Research integrity codes of conduct in Europe: Understanding the divergences

Abstract

In the past decade, policy‐makers in science have been concerned with harmonizing research integrity standards across Europe. These standards are encapsulated in the European Code of Conduct for Research Integrity. Yet, almost every European country today has its own national‐level code of conduct for research integrity. In this study we document in detail how national‐level codes diverge on almost all aspects concerning research integrity—except for what constitutes egregious misconduct. Besides allowing for potentially unfair responses to joint misconduct by international collaborations, we argue that the divergences raise questions about the envisaged self‐regulatory function of the codes of conduct.

Full text

1
Research Integrity Codes of Conduct in Europe: Understanding the Divergences
Hugh Desmond
Kris Dierickx
Published in Bioethics.
Abstract
In the past decade, policy-makers in science have been concerned with
harmonizing research integrity standards across Europe. These standards are
encapsulated in the European Code of Conduct for Research Integrity. Yet,
almost every European country today has its own national-level code of
conduct for research integrity. In this study we document in detail how
national-level codes diverge on almost all aspects concerning research
integrity – except for what constitutes egregious misconduct. Besides
allowing for potentially unfair responses to joint misconduct by international
collaborations, we argue that the divergences raise questions about the
envisaged self-regulatory function of the codes of conduct.
KEYWORDS: Research integrity – Science policy – Self-regulation –
Professionalism – Research Methodology – European Code of Conduct
1. Background
In one of the first surveys of research integrity (RI) standards, the European Science
Foundation concluded that there was a “wide range of approaches” across European countries
and that there was a need for “harmonized standards across Europe”.
1
This was the original
rationale for the European Code of Conduct for Research Integrity (ECoC), which was first
published in 2011 and updated in 2017.
1
ESF, (European Science Foundation). (2008). Stewards of Integrity: Institutional Approaches to Promote and
Safeguard Good Research Practice in Europe. European Science Foundation. Retrieved from
http://www.enrio.eu/wp-content/uploads/2017/03/StewardOfIntegrity.pdf, p. 50.

2
Harmonization does not mean uniformity. The harmonization the ECoC aimed to
provide was that of a reference point for a “common understanding of the demands of research
integrity”.
2
Such a common understanding was intended to be compatible with national-level
differences:
However, unlike the fundamental values of scientific integrity and the violation
thereof, which have a universal character, [poor and inappropriate] practices
3
may be
subject to different national traditions, legislative regulations or institutional
provisions.
4
Similarly, the 2017 ECoC explicitly aims to be a reference point that “allows for local or
national differences in its implementation”.
5
We call this the (European) core versus national periphery model of harmonization:
the ‘core’ aspects of RI – the principles, the good practices, and the definition of misconduct
– must be specified by Europe-wide standards set by the ECoC, but the ‘peripheral’ aspects –
i.e., what counts as merely a questionable research practice may vary from country to country.
In this article we will seek to map patterns of divergence and convergence across
national-level codes and guidelines in Europe, and seek to inquire whether these patterns are
compatible with the core-periphery model of harmonization. Since the publication of the
background document to the first ECoC, over a decade has passed, and as we will document
in this paper, there has been a proliferation of national-level research integrity codes of
conduct since then. It is timely to revisit the original intention of harmonization (as formulated
by the core-periphery model), and enquire to what extent this is currently realized.
Note that the general impact of the ECoC is not in question. It is the standard for all
projects funded by the European Commission, including those by the European Research
Council and those within the Horizon 2020 and successor Framework Programmes. While
more difficult to quantify, it has also undoubtedly stimulated national policy-making efforts.
Nonetheless, it is one thing to state that the values of RI should be held in common, and
another for there to actually be widespread agreement, let alone unanimity, on the core aspects
of RI.
2
ESF-ALLEA, European Science Foundation and All European Academies (2010). A European Code of
Conduct for Research Integrity: Background Paper. Retrieved from https://allea.org/wp-
content/uploads/2015/09/A-European-Code-of-Conduct-for-Research-Integrity_final.10.10.pdf
3
For instance: “questionable procedures for obtaining informed consent, insufficient respect and care for
participants in the research, improper research design and carelessness in observation and analysis, unsuitable
authorship or publishing practices, and reviewing and editorial derelictions” Ibid: 14.
4
Ibid: 14
5
ALLEA, A.E.A. (2017). The European Code of Conduct for Research Integrity. Retrieved from
http://www.allea.org/wp-content/uploads/2017/04/ALLEA-European-Code-of-Conduct-for-Research-Integrity-
2017.pdf

3
Why does harmonization matter? We will normatively evaluate observed divergences
in light of the following two questions:
(1) Do the divergences allow for unfair responses to joint misconduct by
international collaborations?
(2) Do the divergences undermine the credibility of RI codes of conduct as self-
regulatory tools?
The first question regarding fairness is the most straightforward one: if different national-level
codes of conduct define misconduct differently, then – in principle – not all researchers in an
international collaboration may be held accountable (in the same way, or at all) for joint
misconduct. Even slight divergences in the formulation of what constitutes research
misconduct may have outsized impact: when allegations of misconduct are tried in courts of
law, judges may use codes of conduct as ‘soft’ legal documents to guide their judgment.
6
Even
when university bodies investigate allegations of misconduct, accused researchers may hire
the services of a lawyer or team of lawyers, and then the precise wording of codes of conduct
become crucial (personal experience of author 2 [KD] as member of research integrity
commissions). This strong rationale to align definitions of misconduct has been noted by
others,
7
and one of the research aims of this paper is to help map divergences which may be
problematic for reasons of fairness.
The second question is more subtle and concerns what we call the credibility of a
guideline or code of conduct. We flesh this out in more detail in the discussion section, but it
basically refers to how some voices are skeptical of the efficacy of self-regulation, and believe
codes of conduct to primarily be window-dressing, presenting a morally agreeable façade to
the rest of society while hiding a harsher reality of competition and prestige-maximization.
The belief is that the primary purpose is to create a perception that integrous behavior is
valued, and not to promote actual integrous behavior. Thus, skeptics would not be surprised
by, for instance, a large divergence in the fundamental values listed in RI codes: after all, for
window-dressing it is less important which values are listed, and more important the fact that
some values are present. In this way we will evaluate existing divergences according to
whether they are consistent with the skeptical view that codes of conduct are not credible as
effective self-regulatory tools.
6
Kovács, A., Tóth, T., & Forgács, A. (2016). The Legal Effects of European Soft Law and Their Recognition at
National Administrative Courts. ELTE Law Journal. 2, 53–70.
7
E.g. Fanelli, D. (2011). The Black, The White, and the Grey Areas: Towards an International and
Interdisciplinary Definition of Scientific Misconduct. In T. Mayer & N.H. Steneck (Eds.), Promoting research
integrity in a global environment (pp. 79–90). Singapore: World Scientific Publishing.

4
In sum, in this paper we investigate the state of the harmonization of national-level
codes and guidelines, where ‘harmonization’ is understood in terms of the core-periphery
model. While harmonization does not mean complete uniformity across national-level
regulatory documents, for reasons of methodological soundness, we will nonetheless
investigate patterns of replication of the ECoC across national-level documents. The detailed
patterns of divergence and convergence will allow for conclusions to be drawn about
harmonization in the European context, as well as about potential problems arising from the
lack of harmonization.
2. Methods
2.1 Methodological Desiderata
This is not the first study aimed at detailing differences in national-level codes and guidelines
on RI.
8
There is an obvious, content-related reason for undertaking a new study, and that is
that the regulatory situation for RI in Europe has continued to change at a fast pace. A second
European code of conduct was published in 2017, while countries such as France, Estonia,
The Netherlands, Italy, and the UK have published new codes or guidelines in the past few
years (see supplementary materials).
However, there are also methodological reasons for undertaking a new study, and an
examination of previous studies helps identify two methodological desiderata. The first is that
previous studies depended on interpretative decisions about the meanings of values and
virtues. For instance, Godecharle et al. 2013 distinguish between “honesty” and “openness or
open communication”, even though – at least, by dictionary definitions of these words – they
are near-synonyms. An example in Aubert Bonn et al. 2017 is the way in which similar virtues
are lumped together: thus “Openness; Verifiability” form one category, and “Objectivity;
Scrupulousness; Transparency” another. However, with equal justification, ‘transparency’
could have been categorized together with ‘openness’ and ‘verifiability’.
Such interpretations are not necessarily problematic if reasons are given, but this was
not the case. Without such explication, it raises the question if another researcher, looking at
the same basic data, might come to different conclusions about the patterns of divergence. For
8
Aubert Bonn, N., Godecharle, S., & Dierickx, K. (2017). European Universities’ Guidance on Research
Integrity and Misconduct: Accessibility, Approaches, and Content. Journal of Empirical Research on Human
Research Ethics. 12(1), 33–44. https://doi.org/10.1177/1556264616688980; Godecharle, S., Nemery, B., &
Dierickx, K. (2013). Guidance on research integrity: no union in Europe. The Lancet. 381(9872), 1097–1098.
https://doi.org/10.1016/S0140-6736(13)60759-X

5
instance, if a large divergence between national-level codes is noted, to what extent does this
reflect a genuine lack of harmonization, or simply a divergence resulting from interpreting
“honesty” and “openness” as different values? Given this threat to reproducibility, we can
stipulate as a methodological desideratum that, when aiming to understand the differences
between different national-level approaches, one should avoid, as much as possible,
controversial or unsupported interpretative decisions on the basic data – i.e., the words used
by guidelines and codes of conduct.
A second, related methodological desideratum is that previous studies either did not,
or only insufficiently, discuss the importance of the observed differences. For instance, if code
A lists ‘transparency’ as a virtue, and code B lists ‘openness’, then most would find this to be
harmless variation on basically the same virtue. Why should differences actually matter? In
this study we anticipate this issue by focusing the discussion on the core-periphery model of
harmonization.
2.2 Search Methodology
The purpose of the search methodology was to represent national-level approaches to research
integrity (RI) by a single document that can be considered as the leading document in the
national context. In this way the comparative analysis of the ECoC and the national-level
documents was considerably simplified; we will argue later that this simplification does not
call the conclusions into question.
Initial Collection. For the initial collection we cast the net widely and included all
national-level ‘regulatory documents’ directly pertaining to RI. A ‘regulatory document’, as
we understand it, can in principle refer to a document regulating both the actions of individuals
(e.g. an ethics code) or the actions of institutions. Thus, a regulatory document can refer to
any of the following: codes of conduct, guidelines, policy documents, laws (statutes, charters),
and even more descriptive documents such as survey reports, meeting reports, and position
papers. We included a document if it contained substantial normative position statements on
any one of the following: (1) the principles underlying research integrity (e.g., honesty), (2)
behaviors constituting good research practice, (3) behaviors constituting research misconduct,
(4) a plea for the importance of research integrity for science and society.
For the initial collection of regulatory documents, we used the following six
independent search methods for each member of the 32 EFTA countries (EU28 + Norway,
Switzerland, Liechtenstein, Iceland). This means that if, for instance, five documents were
found with method 1, these were double checked with methods 2 through 6. We used these

6
six independent methods to minimize the probability of missing an important regulatory
document.
1. Exhaustive search of following websites, if available:
a. national research council
b. national agency on research integrity
c. national scientific fund
d. national academy of science
2. Search of the websites of academy members of ALLEA
3. Search for “integrity” and “<integrity translated into local language>” of
websites of prominent universities in that country -whenever possible,
websites in original language (through Google translate).
4. A search by means of an internet search engine (Google) with search
terms ((“research integrity” OR “scientific integrity” OR “science
integrity”) AND <name of country>).
5. Search of the resources for that country listed on the website of the
European Network of Research Integrity Officers (enrio.eu).
6. Resources listed on European Science Foundation “Stewards of Integrity”
document (URL:
http://digital.csic.es/bitstream/10261/8663/1/StewardsOfIntegrity.pdf)
This initial collection was completed in March 2019 and updated through August 2020.
Selection. The selection consisted of three steps. In the first step, we selected
candidates for the national ‘leading’ regulatory documents for inclusion in the comparative
study. We considered a document to be a ‘candidate leading document’ if it contained (1)
authoritative formulations of principles of RI, AND (2) definitions of good practices, AND
(3) definitions of misconduct. In many national-level contexts, this procedure was sufficient
to identify the national leading document. However, some countries proved to have multiple
potentially leading documents. For instance, in Denmark or Norway, both a law and a national
code of conduct can be considered leading. In the UK or France, there are multiple national
codes of conduct (authored by Research Councils UK or UK Research Integrity Office; Centre
Nationale de la Recherche Scientifique or Agence Nationale de la Recherche), as well as
national documents aimed primarily at institutions (Concordat in the UK, Charter in France:
see supplementary materials).
When this was the case, in the second step we chose a single candidate leading
document based on following additional criteria:
• We prioritized documents aimed at guiding individual researchers over those
designed for guiding institutions (when the latter seek to author an institution-
specific code or guideline). The reason for this was to facilitate a comparison

7
between each country’s leading document and the ALLEA code, which is
primarily focused on individual researchers.
• If the previous criterion yielded a tie, we prioritized documents with detailed
statements on the constitutive elements of RI over those with few detailed
statements on the elements of RI.
• If the tie was still not broken, we then prioritized the document with the most
institutional signatories.
This decision-making process allowed us to select a single candidate leading document for
each country.
In a final step, we presented this choice, together with our justification, to local RI
experts so they could verify our choice (or offer corrections if necessary). A person was
deemed a local expert if he or she (1) had been appointed as a contact person by a national
research council, a national agency for research integrity, a national scientific fund, or a
national academy, OR (2) was a member of ENRIO (European Network of Research Integrity
Officers), OR (3) was part of an ERC or Horizon2020-funded project on research integrity,
OR (4) a member of EARMA (European Association of Research Managers and
Administrators). The initial search and selection of documents are summarized in Figure 1.
Figure 1: Flowchart for the initial search and selection of national regulatory documents for
inclusion in the comparative analysis
2.3 Methodology of Content Analysis
Initial search (N = 153)
Non-leading regulatory documents (N=127)
Candidate leading
documents (N = 32)
One leading document selected from
multiple candidates (3 in FR, 3 in UK, 2 in DK,
2 in NO, 2 in ES, 2 in PL)
Leading Documents
(N = 24)
Documents submitted
to comparative analysis
(N = 20)
Four leading documents predate ALLEA 2011
(BE, CH, RO, HU)

8
In light of the methodological problems faced by previous studies (see section 2.1), in this
study we opted for a very minimalist method of content analysis.
9
This consisted of simply
verifying whether national-level documents replicated the ECoC in the following three areas:
(1) principles of RI, (2) definition of good research practices, and (3) definition of research
misconduct. ‘Replication’ was defined as the literal copying of the principles listed by the
ECoC, the categories of good research practices, and the categories of misconduct. Thus, for
instance, the 2017 version of the ECoC lists four values of RI: reliability, honesty, respect,
accountability. By contrast, the leading regulatory document in Estonia lists six principles (or
categories of principles): freedom, responsibility, honesty and objectivity, respect and caring,
justice, openness and cooperation. The leading regulatory document in The Netherlands
contains five values: honesty, scrupulousness, transparency, independence, responsibility. We
categorized both Dutch and Estonian documents as ‘non-replications’, thus abstracting away
from the obvious overlap between all three lists.
In focusing on the actual words used rather than their meaning we consciously avoided
attempting to quantify any conceptual overlap between documents, since this would have led
to verbal disputes of dubious importance (see section 2.1). For instance, we were able to avoid
questions such as: Is ‘scrupulousness’ the same as ‘reliability’? Is ‘responsibility’ the same as
‘accountability’? ‘Transparency’ and honesty’?
The strength of this minimalist methodology is that the obtained results are very
robust; the weakness is that it places a greater onus on normatively evaluating the results. (By
contrast, when normative presuppositions are built into describing codes of conduct, then
interpreting the resulting differences is more straightforward.) The measure of literal
replication will categorize both semantically similar terms and dissimilar terms as ‘non-
replications’, and this means that, especially with regard to RI principles, widespread non-
replication could be compatible with widespread semantic agreement.
10
Nonetheless, as
discussed in section 4.2, non-replications raise important questions about the purposes of RI
codes. Hence the importance of the discussion section, where we will argue that the observed
divergences are important, and potentially threaten the core-periphery model. Nonetheless,
the results are such that future research may in principle, using a different normative
framework, evaluate the results differently.
9
Rössler, P. (2013). Comparative Content Analysis. In F. Esser & T. Hanitzsch (Eds.), The Handbook of
Comparative Communication Research (pp. 459–68). New York: Routledge.
https://doi.org/10.4324/9780203149102-39
10
Issues of translation can also lead to non-replication despite relative semantic similarity.

9
3. Results
3.1 Evolution of national regulatory documents
Twenty-four countries (out of a total of 32) were found to have a leading regulatory document
on RI. Figure 2 shows how many countries that lacked a leading national-level regulatory
document in 2012, possess one in 2020. Moreover, among those without such a document
today, two (Bulgaria, Luxemburg) explicitly adopt the European Code of Conduct. A further
two (Greece and Slovenia) have stated the intention to develop a national-level framework.
For four countries – Malta, Liechtenstein, Cyprus, and Iceland – no statement concerning
national-level framework could be found. However, institutional-level RI-regulatory
documents exist in Malta, Iceland, and Cyprus. No institutional code was found for the three
Liechtenstein higher education institutions,
11
but given the fact that the University of
Liechtenstein, the largest institution in Liechtenstein, is small by international standards (1200
students), it is perhaps not surprising that an explicit code has not (yet) been deemed
necessary.
In any case, we concluded from the search that, with the exception of Liechtenstein,
all countries in Europe have a de facto leading regulatory document, whether that is the ECoC
itself, a specific national code or guideline, or, for small countries, a code or guideline in a
large university.
2020
11
See https://www.liechtenstein.li/en/education/higher-education/.

10
2012
Figure 2: Countries with leading regulatory documents concerning RI in 2012 (left) and
2020 (right). The left map is based on the data in
12
.
=
Leading national-level code
not found
=
Leading national-level
code present
3.2 Differentiation of Functions
Our investigation of the regulatory documents corroborates ESF’s prior finding that there is a
broad variance in types of regulatory document.
13
In other words, there is a wide variety
between different groups of envisaged users (all researchers, PhD students, supervisors, policy
makers) as well as different envisaged uses (pragmatic rules for quick consultation vs. in-
depth discussion of scientific methodology and values in science). Since this will be one of
the obvious explanations to be discussed (in section 4), we also aimed to map some of this
variance.
A difficulty here is that the codes of conduct themselves do not always explicitly state
what their envisaged uses are, nor what their target users are. Yet it is clear that the envisaged
uses are different. The longest document (that of Sweden, ca. 44.000 words) is fifty times
longer than the shortest (that of Norway, less than 900 words): whereas the latter contains
only concise, actionable rules and advice, the former contains in-depth discussions of, for
instance, various possible approaches to defining research misconduct. Contrasting with both
of these is the Danish Code of Conduct for Research Integrity (approximately 5.000 words),
which is structurally very similar to the new Danish Research Misconduct etc. Act (of 26 April
2017).
With this example in mind, one can look at ‘orders of magnitude’ differences in word
count as a rudimentary indication that different envisaged uses are at work (Figure 3). The
preceding discussion suggests a rough tri-partite categorization of envisaged uses.
• Portable vademecum, containing actionable advice. Paradigmatic example:
Norwegian leading document.
• A quasi-legal document, containing detailed descriptions of good practices and
categories of misconduct. Paradigmatic example: Danish leading document.
12
Godecharle, Nemery, Dierickx (op. cit. n. 8) : 1097–1098
13
ESF (op. cit. n. 1)

11
• An academic exploration, containing substantial philosophical, sociological, and
legal background to research integrity. Paradigmatic example: Swedish leading
document.
Note that the boundaries between these categories are vague, and that a more detailed
categorization of envisaged uses may be possible. However, even this rudimentary
categorization will help structure the normative evaluation of the observed differences
between leading regulatory documents.
Figure 3: The longest regulatory document analyzed in this paper was over 50 times longer
than the shortest. Even if one cuts out the two largest and two smallest outliers, the variation
in length is still one order of magnitude. This in itself is a strong indication that different
documents are intended for different uses. (Preambles but not tables of contents, annexes or
appendices, or references are included in the word count.)
3.3 Patterns of Replication
Four of the twenty-four leading national-level regulatory documents predate the first ECoC
(2011), and thus could not reflect the impact the core-periphery model has had on national-
level documents. (This is especially relevant with regards to the problem of credibility, in
section 4.2.) The comparative analysis focused on the twenty remaining documents, of which
ten were published in the period 2011-2016, and ten in the period after 2017. We found that
the formulations of the ECoC – whether ECoC 2011 or ECoC 2017, and whether concerning
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
NO
LT
ECoC
AT
SK
LV
CZ
HR
BE
RO
ES
DK
PT
FI
IE
CH
UK
DE
EE
PL
NL
IT
FR
HU
SE
Word Cou nt

12
the values of RI, definitions of misconduct, or definitions of good practices – are almost never
replicated by national-level documents. Of the 60 points of comparison with the ECoC (3 per
country), the formulations of an ECoC were replicated only twice: the Irish code replicates
the values listed in the 2011 ECoC, and the Portuguese code replicates the values listed in the
2017 ECoC. This bird’s-eye view is summarized in Figure 4, which reveals only the
Portuguese and Irish leading documents to replicate core aspects of RI in the relevant ECoC.
Principles
Good practices
Definitions of misconduct
Figure 4: National-level regulatory documents almost never replicate the ECoC with regards
to either principles of RI (left), definitions of good practices (center), or definitions of
misconduct (right). For details, see supplementary materials.
=
Relevant sections from
ALLEA 2011 or ALLEA
2017 replicated
=
Relevant sections from
ALLEA 2011 or ALLEA
2017 not replicated
=
National-level code absent or predates
ALLEA 2011
However, partial overlaps also matter. Note that ‘non-replication’ is a broad category
that contains both partial overlap and a complete lack of overlap. Tables 2 and 3 give, as much
as possible, the literal wording in the leading regulatory documents. For instance, 3 of the 6
principles listed by the Austrian leading regulatory document are adopted from the ECoC
principles (reliability, impartiality, fairness). By contrast, none of the principles listed by the
Estonian leading document overlap with the ECoC principles (exercising methodological
doubt; designing good experiments; managing data; proper use of funding)
14
.
14
Note that “designing good experiments” could be taken as a principle that is very close to reliability. However,
it is not the same, because reliability might be applicable to research in theoretical sciences or in humanities,
where no experiments are involved. Moreover, the “goodness” of an experiment may cover many dimensions, of
which reliability is only one. As a reminder, for such reasons we have consciously avoided making judgments of
this sort (i.e., whether ‘reliability’ and ‘designing good experiments’ refer to the same principle). See section 2.1.

13
We separately mapped the partial overlaps, and how these have changed over time.
Figures 5 and 6 show how probable a principle or category of misconduct listed in the ECoC
would be duplicated by a national-level document. Figure 5 shows that ‘honesty’ is the only
principle to be frequently listed by national-level documents, especially after 2017. Figure 6
shows that FFP is the only category of misconduct that is universally duplicated across
national-level documents (and indeed the consensus on FFP as a form of misconduct is truly
global, with the Federal Policy on Research Misconduct and the Singapore Statement also
referring to FFP). The first version of the ECoC was vague about other forms of misconduct,
and its two other categories of ‘minor misdemeanours’ and ‘failure to meet ethical and legal
requirements’ are never adopted by national-level documents. The second version of the
ECoC was much more detailed in this regard, but the categories it introduced were not nearly
adopted with the same frequency as FFP. Of the thirteen other categories of misconduct, only
‘conflict of interest’, ‘manipulating authorship’ and ‘misrepresentation’ were cited by more
than one national-level document.
We also mapped how the ‘grain’ of the taxonomy of core RI elements varied across
national-level documents. By ‘grain’ we mean the extent to which detailed distinctions were
made detailed. Thus the 2011 ECoC is relatively ‘fine-grained’ with regards to RI principles
(8 principles listed), but ‘coarse-grained’ with regards to categories of misconduct (3
categories). By contrast, the 2017 ECoC is coarse-grained with regards to RI principles (4
principles listed), but fine-grained with regards to categories of misconduct (14 categories).
Figure 7 gives an overview of the results of this exercise, and Figure 8 shows the same
results as a more detailed distribution of the number of listed core RI elements. The first
takeaway is that the number of elements listed per dimension of RI (principles/good
practices/misconduct) vary greatly. For instance, the fine-grained approach to RI principles
adopted by the 2011 ECoC is not followed by a majority of national-level documents: most
list 3 or 4 principles versus the ECoC’s 8 principles.
It is interesting to see the trends in these data, and global changes between the two
periods of 2011-2016 and 2017-present. We caution against reading too much into these data:
the averages and standard deviations are not statistically significant given the low sample sizes
(N=11 including the ECoC). In other words, if one or two national-level documents would
have listed very different numbers of RI principles or types of misconduct, then a very
different picture would emerge.
Nonetheless, with this caveat in mind, in Table 1 is how the granularity of the all
leading documents (plus ECoC) actually evolved over time:

14
2011-2016
2017-present
Number of RI principles
Average: 4.9
Standard deviation: 1.9
Average 4.9
Standard deviation: 1.4
Number of good practises
Average 5.2
Standard deviation 2.4
Avarege: 5.7
Standard deviation 1.3
Number of categories of
misconduct
Average 6.8
Standard deviation: 3.9
Average: 8
Standard deviation: 5.5
Table 1: Time evolution of the granularity of ECoC and leading national-level regulatory
documents.
With respect to RI principles, there has been a weak convergence on listing 4-5 principles.
With respect to good practises, there has been a stronger convergence on listing 5-6 categories
of good practise. Finally, and most interestingly, on average leading regulatory documents have
been more detailed as to the types of behavior are in violation of the code (under 4 to over 5
types), and there is an increased disparity in the approaches of different documents. Some have
opted, with the 2017 ECoC, to list a large number (10+) of categories of misconduct, while
others have opted for a less detailed approach, and almost half (4 out of 10) of the leading
documents for this period only list 3 types of misconduct. The result is a flattened distribution
with a large right skew (bottom graph in Figure 8).
To sum up the results: not a single national-level leading document entirely adopts the
ECoC’s formulation of the core elements of RI. There is a large divergence in the principles
and categories of misconduct listed: the only unambiguous consensus is that fabrication,
falsification, and plagiarism (FFP) count as misconduct. However, this consensus is global
15
,
and is not a particularly surprising finding. With regards to principles of RI, the only candidate
for a quasi-consensus would be honesty. However, also here the consensus seems to be broader
than the European context alone, given how ‘honesty’ is the very first principle listed by the
Singapore Statement on Research Integrity
16
. We will now discuss consequences of these
findings for the envisaged European model of harmonization.
15
WCRI, (World Conference on Research Integrity). (2010). Singapore Statement on Research Integrity.
Retrieved from https://wcrif.org/documents/327-singapore-statement-a4size/file; OSTP, (Office of Science and
Technology Policy, Executive Office of the President). (2000). Federal policy on research misconduct. Federal
Register. Retrieved from https://www.govinfo.gov/content/pkg/FR-2000-12-06/pdf/00-30852.pdf
16
WCRI (op. cit. n. 15)

15
1
2
Figure 5: Number of times a principle listed in the ECoC is also listed in a leading national regulatory document
3
4
0
1
2
3
4
5
Honesty
Reliability
Objectivity
Impartiality
Open communication
Duty of care
Fairness
Responsibility
Number of times listed among the
leading documents (N=10)
Principles listed in ECoC 2011
0
1
2
3
4
5
6
7
8
9
Reliability Honesty Respect Accountability
Number of times listed among the
leading documents (N=10)
Principles listed in ECoC 2017

16
1
Figure 6: Number of times a category of misconduct listed in the ECoC is also listed in a leading national regulatory document
2
0
10
FFP
Minor misdemeanour
Failure to meet clear ethical and legal requirements
Number of times listed among the leading
documents (N=10)
Categories of misconduct listed in ECoC
2011
0
2
4
6
8
10
FFP
Manipulating authorship
self-plagiarism
citing selectively
Withholding results
Conflict of interest
Expanding the bibliography unnecessarily
Malicious accusation of misconduct
Misrepresentation
Exaggeration of research
Delaying research of others
Misusing seniority
Ignoring potential violations of RI
Supporting predatory journals
Number of times listed among the leading
documents (N=10)
Categories of misconduct listed in ECoC 2017

17
Figure 7: The number of listed RI elements fluctuates across leading national-level regulatory documents.
0
2
4
6
8
10
12
14
16
18
ECoC
2011
AT CZ DK ES FI HR IE LT NO PL
Number of listed RI elements
Leading documents
published 2011-2016
0
2
4
6
8
10
12
14
16
18
20
ECoC
2017
DE EE FR IT LV NL PT SE SK UK
Number of listed RI elements
Leading documents published 2017-

18
Figure 8: The distributions of the number of core RI elements across leading national regulatory documents
0
1
2
3
4
5
345678
Number of leading
documents
Principles
2011-2016 2017-
0
1
2
3
4
5
234567810
Number of leading
documents
Categories of Good Practises
2011-2016 2017-
0
1
2
3
4
5
34578910 13 16 18
Number of leading
documents
Categories of Misconduct
2011-2016 2017-

19
4. Discussion
1
2
While the large divergences documented in the previous section results do not necessarily
3
falsify the core-periphery model of harmonization, they do call it into question. If the
4
divergences are compatible with a “common understanding” of the core aspects of research
5
integrity,
17
then it is not obvious at all what this common understanding amounts to precisely
6
– beyond categorizing FFP as a violation.
7
Could one not sidestep this question about “common understanding” by construing it
8
simply as the recognition that research integrity matters rather than actually agreeing about
9
what research integrity entails? Thus, the documented variances could be cast as harmless
10
differences of opinion: what matters is that the importance of integrity is recognized, and all
11
leading national-level documents do this. However, if only the recognition matters, why then
12
the construction of a relatively detailed code of conduct would be unnecessary. Instead,
13
research bodies could put out a joint statement affirming the importance of RI. Since the
14
trouble has been taken to construct substantive codes of conduct, we cannot simply construe
15
the “common understanding” of RI as a recognition of the importance of RI. There must be a
16
shared understanding about substantive content as well.
17
How can we evaluate the divergences mapped in the previous section, and how can
18
we distinguish between divergences that are harmless, and those that are genuinely
19
problematic and put pressure on the core-periphery model? Here we will evaluate the
20
divergences according to two dimensions: threats to fairness, and threats to credibility. If the
21
harmonization between national-level leading regulatory documents is to be genuine, then
22
such documents should give no grounds for unfair treatment of joint misconduct by
23
international collaborations, and they should make clear that they are truly aimed at guiding
24
researcher behavior (as opposed to being merely window-dressing). We will argue that, with
25
regards to fairness, harmonization is clearly lacking, and that with regards to credibility,
26
harmonization is genuinely ambiguous.
27
28
4.1 Fairness
29
It would be unfair for a researcher to be judged more harshly than his or her collaborator for
30
joint misconduct, simply because they are affiliated with an institution in country X rather
31
than country Y. This fairness issue was anticipated by the ECoC 2011, which states that
32
17
ESF-ALLEA (op. cit. n. 2) : 4

20
potential misconduct should be investigated in the country/institute of the project leader.
18
1
However, this passage has been deleted in the 2017 version of ECoC, most likely because of
2
the serious legal obstacles faced by an institution or committee in country X investigating a
3
researcher at a different institution in country Y.
4
Do the current divergences between national-level documents avoid giving grounds
5
for this type of unfairness?
19
While all examined documents categorize FFP as violations,
6
there is a large divergence in what additional categories of behavior constitute misconduct or
7
‘questionable practice’. The 2017 ECoC lists a total 15 categories of misconduct (including
8
FFP), whereas, for instance, the German leading regulatory document only explicitly lists FFP
9
as an example of research misconduct. Even among the documents adopting a coarse-grained
10
approach to defining misconduct, large differences can be seen, even in the strategy of
11
definition. For instance, the Dutch leading regulatory document also only explicitly prohibits
12
FFP, but instead of listing types of prohibited behavior, it seeks to define expected standards
13
of behaviors which can be used to judge if a particular behavior is problematic or not. If one
14
would include failures of cited standards as different types of conduct, the Dutch leading
15
regulatory document in fact takes the most fine-grained approach to defining misconduct, with
16
22 categories in total.
17
Thus, in situations where the precise wording of the definition of misconduct would
18
matter – i.e., when allegations of misconduct end up in courts of law, or when accused
19
researchers hire lawyers for representation in internal disciplinary investigations – it seems
20
that different codes of conduct could lead to different judgments. The harmonization of
21
national-level leading regulatory documents is not such that no grounds are given for this type
22
of potential unfairness.
23
We surmise that the trend towards more detailed definitions of categories of
24
misconduct (see Table 1 and Figures 6-8) is an implicit recognition of the importance of
25
anticipating legal challenges to internal disciplinary investigations. Nonetheless, more work
26
remains to be done not just to add increasing detail in the definition of misconduct, but also
27
to reflect on what is the best way of defining misconduct, and what jurisprudential approach
28
is most appropriate for the goals of RI (see section 4.3).
29
18
ESF-ALLEA, E.S.F. and A.E.A. (2011). The European code of conduct for research integrity. Retrieved from
http://www.allea.org/wp-content/uploads/2015/07/Code_Conduct_ResearchIntegrity.pdf
19
Note that such unfairness can always arise due to different legal or institutional contexts; the question at stake
here is whether the divergent formulations in the codes and guidelines give any grounds to unfair judgments of
joint international misconduct.

21
Besides the definition of the categories of misconduct, a second, and perhaps more
1
fundamental source of potential unfairness lies in divergences between how FFP-behavior is
2
evaluated. While we did not systematically map this for the current study, reading and
3
analyzing the documents was sufficient for finding examples that different codes do not give
4
the same grounds for findings of FFP. We identified two issues: responsibility (whether a
5
researcher is blameworthy is for misconduct) and culpability (the extent to which a researcher
6
is blameworthy).
7
With regards to responsibility: some codes explicitly hold all co-authors of a
8
fraudulent research paper responsible, while many codes are silent on this issue. For instance,
9
the Austrian OeAWI Guidelines for Good Scientific Practice state that all co-authors of a
10
publication are jointly responsible for it.
20
By contrast, the main regulatory documents in
11
France
21
do not contain a provision similar to the OeAWI document.
22
Divergences in such
12
provisions would have concrete consequences for cases of FFP in co-authored papers.
13
With regards to culpability: in various areas of the law, it is common to distinguish
14
between levels of intent in order to determine the proportionate sanction
23
. Codes of conduct
15
have often adopted this legal framework, most explicitly in the Federal Policy on Research
16
Misconduct (USA), which states that a finding of misconduct (defined as FFP) must be shown
17
to have been “committed intentionally, or knowingly, or in reckless disregard of accepted
18
practices”
24
. This explicitly excludes negligent actions as potential violations of research
19
integrity standards, i.e., the individual who violated RI standards did not know but should
20
have known those standards (see Desmond 2019 for a fuller overview of this issue).
21
In European context it has not been possible to take a similarly uniform line with
22
regards to culpability, due to divergent legal contexts between European countries, and this is
23
one reason why the 2017 ECoC is silent on levels of intent and culpability (Maura Hiney, co-
24
author of the 2017 ECoC, personal communication). Yet this means that some codes of
25
20
OeAWI, A.A. for R.I.G. for. (2015). OeAWI Guidelines for Good Scientific Practice. Retrieved from
https://www.cdg.ac.at/fileadmin/main/documents/Sonstige_Dokumente/160418_OeAWI_Richtlinien_Broschue
re_DE_EN.pdf
21
COMETS, C.E.C. (2017). Integrity and responsibility in research practices. Retrieved from
http://www.cnrs.fr/comets/IMG/pdf/guide_2017-en.pdfCNRS & INRA. (2015). French National Charter for
Research Integrity. Retrieved from
http://www.cnrs.fr/comets/IMG/pdf/french_national_charter__research_integrity_29jan2015_-2.pdf
22
Compare these to the latest version of the ICMJE recommendations, where it is only specified that “an author
should be able to identify which co-authors are responsible for specific other parts of the work” and that “authors
should have confidence in the integrity of the contributions of their co-authors.” Hence it stops short of stipulating
that authors should take responsibility for the integrity of the contributions of co-authors.
23
Yaffe, G. (2010). “Intention in Law.” In A Companion to the Philosophy of Action, edited by Timothy
O’Connor and Constantine Sandis. Oxford: Wiley-Blackwell.
24
OSTP (op. cit. n. 15) : 76262

22
conduct include negligent violations of the code as potential misconduct (e.g. DE, AT, FI,
1
NO), while others purposely restrict research misconduct to behaviors with a conscious
2
intention to deceive (e.g., IE, UK).
3
In sum, there is potential for unfair treatment of alleged misconduct by international
4
collaborations with regards to (1) the categories of what behaviors count as misconduct, (2)
5
the extent to which researchers are responsible for the misconduct of their collaborators, (3)
6
the conditions under which researchers are culpable for their own behavior.
7
8
9
4.2 Credibility
10
The second issue concerns ethics more directly (rather than the law): the credibility of the
11
ethical content in codes of conduct. As mentioned in the introduction, with ‘credibility’ we
12
have in mind how there is considerable skepticism about the effectiveness of self-regulation,
13
not only by corporations
25
and but also by professions.
26
After all, codes of conduct are one
14
of the main tools for self-regulation, since they guide the scientific community in regulating
15
their own behavior, including dealing with their own cases of misconduct. How credible are
16
codes of conduct as effective tools for self-regulation? This is increasingly controversial.
17
Some have been calling for more robust external regulation, for instance the form of an
18
increased criminalization of scientific misconduct.
27
Such skeptics would argue that codes of
19
conduct are primarily window-dressing, hiding a harsher, competitive reality. In this view, the
20
only way to regulate credit-maximizers is not to appeal to their sense of ethics, but to change
21
their calculus by means of retribution and punishment.
22
This consideration can be situated in the broader discussion about what precisely the
23
purpose is of professional codes of ethics, and what the reasons are for complying with them.
28
24
25
Short, J.L., & Toffel, M.W. (2010). Making Self-Regulation More Than Merely Symbolic: The Critical Role
of the Legal Environment. Administrative Science Quarterly. 55(3), 361–396.
https://doi.org/10.2189/asqu.2010.55.3.361
26
Carvalho, T., Correia, T., & Serra, H. (2018). Professions under suspicion: what role for professional ethics
and commitment in contemporary societies? Sociologie, Problemas e Práticas. 88, 9–25.
https://doi.org/10.7458/SPP20188814795
27
Collier, R. (2015). Scientific misconduct or criminal offence? CMAJ : Canadian Medical Association
Journal. 187(17), 1273–1274. https://doi.org/10.1503/cmaj.109-5171Sovacool, B.K. (2005). Using
criminalization and due process to reduce scientific misconduct. The American Journal of Bioethics: AJOB.
5(5), W1-7. https://doi.org/10.1080/15265160500313242
28
Spielthenner, G. (2015). Why comply with a code of ethics? Medicine, Health Care and Philosophy. 18(2),
195–202. https://doi.org/10.1007/s11019-014-9594-5; Dobson, J. (2005). Monkey Business: A Neo-Darwinist
Approach to Ethics Codes. Financial Analysts Journal. 61(3), 59–64. https://doi.org/10.2469/faj.v61.n3.2728

23
It is also a particular instance of the yet broader (and perennial) philosophical problem of how
1
ethics can become subordinated to dynamics of power and prestige.
29
2
Mindful of these broader implications, for purposes here we will understand “window-
3
dressing” in the following way: if the ethical content of codes of conduct is merely window
4
dressing, then its primary function is to create a perception to the outside world that a
5
community cares about ethical behavior. By implication, the regulatory function to guide the
6
actual behavior of the members of the community is either secondary or non-existent. In other
7
words, if a code is mere window-dressing, then it is more important that ethical values are
8
present rather than which precise ethical values are listed in the code.
9
This has two implications for the content of codes. The first is that the listed values in
10
the resulting ethical frameworks can expected to be rather generic and ambiguous with regards
11
to what precise behaviors are being promoted. This is, for instance, how Giacomini et al.
30
12
understand the concept of window-dressing in their investigation of the ethical content of
13
Canadian health care policies. By means of a qualitative content analysis, they conclude:
14
Currently, many ethics frameworks seem detached from policies, floating alongside in
15
special text boxes or appendices. They are also set apart by vague, platitudinous
16
language; imperatives are typically expressed in terms that are inspirational, positive,
17
nonspecific, and non-operational. (…) Some degree of flexibility is appropriate and
18
necessary in ethical analysis. However, expansive ethical terms with little definition
19
or operationalization become weak rhetoric, and apparent agreements (e.g., to include
20
“accountability” among guiding principles) may mask fundamentally different aims
21
(e.g., involving public participation vs. transparent reporting).
31
22
23
We cannot rule out an analogous finding out with regards to RI codes of conduct. The values
24
listed by the 2017 ECoC that are most commonly duplicated by national-level documents –
25
‘honesty’ and ‘respect’, see Figure 5 – are arguably precisely those which are applicable to
26
any economic activity or interpersonal relation. These values could be deemed generic insofar
27
they are not specifically applicable to the activity of scientific research, and hence one could
28
legitimately ask what precise guiding function they have for the activity of scientific research
29
(rather than any economic activity or any interpersonal relation).
30
However, a systematic testing of the principles of RI for their precise information
31
content is outside the scope of the minimalist methodology used in this study. Moreover, while
32
29
Evans, J.H. (2012). The history and future of bioethics: a sociological view. New York: Oxford University
Press.Callahan, D. (2005). Bioethics and the culture wars. Cambridge Quarterly of Healthcare Ethics. 14(4),
424–431. https://doi.org/10.1017/s0963180105050577
30
Giacomini, M., Kenny, N., & DeJean, D. (2009). Ethics frameworks in Canadian health policies: Foundation,
scaffolding, or window dressing? Health Policy. 89(1), 58–71. https://doi.org/10.1016/j.healthpol.2008.04.010
31
Ibid. : 67

24
we admit that the principles in the surveyed documents are often generic, we would not wish
1
to go so far as Giocomini et al. in suggesting they are ‘weak rhetoric’, since many of the ethical
2
values are connected to good practices and categories of misconduct. Nonetheless, testing the
3
genericity of ethical content in codes of conduct is a legitimate research question and could
4
be subject to future study.
5
Here we discuss a second textual implication of ‘window dressing’, namely that, if the
6
ethical content is not intended to play a genuine regulatory function, one can expect different
7
codes to list very different ethical values without any discernable rationale for the divergence.
8
Thus, if ethical content is mere window-dressing, then it would not matter if a code listed
9
“honesty, respect, integrity” or “accountability, accuracy, and truthfulness” – as long as there
10
was some ethical content to create the perception of that the research community cares about
11
integrity.
12
Note the direction of the inference in the previous paragraph: from window-dressing
13
to divergence. Hence, conversely, given the observed divergences in ethical values, we cannot
14
simply conclude from this that the leading regulatory documents are by and large intended as
15
window-dressing. There are namely other possible explanations for the divergence that we
16
must consider. We limit the discussion to three explanations: (1) different envisaged uses of
17
the leading regulatory documents, (2) different legal and institutional contexts, (3) honest
18
disagreement about a complex subject matter. Each of these explanations is a possible
19
response to the skeptical stance.
20
Different envisaged uses of the documents? The first explanatory factor to consider
21
is that the content of the different regulatory documents may have been strongly influenced
22
by different envisaged uses (see Figure 3). For instance, the Norwegian General Guidelines
23
for Research Ethics is only a single page and contains concise statements about the nature of
24
science, and advice for action. By contrast, the Swedish Good Research Practise is over fifty
25
times longer than the Norwegian document, and contains detailed discussions about different
26
ways of defining misconduct, or philosophical reflections on the difference between law and
27
ethics. It is clear that these two documents have very different envisaged uses by researchers.
28
Another example is the way in which some guidelines contain directives on institutional
29
responsibilities, such as The Netherlands Code of Conduct for Research Integrity, the Croatian
30
Code of Ethics, or the UK’s Concordat to Support Research Integrity. By contrast, other codes
31
like the Ethical Code of Scientific Research in Belgium do not emphasize institutional
32
responsibilities to the same extent, and thus mainly target individual researchers.
33

25
National-level codes may also wish to be more precise than the ECoC. For instance,
1
among violations the ECoC includes “exaggerating the importance and practical applicability
2
of findings”.
32
However, since the ECoC does not define what constitutes ‘exaggeration’, and
3
since exaggeration is otherwise a context-dependent and slippery term, authors of national-
4
level documents may have believed it necessary to add more concrete detail. This seems to
5
have been the case for the Dutch leading regulatory document, where relatively precise
6
standards are offered (see discussion in section 4.1).
7
Nonetheless, while the envisaged use could explain some divergence – for instance
8
the detail with which misconduct is defined – at stake here is the observed divergence in
9
principles and values. Moreover, the universal character of RI values is a fundamental
10
presupposition of the core-periphery model,
33
and is a presupposition we also make for
11
purposes of this study.
34
Thus what principles are listed should not depend on whether the
12
document is intended as a proto-legal document or an ethical statement. In sum, divergence
13
in envisaged uses can be a sufficient explanation for the divergence in listed RI principles.
14
Different legal and institutional contexts? A second explanatory factor is that each
15
national-level regulatory document is designed to be finely attuned to specific needs or legal
16
constraints within the national context. As mentioned in the introduction, the background
17
paper to the 2011 ECoC acknowledges explicitly that national-level documents could and
18
even should diverge.
19
However, while this factor surely can explain some divergence in content – for
20
instance, divergence in understandings of responsibility and culpability – also this factor
21
hardly can explain much of the observed divergence with regards to ethical values. That there
22
are differences in “national traditions” and “legislative regulations” between European
23
countries is undeniable, but once again, this potential explanation of the observed differences
24
is likewise insufficient given the universal character of RI principles. Even across different
25
national traditions and legislative regulations, the fundamental principles of research integrity
26
remain unchanged.
27
Honest disagreement? A final, and in our view the most plausible, explanation for the
28
divergence is honest disagreement between the authors of the various codes and guidelines.
29
The nature of research integrity is, after all, a very complex issue. It depends crucially on what
30
“good” science is, and defining the latter presupposes an answer to all of the following,
31
32
ALLEA (op. cit. n. 5), p. 8.
33
ESF-ALLEA (op. cit. n. 2), p. 14.
34
Questioning and justifying this presupposition would require a different, more philosophical paper.

26
difficult questions: What is good methodology? What are good collegial and collaborative
1
relations between scientists? How much can scientists be influenced by societal and moral
2
considerations while keeping their integrity?
35
3
Thus, for instance, three examples of a different listing of RI principles are: “generally
4
valid ethical values and thoughtfulness, respect, courtesy, and honesty” (SK), “dignity,
5
responsibility, equity, correctness, diligence” (IT), or “honesty, responsibility, scientific
6
integrity” (FR). If this is due to honest disagreement, then authors simply have a different idea
7
of what the principles of RI are. This is plausible given the complexity of the nature of RI.
8
A reason in support of this explanation is that the authors of national-level documents
9
can be assumed to be experts on RI in their national context, and hence one can assume that
10
any deviation, no matter how small, reflects a conscious deliberation. These experts are
11
presumably familiar with the ECoC, hence even a small substitution of ‘honesty’ for
12
‘transparency’ can be assumed to be reflect a different understanding of what precisely
13
‘research integrity’ amounts to. Even hypothetically, if they chose to disregard the ECoC and
14
wished to construct a new code or guidelines from the bottom up, this would still be a
15
conscious choice insofar it would reflect a dissatisfaction with the ECoC. (By contrast, if we
16
drop the assumptions that the authors are experts who have deliberated over deviations from
17
the ECoC, we end up with the skeptical window-dressing hypothesis, where it matters that a
18
code is written rather than how the code is written.)
19
However, if the divergence is due to honest disagreement, then the skeptical
20
hypothesis is not strongly undermined. The honest disagreement then would point to a lack of
21
common understanding about what precisely the regulatory function of RI principles is. What
22
difference does it make to researcher behavior whether to list ‘respect, courtesy, or honesty’
23
versus ‘dignity, responsibility, equity’? The sophisticated skeptic could acknowledge that the
24
intentions of each individual author may have been noble, and that the disagreement may have
25
been honest, but that, since they did not coordinate with each other to clarify what precisely
26
the regulatory function is of RI principles, authors unwittingly acted to provide window-
27
dressing without ensuring that actual researcher behavior would be optimally impacted. The
28
fact that codes have sometimes been authored in response to high-profile scandals of research
29
35
For a physicist or biologist, the answer to that question might seem easy (i.e., there should be zero influence),
but for an economist or sociologist the issue is not so straightforward (see e.g. Sen 1991).

27
fraud
36
only strengthens the skeptical view that codes are about managing perceptions rather
1
than about changing behavior.
2
In this respect, the contrast between the divergence on RI principles and the
3
convergence on FFP provisions is striking. FFP provisions are understood to play a clear
4
regulatory function, and this is reflected by a total lack of divergence across various national-
5
level documents on this issue (even if they do on the more detailed issues of responsibility
6
and culpability). FFP provisions matter, but it is less clear whether and how RI principles
7
matter. The skeptic would point to this issue, and would explain the state of honest
8
disagreement between authors of codes/guidelines as due to the fact that RI principles do not
9
play any guiding influence on researcher behavior, and so authors are free to pursue their own
10
intuitions, unconstrained by objective regulatory function.
37
11
We do not hold the skeptical view, but do believe it should be taken seriously. The
12
current drive for increasing attention to research integrity issues would be partially self-
13
defeating if the upshot were that the skeptical view would be strengthened. Moreover, this
14
skeptical view is far from being a fringe conspiracy theory: it has underpinned the increased
15
external regulation of other professions such as medicine or law in the recent decades, for
16
instance through New Public Management.
38
We would also like to note the long tradition in
17
sociology and philosophy of taking a skeptical-deflationary view towards professional codes
18
of conduct – and even towards ethics as such.
39
19
Summary. Ultimately, while we are most sympathetic to the explanation that the
20
divergences are due to honest disagreement, we do acknowledge that the current textual state
21
of RI codes does not allow one to satisfactorily falsify the skeptical hypothesis (i.e., that codes
22
of conduct are window-dressing). In most codes of conduct, no reason is given for diverging
23
from the ECoC on core aspects of RI – the divergence is often not even acknowledged. If the
24
36
DFG (German Research Foundation). (2013). Sicherung guter wissenschaftlicher Praxis / Safeguarding Good
Scientific Practice. Retrieved from
http://www.dfg.de/download/pdf/dfg_im_profil/reden_stellungnahmen/download/empfehlung_wiss_praxis_131
0.pdf, p. 64.
37
The skeptic would respond similarly to the idea that the authors of national-level leading documents deviate
from the ECoC in order to inject some originality into the documents. When it concerns FFP provisions, originality
is not only not necessary, but also undesirable. If originality is pursued with regards to RI principles, this is only
a reflection of the fact that there is no objective regulatory function of this part of codes/guidelines, and the skeptic
would charge that the originality is tolerated because it real function of codes/guidelines is to provide window-
dressing. (We thank an anonymous referee for bringing our attention to the possibility of originality.)
38
Carvalho, T., & Correia, T. (2018). Editorial: Professions and Professionalism in Market-Driven Societies.
Professions and Professionalism. 8(3), e3052. https://doi.org/10.7577/pp.3052
39
Foucault, M. (2012). Discipline and Punish: The Birth of the Prison. Knopf Doubleday Publishing Group;
Goldstein, J. (1984). Foucault among the Sociologists: The ‘Disciplines’ and the History of the Professions.
History and Theory. 23(2), 170. https://doi.org/10.2307/2505005; Larson, M.S. (1977). The Rise of
Professionalism: A Sociological Analysis. University of California Press.

28
divergences reflected reasoned and conscious disagreements about RI, then one would expect
1
some preamble stating the presuppositions of the code of conduct. Yet, few had such
2
preambles, as is reflected in Figure 3, where the distribution of word counts is clearly skewed
3
towards the shorter word counts. The predominant goal is providing clear, actionable advice,
4
but since the end result is that different clear and definitive advices are given, some skeptics
5
may seize upon this to undermine the credibility of RI codes of conduct as such.
6
7
4.3 The need for a normative framework of RI codes and guidelines
8
The previous discussion underlines the need for a more systematic normative framework that
9
can be used to evaluate descriptive differences in RI codes and guidelines. The two issues
10
discussed here – fairness in dealing with allegations of misconduct, and the threat that codes
11
of conduct are viewed as mere window-dressing – suggest two dimensions that such a
12
normative framework would incorporate.
13
The first dimension would be a jurisprudential one that would help evaluate legal
14
consequences of RI codes. The need for such a framework can also be observed in previous
15
studies. For instance, Godecharle et al. introduce a distinction between norms and values in
16
order to analyze content of codes of conduct, understanding the latter as “universal” but the
17
latter as “embedded in a specific context: situation, time, and place”.
40
They use this
18
distinction to categorize codes on a value-focused vs. norm-focused continuum, with as an
19
example of the former the Belgian Code of Ethics for Scientific Research in Belgium, and an
20
example of the latter the Danish Code of Conduct. Moreover, they understand the distinction
21
between values and norms to correspond to the distinction between morality-based approaches
22
and legalistic approaches.
23
Yet, while it may be useful in some contexts to distinguish between ‘legalistic’ and
24
‘ethics-focused’ approaches to research integrity and research misconduct,
41
codes of conduct
25
must nonetheless cover both approaches. When findings of misconduct are sanctioned by
26
integrity commissions, researchers often have the legal right to challenge such findings. Thus,
27
the question is what legal implications a code of conduct has, not whether the code of conduct
28
avoids legal terminology.
29
40
Godecharle, S., Nemery, B., & Dierickx, K. (2014). Heterogeneity in European Research Integrity Guidance:
Relying on Values or Norms? Journal of Empirical Research on Human Research Ethics. 9(3), 79–90.
https://doi.org/10.1177/1556264614540594
41
See also Pennock, R.T. (2019). An instinct for truth: curiosity and the moral character of science. Cambridge,
MA: The MIT Press.

29
Such considerations lead to jurisprudential questions about what the purpose is of RI
1
legislation and of the sanctioning of research misconduct: is it, for instance, retribution and
2
deterrence, or is it rehabilitation? These questions matter for how fine-grained the categories
3
of misconduct will be. Thus, if the aim of sanctioning is retribution or deterrence, then the
4
sanction will harm the misbehaving researcher (in a way proportional to the transgression);
5
hence, to avoid a chilling effect on scientific communities, the evidential bar for a misconduct
6
finding would need to be relatively high. This implies that many of the less serious forms of
7
misconduct would not be categorized as transgressions. By contrast, if the aim of sanctioning
8
is rehabilitation or education, then the sanction does not necessarily harm the misbehaving
9
researcher, and lesser forms of misconduct could in principle be sanctioned without a chilling
10
effect on scientific communities.
11
While a further exploration of questions would lead us far beyond the scope of the
12
current paper, this brief discussion illustrates the need for a normative framework with a
13
jurisprudential dimension that identifies the ‘proper aim’ of the sanctioning of misconduct.
14
Such a framework would not only help evaluate current codes of conduct for their legal
15
implications, but would also help in shaping future codes of conduct.
16
The second dimension is sociological. Ethical choices are not made in isolation from
17
the large system of incentives (grants, tenure, promotion, etc.) that may not always promote
18
integrous behavior. What system of incentives is desirable? How should responsibilities be
19
organized? These questions pertain to what social structures are to be deemed as desirable,
20
and today a large choice facing the scientific community is whether to promote professional
21
social structures, where researchers and communities of researchers follow their own
22
judgment, or bureaucratic social structures, where researchers follow procedures and
23
processes that are defined by non-researchers (e.g. corporate leaders or administrators). In the
24
first case, individuals must navigate a system of incentives that is not always oriented towards
25
the ‘service ideal’;
42
in the second case, individuals are averted as much as possible from such
26
dilemma’s.
27
Previous studies sometimes inadvertently depend on this distinction in their normative
28
evaluation of differences between codes of conduct. Thus, in their analysis of ethical
29
frameworks used by health care policies, Giacomini et al. distinguish between values or
30
themes that lie on a continuum between being “more like ethical principles” and “more like
31
42
Freidson, E. (2001). Professionalism, the Third Logic: On the Practice of Knowledge. University of Chicago
Press.

30
policy processes or goals”.
43
Thus, policy processes seek to strengthen bureaucratic structures;
1
ethical principles (as service ideals) serve to strengthen professional structures.
2
The sociological dimension helps evaluate existing codes of conduct, as well as inform
3
future codes of conduct. If the goal is to strengthen professionalism, then listing generic values
4
such as ‘respect’ or ‘honesty’ are likely insufficient to actually impact researcher behavior.
5
Take honesty for instance: honest scientific communication does not mean turning scientific
6
papers into chronicles of all details regarding the research process. There is a distinction
7
between honesty and pedantry; yet, at what point does the selection of data and insights stop
8
being honest, and start being biased or ideological? In true Aristotelian fashion, ‘honesty’ is a
9
virtue situated in the middle between the extremes of pedantry and bias or deceit, but simply
10
listing ‘honesty’ in codes of conduct does not actually guide researchers in how to be honest.
11
In a bureaucratic approach, values such as honesty are not intended to actually impact
12
behavior: what matters is how such values are fleshed out in terms of procedures and processes
13
(e.g., to pre-register research protocols, keeping data for future reproduction efforts, etc.). By
14
contrast, in a professionalism approach, values such as honesty are intended to directly impact
15
behavior, but they would need to be fleshed out in terms of research goals (e.g., truth,
16
understanding), and would leave the details of how to achieve those goals in particular
17
circumstances up to the individual judgment of researchers. Some groundwork in this regard
18
has been done about the desired social structures of the scientific community,
44
but more
19
remains to be said, and the current study underlines the need of a normative framework with
20
not just a jurisprudential dimension, but also a sociological one, such that the sociological
21
implications of a codes of conduct can be evaluated.
22
23
5. Conclusion
24
25
The ECoC was intended to help harmonize research integrity standards across Europe, by
26
identifying the ‘core’ elements of RI: especially the definition of misconduct and the universal
27
principles of RI. In this study we investigated whether this ‘core-periphery’ model of
28
harmonization is realized by examining patterns of replication of the ECoC. We found that
29
the only unambiguous consensus concerns FFP provisions; we also found a relatively large
30
consensus that the value of ‘honesty’ is central to research integrity. Otherwise, with regards
31
43
Giacomini, Kenny, DeJean (op. cit. n. 29), figure 1.
44
Desmond, H. (2019). Professionalism in Science: Competence, Autonomy, and Service. Science and
Engineering Ethics. https://doi.org/10.1007/s11948-019-00143-x

31
to less egregious forms of misconduct, or with regards to other principles of RI, the consensus
1
is not obviously present.
2
This presents some scope for injustice: divergences between national-level documents
3
give grounds for diverging evaluations of joint misconduct by international collaborations.
4
However, most insidiously, we argued that it gives ammunition to skeptics of self-regulation,
5
who believe that the primary purpose of a code of conduct is to provide window-dressing, i.e.,
6
a perception that integrity matters but while actual behavior is driven by the cold competitive
7
calculus of incentive maximization. It is not a coincidence that there is such an unambiguous
8
consensus regarding FFP: findings of FFP, not coincidentally, have very large repercussions
9
for reputations, careers, and even livelihoods, and hence the precise wording of what allows
10
for a finding of FFP is important. Does this mean that the precise wording of RI principles is
11
thought not to matter? While we cautioned against drawing this conclusion (another
12
explanation would be honest disagreement about the principles of RI), we also argued that it
13
cannot be ruled out altogether, and that this alone is problematic enough to be highlighted.
14
In this way, while this study targeted the European regulatory situation, it ultimately
15
raised fundamental questions about how codes of conduct should be constructed and about
16
what precise impact we want these codes to have. What are precisely the values that are
17
specific to the activity of scientific research? What legal approach to sanctioning research
18
misconduct is appropriate? These questions must be addressed if the “common
19
understanding” aimed at by the original ECoC is to be achieved in any substantial way.
20
21
6. Author Contributions
22
23
Hugh Desmond: (1) conceived the idea for an overview study, (2) designed the methodology
24
for documents and identified the core-periphery model, (3) collected the documents, (4)
25
selected the leading documents, (4) analyzed and visually represented the differences
26
between documents, (5) interpreted the differences, and (6) wrote the successive drafts of the
27
paper.
28
29
Kris Dierickx: (1) conceived the idea for an overview study, and (2) commented on important
30
intellectual content in successive drafts of the paper.
31
32
7. Acknowledgments
33

32
The authors wish to thank the audiences at the EQUATOR conference in Berlin and Benoit
1
Nemery de Bellevaux for feedback, and for verifying the choice of leading regulatory
2
document, the authors thank: Ana Sofia Carvalho (PT), Marco Cosentino (IT), Anna
3
Domaradzka (PL), Nicole Foeger (AT), Nina Gennebäck (SE), Bjørn Hofmann (NO), Søren
4
Holm (UK), Peter Kakuk (HU), Juraj Koppel (SK), Petr Kratochvíl (CZ), Simo Kyllonen (FI),
5
Dirk Lanzerath (DE), Niels Mejlgaard (DK), Signe Mezinksa (LV), the Research Ethics
6
Committee of Bucharest University (RO), Margit Sutrop (EE), Emmanuele Valenti (ES), Edita
7
Žiobiene (LT).
8
9
8. Funding
10
11
This research was supported by funding from the European Union’s Horizon 2020 Research
12
and Innovation program under Grant Agreement No. 741782 (ENTIRE consortium). The
13
funders had no role in the study design, data collection and analysis, decision to publish, or
14
preparation of the manuscript.
15

33
9. Raw Data
Values of RI
Good practices
Definition of Misconduct
ECoC
2011
8 values † (Honesty, Reliability, Objectivity, Impartiality and
independence, Open communication, Duty of care, Fairness,
Responsibility for future science generations)
5 areas discussed (Data practices; research procedures; care for
research subjects, publication-related conduct and conflict of
interest; reviewing and editorial issues.
FFP plus ‘failure to meet clear ethical and legal requirements’ plus
‘minor misdemeanors’.
AT
6 values (transparent and sincere communication with scientists;
reliability; impartiality; openness to criticism; fairness; transparent
communication to public)
6 areas† (data practices; transparent handling of other persons’
work; no republication of texts; authorship; conflict of interest;
funding)
FFP plus 5 categories (unjustified refusal to share data; obstructing
others’ research; sabotage; inaccuracies in grant proposals;
discrimination against junior scientists and whistle-blowers)
CZ
7 values ‡ (respect moral values, critical attitude, knowledge,
precision and objectivity, completeness and verifiability,
accountability, confidentiality)
3 areas‡ (publishing; relations with students and co-workers;
Assessment, Evaluation, Expert Activities)
FFP plus 4 categories forgery, distortion, deliberate deception, and theft
DK
3 values (Honesty, Transparency, Accountability)
6 areas (Research planning and conduct; Data management;
Publication and communication; Authorship; Collaborative
research; Conflicts of interest; Teaching, Training, and Supervision)
FFP plus all “serious violations of good scientific practice”
ES
4 values (exercising methodological doubt; designing good
experiments; managing data; proper use of funding)
10 areas† (design; data management; funding; training; supervision;
cooperation; publication; authorship; peer-review; conflict of
interest)
FFP plus exaggerated interpretation of data
FI
3 values (integrity, meticulousness, accuracy)
7 areas† (data acquisition, evaluation, publishing, citation, ethical
review, collaboration, conflict of interest)
FFP or misappropriation categorized as misconduct. Separate
categories for: ‘disregard for the responsible conduct of research’ (5
types of behavior referenced) and ‘other irresponsible practices’ (7
types of behavior referenced).
HR*
4 values (honesty, academic excellence and freedom, mutual
respect and human dignity, personal responsibility and
accountability of institutions)
6 areas† (data management, authorship, protection of respondents,
care of animals, social responsibility, supervision of students,
conflict of interest)
FFP plus seven additional categories (sabotaging the work of other
scientists, duplicate publication, submitting same manuscript to
multiple journals, abuse of authorship, deliberately misrepresentation of
one’s own work, conflict of interests)
IE
Replication (Honesty, Reliability, Objectivity, Impartiality and
independence, Open communication, Duty of care, Fairness,
Responsibility for future science generations)
2 areas (Education, data storage and retention)
FFP plus 5 categories of behavior (data-related poor practice;
Publication-related practice; Personal behaviors; Financial and other
malpractice; Poor research procedures)
LT
6 values † (paraphrase by HD: honesty, pursuit of truth, accuracy
and reliability, objectivity, impartiality and correctness, respect
for human and animal rights)
2 areas dissemination practices and examination/supervision
practices
FFP
NO
4 values (Respect, good consequences, fairness, integrity)
4 area (data collection, informed consent, good reference practices,
conflict of interest)
FFP. From The ‘Research Ethics Act’ (2017): FFP or “other serious
breaches of recognized research ethical norms committed intentionally
or grossly negligent in planning, conducting or reporting research.”
PL
4 values (honesty at all stages of research; accountability for the
research and its exactness; professional kindness and fairness;
appropriate management of research)
6 areas (planning and conduct of research; research results
documentation; publication and communication of results;
authorship; collaborative research; conflicts of interest)
FFP; 5 other categories of behavior qualify as QRPs
Table 2. The elements of RI documented exhaustively, as much as possible in original formulation, for national-level leading documents published after ECoC 2011.
* = extraction based on Google translate
† = some interpretation by HD was necessary to extract values from text.
‡ = a lot of interpretation by HD was necessary to extract values from text.

34
ECoC 2017
4 values (reliability, honesty, respect,
accountability)
8 areas (research environment; training, supervision and
mentoring; research procedures; safeguards; data practices
and management; collaborative working; publication and
dissemination; reviewing, evaluating, and editing)
FFP plus a non-exhaustive list of 13 additional categories (manipulating authorship, self-plagiarism, citing
selectively, withholding results, conflict of interest, expanding the bibliography unnecessarily, malicious
accusation of misconduct, misrepresentation, exaggeration of research, delaying research of others, misusing
seniority, ignoring potential violations of RI, supporting predatory journals)
DE
4 values† (observing professional standards,
documenting results, consistently questioning
one’s own findings, practicing strict honesty)
5 areas (fundamentals of scientific work, cooperation and
leadership responsibility in working groups, mentorship for
young scientists and scholars, securing and storing primary
data, scientific publications.)
FFP
EE
6 values (freedom, responsibility, honesty and
objectivity, respect and caring, justice, openness
and cooperation)
5 areas (planning of research; conduct of research;
authorship/publishing/application of research results;
researcher in the research community; observance, promotion,
and application of research integrity)
Explicit prohibitions of†: FFP plus 5 categories (drawing unsubstantiated conclusions from data (2.2.2), gift
authorship (3.2.2), biased reviewing (3.5.2), simultaneous submission (3.5.3), switching publisher after
conditional acceptance (3.5.4))
FR
3 values (honesty, responsibility, scientific
integrity. The latter is defined as the “refusal to
allow scientific values to be corrupted by
motivations for financial gain or public
recognition”, see p. 4 of FR2 in supplementary
materials)
7 areas (compliance with legislative and regulatory
requirements; reliability of research work; communication;
responsibility in collective work; impartiality and
independence in assessment and expertise; collaborative work
and plurality of activities; training)
FFP, plus 10 categories (conflicts of interest; intentional misrepresentation or erroneous quotation of research
carried out by competitors; deliberate omission of contributions made by other authors in references;
incorrect indications on the progress of the researcher's own work; overestimation of the applicability of the
research findings; addition of "guest" or "ghost" authors to the list of authors; omission of anyone who made
a significant contribution to the project from the list of authors; listing co-authors without their consent;
republishing parts of previous publications without citing the original source)
IT
5 values (dignity, responsibility, equity,
correctness, diligence)
6 areas (research planning; research execution; publication;
reviewing people, projects or publications; relationships
within research institutions, relations with colleagues and
supervision; public communication and dissemination of
results)
FFP plus 15 categories (mismanagement of conflicts of interest; carelessness and misuse of data; data theft;
multiple publications; sending multiple proposals; carelessness and abuse of article signing; failure to
correct and correct their scientific production; neglect and abuse in performing the role of auditor or
manager / director of a research group; carelessness and abuse in the performance of the role of editorial
manager; falsification of scientific credentials in the submission of publications or projects, or in the
participation in a call for tenders; sabotage of colleagues; instigation, facilitation, connivance, omertà;
malevolent accusations and obstacles to the investigation of misconduct; non-transparent or inappropriate
use of research funds; neglect and abuse of one's role)
NL
5 values (honesty, scrupulousness, transparency,
independence, responsibility)
6 areas (design; conduct; reporting results; assessment and
peer review; communication; general standards)
FFP, and give a list of criteria by which to judge individual cases
LV*
6 values‡ (a call to understand, respect for moral
norms, respect for professional norms, critical
attitude, integrity, respect for colleagues)
6 areas† (research; teaching; reviewing articles; reviewing
proposals; debate; public dissemination)
Explicit prohibitions of plagiarism, falsification, duplicate publication.
PT*
Replication: 4 values (reliability, honesty,
respect, accountability)
No explicit discussion of good practices.
FFP plus 6 categories (conflict of interest; manipulation and violation of authorship criteria; inadequate
protection of persons participating in investigations and protection of animals in research; absence of
adequate publication criteria; shared responsibility between investigators and other inadequate team
members; and ineffective guidance and supervision)
SE
No explicit stipulation of RI values is made. The
8 ALLEA 2011 values are mentioned as helpful,
but do not structure the discussion.
5 areas (ethics review; handling of research material;
collaboration; publishing; supervision, teaching, reviewing,
committee work)
FFP (“Research misconduct entails actions or omissions in research, which – consciously or through
carelessness – lead to falsified or manipulated results or give misleading information about someone’s
contribution to the research.”)
SK*
5 values (generally valid ethical values and
thoughtfulness, respect, courtesy, and honesty)
6 areas (research; publishing; assessment; review, evaluation
and expert activities; behavior towards co-workers and
students)
FFP + 4 categories (e.g. manipulation of authorship, delaying publication of an article through peer review,
malicious accusation of breach of RI, misrepresentation)
UK
4 values (honesty, rigour, transparency and open
communication, care and respect)
3 areas (standards of RI, culture of RI, strengthening RI). 3
types of agent (researchers, employers, funders)
Definition is FFP + 2 categories (failure to meet ethical, legal and professional obligations, improper
dealing with allegations of misconduct)
Table 3. The elements of RI documented exhaustively, as much as possible in the original formulation, for national-level leading documents published after ECoC 2017.
* = extraction based on Google translate
† = some interpretation by HD was necessary to extract values from text.
‡ = a lot of interpretation by HD was necessary to extract values from text

35
10. List of Selected Leading Documents
1
2
1. AT
3
Title: OeAWI Guidelines for Good Scientific Practice
4
Author: Austrian Agency for Scientific Integrity
5
Date: 2015
6
URL: https://oeawi.at/wp-content/uploads/2018/06/Brosch.-GWP-
7
Richtlinien-WEB-2017_neu-1.pdf
8
Word Count: ca. 1860
9
Languages of document: English and German
10
11
2. BE
12
Title: Code of Ethics for Scientific Research in Belgium.
13
Author: Royal Flemish Academy of Belgium for Science and the Arts, The
14
Royal Academy of Science, Letters and Fine Arts of Belgium.
15
URL: http://www.belspo.be/belspo/organisation/publ/eth_code_nl.stm
16
Date: 2009
17
Word Count: ca. 2290
18
Language of document: Dutch
19
20
21
3. CH
22
Title: Scientific integrity: principles and rules of procedure.
23
Author: Swiss Academies of Arts and Sciences
24
Date: 2008
25
URL: http://www.akademien-
26
schweiz.ch/en/dms/E/Publications/Guidelines-and-
27
Recommendations/e_Integrity.pdf
28
Word Count: ca. 5350
29
Language of document: English
30
31
4. CZ
32
Title: Code of Ethics for Researchers of the Czech Academy of Sciences
33
Author: Czech Academy of Sciences
34
Date: 2016 (Latest version)
35
URL: http://www.avcr.cz/opencms/export/sites/avcr.cz/.content/galerie-
36
souboru/INc-16-12_AJsmm.pdf
37
Word Count: ca. 1980
38
Language of document: English
39
40
41
42
5. DE
43
Title: Guidelines for Safeguarding Good Research Practice
44
Date: 2019
45
Authors: German Research Foundation (DFG)
46
URL:
47
https://www.dfg.de/download/pdf/foerderung/rechtliche_rahmenbedingunge
48
n/gute_wissenschaftliche_praxis/kodex_gwp.pdf
49
Word Count: ca. 5820
50

36
Languages of document: English and German
1
2
3
6. DK
4
Title: Danish Code of Conduct for Research Integrity
5
Date: November 2014
6
Authors: Ministry of Higher Education and Science
7
URL:https://ufm.dk/en/publications/2014/the-danish-code-of-conduct-for-
8
research-integrity
9
Word Count: ca. 4610
10
Language of document: English
11
12
13
7. EE
14
Title: Estonian Code of Conduct for Research Integrity
15
Author: the Estonian Academy of Sciences, the Estonian Research Council,
16
and the Ministry of Education and Research.
17
Date: 2017
18
URL:
19
https://www.eetika.ee/sites/default/files/www_ut/hea_teadustava_eng_trukis.
20
pdf
21
Word Count: ca. 6380
22
Language of document: English
23
24
8. ES
25
Title: Code of Good Scientific Practices of CSIC.
26
Author: Spanish Research Council (CSIC)
27
Date: 2011
28
URL: https://www.cnb.csic.es/documents/CBP_CSIC.pdf
29
Word Count: ca. 3350
30
Languages of document: English and Spanish
31
32
9. FI
33
Title: Responsible conduct of research and procedures for handling
34
allegations of misconduct in Finland
35
Author: Finnish Advisory Board on Research Integrity
36
Date: 2012
37
URL: http://www.tenk.fi/sites/tenk.fi/files/HTK_ohje_2012.pdf
38
Word Count: ca. 5140
39
Language of document: English
40
41
10. FR
42
Title: Integrity and responsibility in research practices: a guide
43
Author: CNRS-CPU
44
Date: March 2017
45
URL: http://www4.cnrs-dir.fr/comets/IMG/pdf/comets-guide-en.pdf
46
Word Count: ca. 8090
47
Language of document: English
48
49

37
11. HR
1
Title: Etički Kodeks Odbora za Etiku u Znanosti i Visokom Obrazovanju
2
(Ethical Code of the Board of Ethics in Science and Higher Education).
3
Author: Agency for Science and Higher Education (ASHE)
4
Date: 2015 (consolidated text).
5
URL:
6
https://www.azvo.hr/images/stories/tijela_agencije/Eticki_kodeks_OE
7
ZVO_pro%C4%8Di%C5%A1%C4%87eni_tekst_nakon_izmjena_i_d
8
opuna_s_8._sjednice_15.6.15.doc
9
Word Count: ca. 2170
10
Language of document: Croatian
11
12
12. HU
13
Title: Science Ethics Code of the Hungarian Academy of Sciences
14
Author: Hungarian Academy of Sciences
15
Date: 2010
16
URL:
17
http://mta.hu/data/dokumentumok/english/background/Science_Ethics_Code
18
_English.pdf
19
Word Count: ca. 10670
20
Language of document: English
21
22
23
13. IE
24
Title: National Policy Statement on Ensuring Research Integrity in Ireland
25
Date: June 2014
26
Authors:
27
Irish Universities Association (IUA)
28
Health Research Board (HRB)
29
Royal Irish Academy (RIA)
30
Science Foundation Ireland (SFI)
31
Institutes of Technology Ireland(IoTI)
32
Higher Education Authority (HEA)
33
Dublin Institute of Technology (DIT)
34
Enterprise Ireland (EI)
35
Teagasc
36
Irish Research Council (IRC)
37
Royal College of Surgeons in Ireland (RCSI)
38
Quality and Quali cations Ireland (QQI)
39
URL:
40
http://hea.ie/assets/uploads/2017/04/National-Policy-Statement-on-
41
Ensuring-Research-Integrity-in-Ireland-2014.pdf
42
Word Count: ca. 5170
43
Language of document: English
44
45
46
14. IT
47
Title: Linee guida per l’integrità nella ricercar (Guidelines for Research
48
Integrity)
49
Date: 10 June 2015
50

38
Authors: Commission for Research Ethics and Bioethics of the Natinoal
1
Center for Research (Commissione per l’Etica della Ricerca e la Bioetica del
2
CNR)
3
URL:
4
https://www.cnr.it/sites/default/files/public/media/doc_istituzionali/linee-
5
guida-integrita-nella-ricerca-cnr-commissione_etica.pdf?v=1
6
Word Count: ca. 7650
7
Language of document: Italian
8
9
10
15. LT
11
Title: Mokslininko etikos kodeksas (Scientist’s Code of Ethics)
12
Author: Lithuanian Academy of Sciences (Lietuvos mokslu akademija)
13
Date: 2012
14
URL: http://www.lma.lt/mokslininko-etikos-kodeksas
15
Word Count: ca. 990
16
Language of document: Lithuanian
17
18
19
16. LV
20
Title: Scientist’s Code of Ethics
21
Date: 2017 (reapproval of 1997 code)
22
Author: Latvian Academy of Science, Latvian Council of Science
23
URL:
24
https://www.lzp.gov.lv/index.php?option=com_content&task=view&id=149
25
&Itemid=113
26
Word Count: ca. 1920
27
Language of document: English
28
29
17. NL
30
Title: Netherlands Code of Conduct for Research Integrity
31
Date: 2018
32
Authors: Koninklijke Nederlandse Akademie van Wetenschappen (KNAW),
33
et al.
34
URL: https://doi.org/10.17026/dans-2cj-nvwu
35
Word Count: ca. 7250
36
Language of document: Dutch
37
38
18. NO
39
Title: General guidelines for research ethics
40
Date: September 2014
41
Authors: The Norwegian National Research Ethics Committees
42
URL: https://www.etikkom.no/globalassets/general-guidelines.pdf
43
Word Count: ca. 880
44
Language of document: English
45
46
19. PL
47
Title: The Code of the National Science Centre on Research Integrity and
48
Applying for Research Financing
49
Author: National Science Centre
50

39
Date: 2016
1
URL: https://ncn.gov.pl/sites/default/files/pliki/Code-of-the-National-
2
Science-Centre-on-Research-Integrity.pdf
3
Word Count: ca. 6580
4
Language of document: English
5
6
20. PT
7
Title: Integridade na Investigação Científica: Recomendação (Integrity in
8
Scientific Research: Recommendation)
9
Author: Conselho Nacional de Ética para as Ciências da Vida (National
10
Council of Ethics for the Life Sciences)
11
Date: 2018
12
URL:
13
http://www.cnecv.pt/admin/files/data/docs/1523888172_IntegridadeCNECV
14
2018.
15
Word Count: ca. 4850
16
Language of document: Portuguese
17
18
21. RO
19
Title: Codul General de Etică în Cercatarea Ştiinţifică (General Code of Ethics
20
in Scientific Research)
21
Author: Ministerul Educaţiei, Cercetarii şi Tineretului; Autoritatea Natională
22
Pentru Cercatare Ştiinţifică; Consiliul National de Etică (Ministry of
23
Education, Research and the Youth; National Authority for Scientific
24
Research; The National Council of Ethics)
25
Date: 2007
26
URL:http://www.academiaromana.ro/consiliuCercetare/doc2007/ccc2007-
27
0913-IEI-CodEtica.doc
28
Word Count: ca. 3000
29
Language of document: Romanian
30
31
22. SE
32
Title: Good Research Practice
33
Author: Swedish Research Council
34
Date: July 2017
35
URL:
36
https://www.vr.se/download/18.5639980c162791bbfe697882/152948052947
37
2/Good-Research-Practice_VR_2017.pdf
38
Word Count: ca. 44 000
39
Language of document: English
40
41
23. SK
42
Title: Etický kódex SAV (Ethics Code of the Slovak Academy of Sciences)
43
Author: Slovak Academy of Sciences
44
Date: 2018
45
URL (original): https://www.sav.sk/php/download_doc.php?doc_no=7663
46
URL (appendix): https://www.sav.sk/php/download_doc.php?doc_no=7664
47
Word Count: ca. 1880
48
Language of document: Slovak
49
50

40
24. UK
1
Title: The Concordat to Support Research Integrity
2
Date: October 2019
3
Author: Universities UK
4
URL: https://www.universitiesuk.ac.uk/policy-and-
5
analysis/reports/Documents/2019/the-concordat-to-support-research-
6
integrity.pdf
7
Word Count: ca. 5370
8
Language of document: English
9
10
11