernik 2013). If available, also more
information can be included, like the type of resource or a version number.

Figure 13: Example citation

When you are a data user, it is useful to provide feedback about the datasets you have
used. By providing feedback it is possible to reduce repetition of mistakes, improve
data quality and provide chances for further publications (Duerr 2011).

Resources
Digital Curation Centre (DCC): How to Cite Datasets and Link to Publications.
http://www.dcc.ac.uk/resources/how-guides/cite-datasets

Digital Curation Centre (DCC): Data Citation and linking.

www.dcc.ac.uk/resources/briefing-papers/introduction-curation/data-citation-and-
linking

Data Citation what you need to know.

www.esds.ac.uk/news/publications/data_citation_online.pdf

32
 3.9 Analyse

Analysis comprises the actions used to derive and understand information from data.
The types of analyses depend on the discipline and on the research questions to be
answered. Furthermore, the software and hardware used to analyse data also vary.

Statistics and visualisation

Statistics are one of the most common types of analyses used for quantitative data.
They comprise for example (multivariate) analyses of variance or regressions.
Conventional statistics tend to rely on assumptions such as random error and
homogeneous variance. Descriptive statistics are traditionally applied to observational
data. Descriptive data might include the distribution of organisms in space or
community-habitat relationships. Statistics used to understand these types of data
include diversity indices, cluster analysis, and principle components analysis, among
many others. Statistical analyses might also include temporal or spatial analyses, and
nonparametric approaches which do not rely on specific assumptions about the data’s
distribution. Other kinds of data analyses are simulations, model runs, parameter
estimations and visualisations (also compromising plots and graphs; DataONE
2012b). Visual representation allows to detect patterns within the data. Plots can be
used to assure the quality of data as well. They can quickly show you potential data
errors such as impossible values. In Figure 14, on the left is a scatter plot of
temperatures for the month of August. The general pattern is easily discernible, and
particularly warm measurements are readily apparent. On the right is a box and
whisker plot of monthly temperatures. The boxes indicate averages, and
measurements far from the averages are visible as red dots outside of the error bars.

Scatter plot of August Box and whisker plot of temperature by

temperatures month

Strasser, unpub. data Strasser, unpub. data

Figure 14: Examples of visualisations. DataONE 2012b

Before starting to analyse the data, it can be necessary to process it to be more usable.
Processing can include for example selecting a subset, merging multiple datasets or
reduce the amount of data. It can also be necessary to transform data prior to analysis.
That includes the conversion of values to common units or the adjustment of data

33
collected by multiple people. Analysing very large datasets requires special
considerations. Data of interest have to be discovered and identified. Often high-
performance computing systems are necessary to process big amounts of data
(DataONE 2012b).

Documentation
Data analysis often consist of several steps. Analyses are run, hypotheses modified,
data transformed and complemented, and again analysed. This complex process can
be hard to reproduce if it is not properly documented. The documentation of data
cleaning, transformation and analysis is called process metadata. Related to process
metadata is the concept of data provenance, which explains the origins of the data at
hand. In order to make data provenance clear, process metadata allow others to follow
the data through their life cycle and understand all steps used to create results and
output. In other words, the scientific procedure is described by the process metadata.
Good provenance allows for the ability to replicate analyses and reproduce results
(DataONE 2012b).
Process metadata can be documented in informal or formal workflows. A workflow is
essentially higher-level metadata. It is a conceptualized series of data ingestion,
transformation, and analytical steps. Informal workflows consist of diagrams or
flowcharts, guiding you through data processing. In Figure 15, a flowchart is
displayed. The green boxes in the middle indicate the transformation rules (what you
are doing to your data), and the blue boxes on the left and right represent the actual
data. First, temperature and salinity data are imported in R. The output is the data in R
format. These data are then controlled and cleaned, resulting in “clean” data. These
clean data are the input for the analysis step. The output, the summary statistics, is
used to produce a graph.

Figure 15: Informal workflow in form of a flowchart. DataONE 2012b

Another type of an informal flowchart would be commented scripts. These can be

applied when working with scripted languages, e.g. R. A commented script is a text
file containing the used commands and additional information, like descriptions of the

34
commands. The advantage of using R or Matlab for analysis instead of Excel is that
you can produce a script and therefore know exactly what transformations the data
have gone through. Michener and Jones (2012) state:
Ecologists have traditionally used tools such as Excel to manipulate and convert
data manually for integration; however, this process is error-prone and is not
reproducible because of the lack of provenance regarding these operations.
Scripted analysis environments, such as R and Matlab, improve this by providing
a record of data manipulations, but are still largely a record of procedural
manipulations of data.

Besides informal workflows, also formal ones exist. Formal workflows are produced
using a specialized software like Kepler or VisTrails. They allow to integrate different
software systems in one workflow. In the process of performing analyses, each step
and its parameters and requirements are formally recorded (DataONE 2012b). This
allows both the individual steps and the overall workflow to be re-used, either by the
original scientist or someone else. The open-source and free cross-platform
programme Kepler uses a drag-and-drop interface for scientists to construct their
workflow. There exist pre-programmed components e.g. for R for easy application
(Kepler 2008). Additionally, Kepler provides access to data repositories, computing
resources and workflow libraries (Kepler 2008).

Resources
Data Pub articles about workflows (Blog by California Digital Library):
http://datapub.cdlib.org/2012/06/05/workflows-part-i-informal/
http://datapub.cdlib.org/2012/06/12/workflows-part-ii-formal/

Kepler: https://kepler-project.org

VisTrails: http://www.vistrails.org/index.php/Main_Page

35
 3.10 Publish

The last step of the Data Life Cycle deals with the publication of data, especially with
datasets linked with the publication of related academic papers. It is based on the fact
that datasets are a fundamental part of the research process, as important as
discussions and conclusions derived from them (AGU 2012):
The scientific community should recognize the professional value of such [data]
activities by endorsing the concept of publication of data, to be credited and cited
like the products of any other scientific activity, and encouraging peer-review of
such publications.

Furthermore, publication of datasets promotes transparency in the research cycle and

facilitates verification and reproducibility. Therefore, it is important to establish
mechanisms which allow sharing datasets while at the same time promoting rewards
mechanisms for authors, as journals do for academic publications.
More and more academic journals (e.g. Nature, PLOS) include the publication of the
datasets within their publication requirements. Publishing research data also brings
advantages to the researcher. The work becomes more visible and the citation rate
may increase.

Best practices for publishing and sharing data

• Work with journal publishers and data repositories to archive data during the
publication process.
• Information about how to access your data set can be published with your
article. Data repositories typically offer to embargo archived data for a pre-
determined time period after publication.
• Archived data sets should get a persistent identifier (DOI) through the holding
data centre so that they can be uniquely cited.
• Encourage other data authors to cite data and to make their own data available
for re-use.
• Provide full citation information for data whenever you publish work that
makes use of another author’s data.
• Archive your own data in a repository that supports data discovery and re-use.
• Update your archived data sets when newer versions are available.
(DataONE 2012c)

Where to publish your data

There are many different possibilities where to publish your data. There exist a lot of
data centres, often specialised on one discipline or a field of research (e.g. GFBio for
biological end environmental data). See the resource list in chapter 3.7 Discover for
data centre suggestions. Specialized data journals offer the possibility to publish short
papers cross-linked to and describing a data set which is deposited in a data centre. If
a dataset is published in relation to a paper, the journal might prescribe a data centre
where to archive the data.

36
 Data publishing process
The following steps are a general overview about how the deposition of data in the
UK Data Archive works (UK Data Archive 2002-2015):
1. Data transfer: depositor transfers data to the archive
2. Data processing: curators (in cooperation with depositors) check data against
the documentation supplied, prepare archival and dissemination versions and
document processing
3. Metadata creation: a catalogue record based on specific standards is created
4. Additional user information: additional information is provided e.g. by a “read
me”-file
5. Publishing data: after data and documentation are complete, material is
transferred to preservation systems. The catalogue record is published and the
data becomes available to users.

Licences
Published data is ideally licenced. By this, it is clear for others how they can use the
data and what they are allowed to do with them. Two common licences are described
below.

Creative Commons Licences: http://creativecommons.org/

As data do not underlie the copyright, the Creative Commons Licence Zero CC0
applies. It means that no rights are reserved and data may be re-used and enhanced.
Nonetheless, appropriate citation can be requested. Data repositories like PANGAEA
or Dryad publish data under the CC0 licence.
For data other than raw data like figures, media, posters and papers, the CC-BY
licence can be used. This licence states that the work has an author. It allows to use
and enhance as long as the author is credited.

Open Data Licences: http://opendatacommons.org/licenses/

Open Data Commons offers three different licences:
Public Domain Dedication and License (PDDL): You are free to copy, distribute and
use the database; to produce works from the database and to modify, transform and
build upon the database without restriction.
Attribution License (ODC-By): You are free to copy, distribute and use the database;
to produce works from the database and to modify, transform and build upon the
database as long as you attribute as specified in the licence.
Open Database License (ODC-ODbL): You are free to copy, distribute and use the
database; to produce works from the database and to modify, transform and build
upon the database as long as you attribute, share alike and keep the data open (no
technological measures to restrict the work).

Sharing
Sharing and publication of data have already been discussed previously in this reader
(Chapters 2, 3.5). Its advantages were explained and possible barriers discussed, some
of them which are hard to solve. But there are also some less serious opinions against

37
sharing and publication which can be resolved. See Figure 16 for some arguments
from people reluctant to data sharing. But there are answers and solutions at hand,
many of them discussed in the proceeding chapters. So, go for it, share your data and
enjoy the benefits (Figure 17)!

Figure 16: Solutions to doubts about data sharing. Own design, after Corti et al. 2011 and Open Data Bingo

38
Figure 17: Benefits of sharing data. GFBio-Postcard realised by Seppo

Resources
Korn, N., Oppenheim, C. (2011). Licensing Open Data: A Practical Guide.
http://discovery.ac.uk/files/pdf/Licensing_Open_Data_A_Practical_Guide.pdf
Digital Curation Centre: How to license research data.
www.dcc.ac.uk/resources/how-guides/license-research-data

Research Data Netherlands: Addressing a researcher's data sharing concerns.

https://www.youtube.com/watch?v=8qRLgQa1wT4

Data Pub article about sharing excuses (Blog by California Digital Library):
http://datapub.cdlib.org/2013/04/24/closed-data-excuses-excuses/

Data Journals:
Nature Scientific Data: http://www.nature.com/sdata/
Biodiversity Data Journal: http://biodiversitydatajournal.com/
UP meta journals (Journal for meta papers): http://metajnl.com/
Geoscience data journal:
http://onlinelibrary.wiley.com/journal/10.1002/%28ISSN%292049-6060

Re3data: Registriy of research data repositories. http://www.re3data.org/

39
 4. Data Management with BExIS, version 1

BExIS is a platform for data storage and information exchange in scientific projects,
especially aimed at joint research projects, collaborative research centres or research
training groups. It provides central storage, access and exchange facilities for data,
with a focus on tabular data.
In this chapter, some functionality of BExIS is described. More detailed descriptions
and How-to-guides are distributed with BExIS-instances.

Metadata
The first action in the process of uploading data in BExIS is to create metadata. This
ensures that no data can be uploaded without corresponding metadata. The metadata
created by BExIS is compatible to EML. Nine steps, each on a separate page, guide
you through metadata creation:
1. general information: title of the dataset and information about the owner
2. research objects
3. methodology: objectives of the project, research methods
4. description: acronyms and keywords
5. time: set format of time and date used in the data set, start and end of data
collection
6. data: all information to understand the data set, data quality, data structure,
variables (name, type, units, description)
7. references: links to additional documentation, publications
8. comment
9. validation

Especially the description of the data in step 6 is important. The structure of the data
set defined here will determine what data can be uploaded later on. If the data is
structured data, BExIS reads the data while uploading them and compares them to the
predefined structure. This function is part of quality assurance as it helps to ensure
that data are consistent and detects erroneous values (if they have another type). If for
example a variable is defined as integer number, the upload of a dataset containing
characters in a field of that variable will fail.

Upload structured and unstructured data

Once a metadata entry is generated, data can be uploaded. In the metadata entry, the
file type to be uploaded is already determined. Structured data allow for filtering or
exporting selected parts. The data are stored in spreadsheets, csv or tsv. The file type
unstructured data is intended for the upload of documents, models, images and other
data with no tabular arrangement.

40
 Edit structured data
Single records in structured data sets can be directly edited in BExIS using the online
interface. Extensive transformations are better carried out by downloading the data,
manipulating it with a programme of choice and then uploading it again. BExIS
automatically creates a new version number if any changes have been performed on
the data set.

File Versioning
BExIS generates file version numbers automatically. The versioning system is
explained in Figure 18.

Figure 18: BExIS versioning system. BExIS How To: Version numbers in BExIS.

Access data and right management

BExIS is designed for an easy exchange of data. Metadata uploaded in BExIS is
visible to everybody. BExIS is conservative with access rights. Data is by default only
visible to the owner and available to others on request. The owner can decide to make
data available to a certain group or the entire project by asking the administrator to
grant access rights. The data owner gets an overview about who downloaded the data
and when.

Download data
Download of data is only possible when you have appropriate access rights. If you
download any data you will receive an automatically generated email once a month if
there have been any changes or new versions of the data set. If somebody downloaded
your data, you will get a notification E-Mail. Under Upload > Owner Info you can
see, who downloaded the data.

41
 5. Data Management at a glance: Summary

Data management is about work with research data in a structured, efficient, and
conscious way, all throughout the 10-step Data Life Cycle:
1. Propose: To get the most benefit out of data management, start early. Establish
a Data Management Plan at proposal stage.
2. Collect: Core activity for generating data. Take special care with data entry in
spreadsheets. Organize files in a structured way, optimized for analysis. Avoid
redundancies.
3. Assure: Assure data quality prior to analysis. Document quality using data
labels and value flags. Also transformations should be documented so people
can judge if data are fit for use.
4. Describe: Metadata makes data understandable (values, parameters,
methods…) and findable. It also provides information about the creator, which
is necessary for correct citation.
5. Submit: Submit data to a curated environment, especially for long-term
storage and for sharing and publication.
6. Preserve: Back-ups and short-term storage within the project to secure data.
Long-term storage and special preservation actions are carried out by
repositories.
7. Discover: Search, find and include already existing data to enrich your
research.
8. Integrate: Merging of data sets, data may be created by others. Special focus in
this phase lies on compatibility and citing.
9. Analyse: Statistical, visual and other analyses to derive information from data.
10. Publish: Publish a data set individually or underpinning a publication.

The way we work with data is currently undergoing a period of change. Access to and
re-use of data is a growing concern of funding agencies the world over. The digital
transformation causes a deluge of data as well as collaboration capabilities and
changes the way research is conducted. What proper data management should look
like in the future will very much depend on the discipline and research approach.
In this handbook we tried to cover basic concepts of proper data management,
including practical advices, following the Data Life Cycle. Success will hinge on how
well introduced concepts can be integrated into day-to-day research. It will depend on
both an understanding of researchers about data management and data management
providers about research processes and needs. Tangible benefits of data management
are already evident in form of security, compliance, access, quality and efficiency and
we hope that interest in and adoption of data management will further grow among
the scientific community.

42
 6. More Data Management: Recommended further reading

German Federation for the Curation of Biological Data (GFBio):

http://www.gfbio.org
Provides access to data, tools for visualisation, integration and management of data,
as well as training resources, like fact-sheets for each step in the Data Life Cycle
(http://www.gfbio.org/fact-sheets).

Digital Curation Centre (DCC): http://www.dcc.ac.uk/

Centre of expertise in digital curation, with a focus on building capacity, capability
and skills for research data management across the UK's higher education research
community. Provides a lot of detailed information material, including checklists,
briefing papers, case studies and workshop materials for professional curators.

UK Data Service: http://ukdataservice.ac.uk/

Guides, advice and training on data management, with focus on social data.

UK Data Archive: http://www.data-archive.ac.uk/

Collection of digital research data in social sciences and humanities, with information
about curation und data management.

Handbook on Data Management:

Corti L, Van der Eynden V, Bishop L, Morgan-Brett B (2011) Managing and sharing
data: training resources. UK Data Archive (2014 edition published at SAGE)
http://repository.essex.ac.uk/2398/1/TrainingResourcesPack.pdf

Van der Eynden, V (2014): Managing and sharing your research data. UK Data
Archive http://ukdataservice.ac.uk/media/455282/dm_stirling_2014.pdf (extensive
presentation corresponding to the handbook)

Data Observation Network for Earth (DataONE): https://www.dataone.org/

A US based cyberinfrastructure or the environmental science, huge best practice
collection and education modules prepared in presentations and fact sheets.

DataONE Best Practice Primer:

https://www.dataone.org/sites/all/documents/DataONE_BP_Primer_020212.pdf
Primer which gives a first overview about best practices in data management.

Mantra: http://datalib.edina.ac.uk/mantra/
Research Data Management Training online course by the University of Edinburgh.
Includes material for students, researchers and information professionals as well as
data handling tutorials in R and ArcGIS.

43
 7. Glossary

Access Possibility to search and retrieve data. Covers also rights to download or re-use data

Archiving To place or store data in a data centre or otherwise managed environment; typically done
for ensuring long-term preservation of the data and to promote discovery and use of the data
(Strasser et al. 2012). Also refers to a curation activity which ensures that data is properly
selected, stored, can be accessed and that its logical and physical integrity is maintained over
time (RDA Europe 2014b).

Big data Data sets encompassing a big amount of records, normally collected automatically by
satellites, remote sensing, instruments or sensor networks. Often personal computers cannot
handle such data, grid computing or clouds are used. The term Big Data is not defined by a
specific quantity but “is often used when speaking about petabytes and exabytes of data”
(NIST 2015:10 after Dutcher 2014).

Bit stream Sequence of bits that is identified as a unit, constituting a digital object.

Citable data Citable data is a type of referable data that has undergone quality assessment and can be
referred to as citation in publications (RDA Europe 2014b).

Curation Work of specialised curators to ensure good management of data, especially ensuring
preservation and promoting re-use.

Data Information used for research.

Database An organized collection of data. A database can be classified by the type of content
included in it (e.g. bibliographic, statistical, document-text) or by its application area (e.g.
biological, geological; Strasser et al. 2012).

Data catalogue Curated collection of metadata about datasets (RDA Europe 2014b).

Data centre (see also repository) Facility specialized in the professional work with data (e.g.
acquiring data from providers and make it available to re-users), can be a repository, archive
or other kind of centre. Data centres also offer e.g. user help desk support and training,
support data processing activities and other services.

Data citation A reference to data created by others, used analogical to bibliographic references like
citation of journal articles.

Data documentation Metadata or information about data that enables one to understand and use the
data. Comprises also the documentation of data transformations and analyses (process
metadata; Strasser et al. 2012).

Data element A logical, identifiable unit of data that forms the basic organizational component in a
database. Usually a combination of characters or bytes referring to one separate piece of
information. A data element may combine with one or more other data elements or digital
objects to form a digital record.

Data entropy Normal degradation in information content associated with data and metadata over
time.

44
Data Life Cycle Steps data undergo in their life cycle concerning data management. The Data Life
Cycle encompasses all facets of data generation to knowledge creation, including the
proposal, collection and quality assurance, metadata creation, data submission and
preservation, discovery, integration, analysis and data publication.

Deposit The act of submitting data, as to a repository (Strasser et al. 2012).

Derived data set New dataset created by using multiple existing datasets and their data elements as
sources (Strasser et al. 2012).

Digital curation: see curation

Digital data Refers to a structured sequence of bits/bytes that represents information content. In many
contexts digital data and data are used interchangeably implying both the bits and the content
(RDA Europe 2014b).

Digital object Consists of a structured sequence of bits/bytes. As an object it is named. This bit
sequence can be identified and accessed by a unique and persistent identifier or by use of
referencing attributes describing its properties.

DOI (Digital Object Identifier) A DOI is a string of letters and numbers that can be used to make
resources directly available to anyone over the internet. Used for data products so that they
can be uniquely identified and cited (Michener and Jones 2012).

Faceted search Enables users to discover specific data products by filtering a set of available
descriptors (such as author, repository where the data are stored, sensors used to collect the
data, geographic location; Michener and Jones 2012).

Header Meaningful name for referencing the content contained in a row or column, as in a
spreadsheet (Strasser et al. 2012).

Integrity Accuracy, validity and consistency of data. Integrity means that data has not been changed
or manipulated undocumented.

Meta-analysis An analysis that combines the results of multiple studies.

Metadata Provides information about and documentation of data content, context, quality and
structure. Data without metadata cannot be used.

Metadata editing tool Software tool to input, edit, and view metadata using a metadata standard.

Metadata standard Machine readable metadata using extensible markup language (XML) and
following a prescribed structure, depending on the chosen standard. Ensures consistency in
content and format of metadata.

Non-proprietary Refers to software not protected by patent, copyright, or trademark, often with open
source code.

Ontology A framework for interrelated concepts within a domain. An ontology would for example
link the terms “water vapour”, “relative humidity”, and “H2O vapour pressure”, so that a user
searching for one, would also see the other related terms and their relationships (Strasser et al.
2012).

Operations Data manipulation

45
Parameter A variable and measurable factor that determines or characterizes a system.

Persistent identifiers Persistent identifiers are labels for digital objects that remain the same
regardless of where the object is located. The use of persistent identifiers is not limited to web
material. It is equally essential for linking and citation of primary research with datasets
(DCC Glossary).

Provenance Data provenance refers to the ability to track data from creation through all
transformations, analyses and interpretations, enabling full understanding of the processes
used to create derived scientific products (Michener and Jones 2012).

Quality assurance/quality control (QA/QC) Refers to the mechanisms preventing errors from
entering a data set and to monitor and maintain data quality (Michener and Jones 2012).
Quality control is testing and other activities designed to identify problems and errors
(Strasser et al. 2012).

Quality level flag Indicator within the data file that identifies the level of quality of a particular data
point or data set. Flags can be defined within the metadata (Strasser et al. 2012).

Referable data Type of data that is persistently stored and has a persistent identifier (RDA Europe
2014b).

Relational database Collection of tables (often called relations) with defined relationships to one
another (Strasser et al. 2012).

Repository Facility specialized on storing data and make it available to re-users. Repositories provide
curation and stewardship services, access to data products and search interfaces.

Re-use Use of data by other researchers or for a purpose other than that for which they were collected.
Re-use of data can be limited by file formats or missing and incomplete metadata.

Scripted programme Programme (requiring a command line interface or similar) that performs an
action or task. Scripts can be saved, modified and re-used as necessary (Strasser et al. 2012).

Scientific workflow Description of the scientific procedure, for example what analyses have been
carried out on data. Can be informal (e.g. flowchart) or formal (e.g. workflow systems).

Structured data Tabular data stored in spreadsheets, csv or tsv. Structured data allow for filtering or
exporting selected parts.

Unstructured data Data with no tabular arrangement, like documents or models.

46
 8. References

AGU - American Geophysical Union. (2012). Data Position Statement. Retrieved from http://
sciencepolicy.agu.org/files/2013/07/AGU-Data-Position-Statement_March-2012.pdf

Corti, L., Van der Eynden, V., Bishop, L., & Morgan-Brett, B. (2011). Managing and sharing data:
training resources. UK Data Archive.

DataONE. (n.y.). Best Practices. Retrieved May 28, 2015, from https://www.dataone.org/best-
practices

DataONE. (2012a). DataONE Education Module: Data Management. Retrieved May 22, 2015,
from http://www.dataone.org/sites/all/documents/L01_DataManagement.pptx

DataONE. (2012b). DataONE Education Module: Analysis and Workflows. Retrieved May 29,
2015, from http://www.dataone.org/sites/all/documents/L10_Analysis Workflows.pptx

DataONE. (2012c). DataONE Education Module: Data Citation. Retrieved April 22, 2012, from
http://www.dataone.org/sites/all/documents/L09_DataCitation.pptx

DataONE. (2012d). DataONE Education Module: Data Entry and Manipulation. Retrieved April
29, 2015, from http://www.dataone.org/sites/all/documents/L04_DataEntryManipulation.pptx

DataONE Community. (2014). DataONE Education Module Lesson 1: Data Management Handout.
Retrieved April 7, 2015, from https://www.dataone.org/sites/all/documents/
L01_DataManagement_Handout_FINAL.pdf

DCC - Digital Curation Centre. (2008a). Digital Curation 101 Incentives for Digital Curation.
Retrieved from http://www.dcc.ac.uk/sites/default/files/documents/DC%20101%
20Incentives%20for%20Curation.pdf

DCC - Digital Curation Centre. (2008b). Digital Curation 101 Migrate. Retrieved from http://
www.dcc.ac.uk/sites/default/files/documents/DC%20101%20Migrate.pdf

DCC - Digital Curation Centre. (2008c). Digital Curation 101 Preservation Methods. Retrieved
from http://www.dcc.ac.uk/sites/default/files/documents/DC%20101%20Preservation%
20Methods.pdf

DCC - Digital Curation Centre. (2008d). Digital Curation 101 - Access, use and reuse. Retrieved
from http://www.dcc.ac.uk/sites/default/files/documents/DC%20101%20Access%20and%
20Use.pdf

DCC - Digital Curation Centre. (n.y.). Glossary. Retrieved July 2, 2015, from http://
www.dcc.ac.uk/digital-curation/glossary#B

DCC - Digital Curation Centre. (2004-2015). Using Metadata Standards. Retrieved June 2, 2015,
from http://www.dcc.ac.uk/resources/briefing-papers/standards-watch-papers/using-metadata-
standards

47
DCC - Digital Curation Centre. (2008-2015). DC 101 materials. Retrieved from http://
www.dcc.ac.uk/training/train-the-trainer/dc-101-training-materials

DCC - Digital Curation Centre. (2009). Preservation Action Checklist. Retrieved from http://
www.dcc.ac.uk/sites/default/files/documents/Preservation%20Action%20Checklist.pdf

DCC - Digital Curation Centre. (2010). How to Appraise and Select Research Data for Curation.
Retrieved May 29, 2015, from http://www.dcc.ac.uk/resources/how-guides/appraise-select-
data

DCC - Digital Curation Centre. (2014). Five steps to decide what data to keep. Retrieved May 29,
2015, from http://www.dcc.ac.uk/resources/how-guides/five-steps-decide-what-data-keep

DFG - Deutsche Forschungsgemeinschaft. (2014). Leitfaden für die Antragstellung: Projektanträge.

Retrieved from http://www.dfg.de/formulare/54_01/54_01_de.pdf

DFG - Deutsche Forschungsgemeinschaft. (2015). Guidelines on the Handling of Research Data in

Biodiversity Research. Working Group on Data Management of the DFG Senate Commission
on Biodiversity Research. Information for Researchers No. 36. Retrieved from http://
www.dfg.de/download/pdf/foerderung/antragstellung/forschungsdaten/
guidelines_biodiversity_research.pdf

Downs, R. R. (2013). Responsible Data Use: Data restrictions. Retrieved from

commons.esipfed.org/sites/default/files/RDUDataRestrictionsDowns_final.pptx

Duerr, R. (2011). Responsible data use. Presented at the AGU Data Management 101 for the Earth
Scientist. ESIP. Retrieved from http://wiki.esipfed.org/images/7/71/
ResponsibleDataUse_FINAL.ppt

ESRC - Economic and Social Research Council. (2012). Data Citation. What you need to know.
Retrieved from http://www.esrc.ac.uk/_images/Data_citation_booklet_tcm8-21453.pdf

Heidorn, B. (2008). Shedding Light on the Dark Data in the Long Tail of Science. Library Trends,
57(2). http://doi.org/http://hdl.handle.net/2142/9127

Houghton, J. (2011). Costs and Benefits of Data Provision. Centre for Strategic Economic Studies,
Victoria University. Retrieved from http://ands.org.au/resource/houghton-cost-benefit-
study.pdf

Kepler. (2008). Getting started with Kepler. Retrieved from https://kepler-project.org/users/

documentation

Ludwig, J., & Enke, H. (Eds.). (2013). Leitfaden zum Forschungsdaten-Management:

Handreichungen aus dem WissGrid-Projekt. Glückstadt: Hülsbusch. Retrieved from http://
www.univerlag.uni-goettingen.de/bitstream/handle/3/isbn-978-3-86488-032-2/
leitfaden_DGRID.pdf?sequence=1

Mantra, EDINA, & Data Library, University of Edinburgh. (2014). Research Data MANTRA
(online course). Retrieved April 20, 2015, from http://datalib.edina.ac.uk/
mantra#sthash.c9HnlX89.dpuf

48
Mayernik, M. (2013). Responsible Data Use: Citation and Credit. Presented at the Data
Management for Scientists Short Course; Federation of Earth Science Information
Partners: ESIP Commons. National Center for Atmospheric Research, DOI 10.7269/
P3QC01D5. Retrieved from http://live.commons.esipfed.bluedotapps.org/node/1428

Michener, W. K. (2006). Meta-information concepts for ecological data management. Ecological

Informatics, 1(1), 3–7. http://doi.org/doi:10.1016/j.ecoinf.2005.08.004

Michener, W. K., & Jones, M. B. (2012). Ecoinformatics: supporting ecology as a data-intensive

science. Trends in Ecology & Evolution, 27(2), 85–93. http://doi.org/10.1016/
j.tree.2011.11.016

NIST - National Institute of Standards and Technology. (2015). NIST Big Data Interoperability
Framework: Volume 1, Definitions. DRAFT. Big Data Public Working Group.

Open Data Bingo. (n.y.). Retrieved May 15, 2015, from http://data.dev8d.org/devbingo/

Piwowar, H. A., & Vision, T. J. (2013). Data reuse and the open data citation advantage. PeerJ,
1(e175). http://doi.org/https://dx.doi.org/10.7717/peerj.175

RDA Europe - Research Data Alliance. (2014a). Report on the RDA-MPG Science Workshop on
Data. Retrieved from https://europe.rd-alliance.org/sites/default/files/report/RDA-Europe-
Science-Workshop-Report_final_April2014.pdf

RDA Europe - Research Data Alliance. (2014b). DFT Model Overview & Term Definitions -
Prepared for Plenary 3. Dublin.

Rüegg, J., Gried, C., Bond-Lamberty, B., Bowen, G. J., Felzer, B. S., McIntyre, N. E., Soranno,
P.A., Vanderbilt, K.L., Weathers, K. C. (2014). Completing the data life cycle: using
information management in macrosystems ecology research. Frontiers in Ecology and the
Environment, 12(1), 24–30. http://doi.org/10.1890/120375

Strasser, C., Cook, R., Michener, W., & Budden, A. (2012). Primer on Data Management: What
you always wanted to know. (DataONE, Ed.). Retrieved from www.dataone.org

UK Data Archive. (2002-2015). How we curate data: the process. Retrieved June 1, 2015, from
http://www.data-archive.ac.uk/curate/process

UK Data Service. (2012-2015a). Prepare and manage data. Retrieved May 22, 2015, from http://
ukdataservice.ac.uk/manage-data.aspx

UK Data Service. (2012-2015b). Why share data? Retrieved May 22, 2015, from http://
ukdataservice.ac.uk/manage-data/plan/why-share.aspx

UK Data Service. (2012-2015c). Recommended formats. Retrieved June 1, 2015, from http://
ukdataservice.ac.uk/manage-data/format/recommended-formats.aspx

UK Data Service (Ed.). (2014a). Benefits of managing and sharing your data. University of Essex.
Retrieved from http://ukdataservice.ac.uk/media/440285/whysharedata.pdf

49
UK Data Service. (2014b). Formatting and organising research data. Retrieved June 2, 2015, from
http://ukdataservice.ac.uk/media/440281/formattingorganising.pdf

University of Western Australia. (2015). Research Data Management Toolkit. Retrieved May 20,
2015, from http://guides.is.uwa.edu.au/RDMtoolkit

50