ArticlePDF Available

Maturity Level Assessments of Information Security Controls: An Empirical Analysis of Practitioners Assessment Capabilities

  • Capgemini Invent
  • Continental Automotive Technologies GmbH

Abstract and Figures

Maturity models are a widely used concept for measuring information security. The idea is to systematically evaluate the maturity of security-relevant processes in an organisation. This enables decision makers to get an overview of the implementation status of relevant processes to identify neuralgic points. Maturity models thus play a central role in the conception of information security management systems (ISMS). Some industries, for instance, the German automotive industry, have even established security maturity levels as the de facto standard for measuring information security. However, the quality of security maturity level assessments has not been sufficiently investigated yet. We have analysed to what extent security managers can accurately assess the maturity levels of security controls. To verify the quality of maturity level assessments a case study was conducted where security experts assessed a subset of the ISO/IEC 27002 security controls for a hypothetical scenario using the COBIT maturity levels. Additionally, ex-post interviews have been conducted with several study participants to verify some of the hypotheses developed during the previous analyses. Our results show that many security experts struggled with the task and did not perform well. However, we discovered professional characteristics that have a strong significant effect on the assessment capabilities. We also identified various types of additional support that can help practitioners to make more reliable assessments in practice. Moreover, the experts self-perception was overly optimistic when asked to assess their performance. We even found a weak inverted correlation for more experienced experts, also known as Dunning-Kruger effect. Our results have a strong impact on practise since they indicate that practitioners need support to carry out high-quality assessments and they also show what kind of support addresses the identified challenges.
Content may be subject to copyright.
Maturity Level Assessments of Information Security Controls:
An Empirical Analysis of Practitioners‘ Assessment Capabilities
Christopher Schmitz, Michael Schmid, David Harborth and Sebastian Pape
Goethe University Frankfurt, Germany
{christopher.schmitz, michael.schmid, david.harborth, sebastian.pape}
Maturity models are a widely used concept for measuring information security. The idea is to systematically evaluate the maturity
of security-relevant processes in an organisation. This enables decision makers to get an overview of the implementation status of
relevant processes to identify neuralgic points. Maturity models thus play a central role in the conception of information security
management systems (ISMS). Some industries, for instance, the German automotive industry, have even established security matu-
rity levels as the de facto standard for measuring information security. However, the quality of security maturity level assessments
has not been suciently investigated yet. We have analysed to what extent security managers can accurately assess the maturity
levels of security controls. To verify the quality of maturity level assessments a case study was conducted where security experts
assessed a subset of the ISO/IEC 27002 security controls for a hypothetical scenario using the COBIT maturity levels. Addition-
ally, ex-post interviews have been conducted with several study participants to verify some of the hypotheses developed during
the previous analyses. Our results show that many security experts struggled with the task and did not perform well. However,
we discovered professional characteristics that have a strong significant eect on the assessment capabilities. We also identified
various types of additional support that can help practitioners to make more reliable assessments in practice. Moreover, the experts’
self-perception was overly optimistic when asked to assess their performance. We even found a weak inverted correlation for more
experienced experts, also known as Dunning-Kruger eect. Our results have a strong impact on practise since they indicate that
practitioners need support to carry out high-quality assessments and they also show what kind of support addresses the identified
Keywords: Usability Evaluation, Security Controls, Maturity Levels, ISO/IEC 27002, COBIT
1. Introduction
During the last years, data breaches have been broadly re-
ported in the mainstream media and the number of security
breaches has increased by 11% since 2018 and 67% since
2014 [1]. Data breaches are not only expensive for organisa-
tions [2], organisations are even obliged by law to fulfil certain
regulatory requirements. For example, the General Data Pro-
tection Regulation (GDPR) in Europe requires organisations to
implement “a process for regularly testing, assessing and evalu-
ating the eectiveness of technical and organisational measures
for ensuring the security of the processing” [3, Art. 32]. How-
ever, following DeMarco, “you can’t control what you can’t
measure” [4, p. 3]. Thus, organisations face the problem that
information security can only be measured indirectly [5]. To
demonstrate compliance, organisations make use of maturity
level metrics of security controls which are mainly described
in standards and rely on the security knowledge of security ex-
perts [6].
The approach of assessing security controls with the aid of
maturity levels is widely used in many industries. It is, for
instance, the de facto standard for the German automotive in-
dustry when it comes to measuring information security. Or-
ganisations in this domain have to complete a standardised se-
curity self-assessment questionnaire (VDA-ISA) provided by
the German Association of the Automotive Industry (VDA)
where maturity levels are applied on security controls follow-
ing ISO/IEC 27002 [7]. The idea to assess the maturity levels
of ISO/IEC 27002 controls is also supported by standard GRC
(governance, risk, compliance) tools like risk2value which is
also used by major companies [8]. There also exist several
academic approaches relying on similar maturity-based ap-
proaches [9, 10, 11].
An established model for assessing security maturity levels is
oered by the ISACA COBIT framework. This 6-level model
makes it possible to evaluate the prevailing security situation in
an organisation. In combination with the de facto standard in in-
formation security, the ISO/IEC 27001 and 27002, a very gran-
ular evaluation can be carried out, namely on a control level.
Since maturity levels are assessed by humans, the human fac-
tor of the evaluation may lead to possible uncertainties, e. g.
because of dierent evaluators or dierent physical and mental
conditions of the same evaluator on dierent days. The quality
of these assessments has not yet been systematically evaluated.
Therefore, we conducted a case study where practitioners as-
sessed a subset of the ISO/IEC 27002 security controls for a
sample scenario (see Section Appendix A) using the maturity
Preprint submitted to Computers &Security April 12, 2021
levels of COBIT 5. We investigated deviations between dier-
ent participants, e. g. possible inaccuracies, in the assessment of
information security control maturity levels. Besides identify-
ing the possible range of answers, we also investigated whether
the participants’ results depend on professional factors such as
experience, industry, familiarity with maturity levels and re-
ceived certifications.
The remainder of this paper is structured as follows: Sec-
tion 2 provides background and reviews related work. Sec-
tion 3 introduces the methodology of our research. Section 4
describes the results along with an analysis in Section 5 and a
discussion in Section 6. Section 7 concludes our work. The
scenario and the questionnaire can be found in the appendix.
2. Background and Related Work
We briefly discuss dierent security maturity models, the en-
vironment where they are used and their relation to security
controls in the background section before we present related
work and compare it to our study.
2.1. Background
An established way to monitor and steer the information se-
curity is the implementation of an information security man-
agement system (ISMS). An ISMS is a comprehensive man-
agement framework through which an organisation identifies,
analyses and addresses its information risks. Using an ISMS
ensures that the security arrangements are fine-tuned to keep
pace with new security threats, vulnerabilities and their impacts
on the business processes. Many dierent standards and frame-
works are available providing guidance to correctly implement
and maintain an ISMS. The most popular standard in this envi-
ronment is ISO/IEC 27001 (incl. ISO/IEC 27002). It contains
guidelines for organisational information security management
practices including the selection, implementation and manage-
ment of controls, taking into consideration the organisation’s
information security risk environment(s). Another notable stan-
dard is from the National Institute of Standards and Technology
(NIST) within the 800-series publications [12] and Cybersecu-
rity Framework (NIST CSF) [13] which include information se-
curity management, information security evaluation, authenti-
cation and authorisation, etc. A country-specific standard from
the German Federal Oce for Information Security (BSI) is
called IT-Grundschutz (former English name: IT Baseline Pro-
tection Manual). BSI standards 200-1 to 200-3 and 100-4 cover
technical, organisational, infrastructural and personnel aspects
of information security [14]. The IT-Grundschutz is a collec-
tion of standards and catalogues that describe generalised pro-
cedures for the protection of the information technology used.
They are defined in exemplary modules, threats and measures
that were used to systematically construct the scenario accord-
ing to predefined scenario maturity levels. Another standard
which is very popular in the automotive industry is the VDA-
ISA [7], a questionnaire for assuring information security com-
pliance by the German Association of the Automotive Industry
(VDA) which is based on the ISO/IEC 27001 and 27002 stan-
These standards and frameworks have security controls in
common. Security controls describe a set of security measures
to fulfil a security requirement, and therefore to mitigate secu-
rity risks to physical properties, information, computer systems,
or other (also intangible) assets. These security controls address
various topics e. g. electronic signatures, software development,
business continuity, incidents etc.
Maturity models are a popular concept to assess the status of
information security as well as the quality of the process [15].
There is a large number of dierent security maturity models.
Proenc¸a et al. have identified more than 20 dierent security
maturity models for information systems. Almost all of them
dierentiate between 5 or 6 maturity levels [16]. Examples are
the Cybersecurity Capabilities Maturity Model (C2M2) devel-
oped by the United States Department of Energy in partner-
ship with the Department of Homeland Security [17], the In-
formation security maturity model (ISMM) [18] and the Open
Information Security Maturity Model (O-ISM3) [18]. ISMM
is intended as a tool to evaluate the ability of organisations
to meet the objectives of security and O-ISM3 aims to ensure
that security processes operate at a level consistent with busi-
ness requirements. Some of the most prominent models are
CMMI [19], SSE-CMM (ISO/IEC 21827:2008) and the CO-
BIT maturity model that is derived from the ISO/IEC 15504
(SPICE) [20, 21].
Using such models, it is possible to assess the quality of
controls on 6- to 7-point (from 0 to 5 or 6). The assessment
provides a management perspective in the fulfilment of regula-
tory requirements. The maturity levels are used as a measure
to quantify the implementation status of a security control. The
higher the maturity level of a control, the higher the chance that
it is performed in an eective and secure way so that it con-
tributes more to the organisational security. COBIT defines six
maturity levels. Each maturity level adds on the requirements of
the level(s) below. The criteria for each maturity level provided
to the participants are depicted in Table 1. They range from
level 0 where the control is not implemented to level 5 where the
respective control is eectively implemented, monitored, con-
trolled, and continuously improved. For the remainder of this
paper, we refer to maturity models as security maturity models.
2.2. Related Work
Although maturity models are widely used in industry and
are quite common in academia, there are only a few studies
analysing the practitioners’ assessment capabilities.
Zhang and Fever [22] summarise the theoretical values and
weaknesses of the COBIT framework identified by previous re-
searchers. They found that most organisations are mainly in-
terested in the maturity model from the COBIT framework be-
cause it is easy to understand and can be quantified. However,
their work is solely based on existing literature and they state
themselves that it is necessary to collect more input and criti-
cism from practitioners and COBIT experts.
El Emam et al. have investigated the reliability for SPICE
maturity levels for version 1.0 of the SPICE rating scheme on
the basis of process groups in software engineering [23]. In
Table 1: Description of the COBIT 5 Maturity Levels
Level Maturity Levels Description
0–Incomplete The control is not implemented or fails to
achieve its purpose.
1–Performed The implemented control achieves its pro-
cess purpose.
2–Managed The level 1 performed control is now im-
plemented in a managed fashion (planned,
monitored and adjusted) and its work prod-
ucts are appropriately established, con-
trolled and maintained.
3–Established The level 2 managed control is now imple-
mented using a defined process that is ca-
pable of achieving its process outcomes.
4–Predictable The level 3 established control now oper-
ates within defined limits to achieve its pro-
cess outcomes.
5–Optimising The level 4 predictable control is contin-
uously improved to meet relevant current
and projected business goals.
their study, the degree of reliability strongly varied between dif-
ferent processes. For some of them, the maturity levels could
be assessed very reliably (perfect agreement), whereas the as-
sessments for other processes only showed moderate reliability.
The authors explain these higher grades of disagreements pri-
marily by saying that the assessments were conducted by inter-
nal assessors (five) and external assessors (three). This plays a
significant role because internal assessors typically know their
organisation’s processes better, hence a dierent perception.
Another similar study has been conducted by Lee et al. [24].
They also analyse the reliability of SPICE assessments for soft-
ware engineering processes. Their results show a substantial
to excellent reliability in the maturity level assessment. How-
ever, current studies only focus on measuring the reliability but
do not qualify their findings with respect to the validity of the
assessor’s ratings. This is important since even ratings with
high reliability may significantly dier from the theoretical rat-
ing. More importantly, most existing studies do not only deal
with the evaluation of entire software engineering processes but
they also do not address lightweight variants. This means that
the maturity levels are not determined directly but result from
the complex process to assess the implementation status (not
achieved, partially achieved, largely achieved, fully achieved)
of the process attributes defined for each process and level. But
in practice, oftentimes, less complex variants are implemented,
for example, in the VDA-ISA questionnaire, in which the con-
trols’ maturity levels are directly assessed [7]. There are several
other approaches in security management based on the maturity
levels of the ISO/IEC 27002 security controls [9, 10]. To the
best of our knowledge the quality of this kind of maturity level
assessments has never been analysed in the literature.
3. Methodology
To evaluate practitioners’ capabilities to assess the COBIT
maturity levels of security controls and to gain insights into po-
tential challenges and how to address them, an experiment has
been conducted. The experiment was set up as an online sur-
vey in which a hypothetical IT infrastructure, including security
measures and processes, of a small company was presented to
the participants. The description of the scenario was system-
atically constructed to represent predefined maturity levels for
a number of the scenario’s security controls. The participants’
task was then to assess the maturity levels for these controls,
and also to provide a rationale for their decision. These data
build the basis for a quantitative and qualitative analysis on the
quality of their assessments. In addition to this, ex-post inter-
views have been conducted to verify some of the hypotheses
developed during the previous analyses.
3.1. Scenario Design and Scenario Maturity Levels
In the following, it is described how the scenario was con-
structed on the basis of the predefined maturity levels. Fig. 1
visualises the scenario.
3.1.1. Scenario Design
The scenario reflects a fairly generic and realistic situation to
ensure that a high number of participants will understand it and
can better identify themselves with the scenario.
It aims to address a typical situation in the real world. It
describes a hypothetical cloud service provider, called Cloud-
Sec, which is a small company oering various cloud services
(IaaS, PaaS and SaaS) to their customers. From the perspec-
tive of information security, policies, physical and application
security are very important for a cloud service provider. The
scenario description consists of a textual description and its vi-
sual illustration (see Figure 1 in the Appendix). The illustration
shows the physical characteristics of the building and schemat-
ically refers to the respective paragraphs in the text. The text
is structured by high-level security categories like information
security guidelines, access control, and vulnerability manage-
ment. In each paragraph, dierent security measures and their
implementation status are described. We decided to use security
measures instead of the controls itself to derive a more realis-
tic scenario since in a real-world situation practitioners have to
map security measures to the corresponding controls as well.
The starting point to describe the scenario’s security mea-
sures were as set of ISO/IEC 27002 security controls. Since
policies, physical security and application security play a ma-
jor role for cloud service provider controls from this topic area
were selected. However, not all controls could be selected
since, otherwise, the size and the length of the scenario would
go much beyond what a study participant is able to compre-
hend. For each control, we then defined a scenario maturity lev-
els. Next, we used the mapping of security measures from the
BSI IT-Grundschutz to ISO/IEC 27002 controls [25] to iden-
tify corresponding measures. The mapping was developed in
cooperation between ISACAs Information Security (German
Figure 1: Visualisation of the Scenario
Chapter) professional group and the BSI, and thus has a fo-
cus on practitioners. Since the BSI IT-Grundschutz is based on
the ISO/IEC 27002 standard, it is straightforward to derive de-
scriptions for specific measures from the controls. Examples
for three of the controls used in our scenario are shown in Ta-
ble 2. These security controls concern policies (A.5.1), physical
security (A.11.1) and application security (A.12.6).
3.1.2. Predefined Scenario Maturity Levels
To validate and to analyse the quality of the maturity level
assessments of the study participants, it is crucial to know the
predefined maturity levels of the security controls (see Table 3)
and to understand how the scenario has been constructed ac-
The predefined scenario maturity levels (SML) reflect a real-
istic distribution and show a wide range of maturity levels up
to level 4. Level 5, the highest level possible, is very dicult
to achieve in relative terms and is therefore not applied in the
scenario. Most of the predefined scenario maturity levels are
between 0 and 3 since a medium-term goal of many companies
is to achieve at least a control maturity level of 3 (a defined
process) [26, 7].
In the following, three exemplary controls (each from a dif-
ferent control category) are used to demonstrate how the sce-
nario was constructed on the basis of the predefined scenario
maturity levels. For this purpose, the maturity level of each
control was systematically demarcated to the next lower and to
the next higher level by (not) fulfilling the respective criteria
(see Table 4).
For control 5.1.1 (policies for information security) a sce-
nario maturity level of 2 has been defined respecting the criteria
defined in Table 1. To achieve this, 2 (out of 3) IT-Grundschutz
measures for control 5.1.1 (that are most relevant for the sce-
nario) are described to point out that the policy was defined in
a systematic and structured way (see Table 2). Additionally,
essential aspects like the high resource-intensity and the man-
agement support have been emphasised which go beyond stan-
dard level 1 requirements. On the other hand, it is emphasised
that no information security policy has been published before,
indicating that no process has been defined there, as this is the
key characteristic of the next higher maturity level 3.
For control 11.1.1 (physical security perimeter) a maturity
level of 2 has been defined so that it is eectively managed. To
ensure the adequacy of the process all IT-Grundschutz measures
for this control have been described in the scenario (see Table
2), ranging from the creation of security zones to burglary pro-
tection and the server room’s arrangement in the rear part of the
building. The security concept reflects the result of a solid plan-
ning in order to eectively meet the quasi-static requirements.
On the other hand, the description of the control implementa-
tion does not fulfill the requirements for level 3 since it does not
follow a defined process.
Control 12.6.1 (management of technical vulnerabilities) is
described according to maturity level 4. All IT-Grundschutz
measures (see Table 2) and several additional ones are men-
tioned in the description clarify that the level 4 requirements
are met, e. g. that an eective process is in place to measure the
vulnerability exposure and to react accordingly. However, since
a continuous improvement process is missing level 5 cannot be
The other 7 ISO/IEC controls have been described in the
same way to represent the predefined maturity levels.
3.2. Questionnaire Design
The questionnaire comprises four sections (see Appendix B).
First, the practitioners were asked for demographic information
(see Section [A and B]) (such as professional background, cer-
tifications, employer characteristics, experience with ISO/IEC
27002 and maturity levels). Second, the general concepts that
are required to understand the experiment were explained (see
Section [C]). This includes the term security control, the con-
cept of maturity levels (particularly the COBIT maturity lev-
els) as well as a description and an illustration of the scenario
Table 2: Security measures for exemplary controls used in the scenario
Control Measure ID Description of Measure Paragraphs in the scenario description
C 5.1.1 M 2.192 Preparation of a guideline on information security (2) Information security policy
M 2.335 Definition of the security objectives and strategy (1) Security assessment
C 11.1.1 M 1.79 Creation of security zones (3) Access control by gatekeeper,
(6) Working guidelines and (9) Backups
M 1.17 Gatekeeping (3) Access control by gatekeeper
M 1.53 Video surveillance (4) Video surveillance
M 1.19 Burglary protection (5) Protection measures for server room
M 1.13 Arrangement of parts of buildings requiring protection (3) Access control by gatekeeper
M 1.55 Perimeter protection (5) Protection measures for server room
C 12.6.1 M 2.35 Obtaining information on system vulnerabilities (8) Vulnerability management
M 2.273 Prompt application of security-relevant patches and up-
(8) Vulnerability management
Table 3: Scenario maturity levels per control
Control Control Description Scenario Maturity Level Qualitative Feedback
C 5.1.1 Policies for information security 2 - Managed Question F1
C 5.1.2 Review of the policies for information security 0 - Incomplete
C 11.1.1 Physical security perimeter 2 - Managed Question H1
C 11.1.2 Physical entry controls 3 - Established
C 11.1.3 Securing oces, rooms and facilities 2 - Managed
C 11.1.4 Protecting against external and environmental threats 3 - Established
C 11.1.5 Working in secure areas 0 - Incomplete
C 11.1.6 Delivery and loading areas 3 - Established
C 12.6.1 Management of technical vulnerabilities 4 - Predictable Question J1
C 12.6.2 Restrictions on software installation 0 - Incomplete
(see Section [D]). In the third section, the practitioners were
requested to assess the maturity levels for a number of secu-
rity controls described in the scenario (see Section [E, G and
I]). Additionally, they were asked to explain for three controls
how the next maturity level could be achieved (see Section [F,
H and J]). This enables qualitative analysis whether practition-
ers might have misunderstood, for instance, the scenario or the
given maturity levels. A control question has also been inte-
grated into this section, requesting the practitioners to set the
maturity level of the (non-existing) security control 12.6.0. to
5. Finally, the practitioners were supposed to indicate how chal-
lenging they perceived the task, and the reasons for these chal-
lenges (see Section [K, L and M]). This helps to get a better
understanding of the type of additional support that might be
3.3. Validation of Scenario and Questionnaire
The scenario and the questionnaire have been validated in
two rounds of pre-tests with experienced information security
The first round was conducted to gather initial feedback and
to identify neuralgic points. The survey has been given to three
experts, followed by a group discussion on the general ques-
tionnaire design and the scenario description. As a result of
their feedback, the textual scenario description has been struc-
tured in a more comprehensible form and has been extended by
a graphical illustration of the scenario. In addition, several parts
of the questionnaire have been changed. A control question has
been added, for instance, and the maturity level definitions and
a link to the scenario description have then been presented on
each page where a control’s maturity level has to be assessed,
and not only in the questionnaire’s introduction. Altogether, the
experts confirmed the practical relevance of the research ques-
To validate the modified version, the second round of pre-
tests has been conducted in which the survey has been sent to
three chief information security ocers from large organisa-
tions. Afterwards, the results have been discussed with each
of them in 20- to 30-minute phone interviews. As a result, the
description of some controls has been reworked to reduce ambi-
guities. Furthermore, three additional feedback questions have
been added to better understand the practitioners’ rationale be-
hind their maturity level assessments.
3.4. Ethical Considerations
We have considered potential ethical issues of the study by
evaluating an extensive check-list provided by the ethics board
of the authors’ university. This check-list qualifies our study
as exempt from an ethics review. However, in order to inform
participants about our data collection process, we provided in-
formation about their right to information and deletion of their
personal data, and that they can revoke their consent at any time.
Table 4: Demarcation of security maturity levels
Control SMLDemarcation to the next lower level Demarcation to the next higher level
C 5.1.1 2 Systematic realisation accompanied by mgmt. support (2) No defined process (2)
C 5.1.2 0 No future review of the policy planned (2)
C 11.1.1 2 Eectively managed to meet the requirements (3, 4, 5) No defined process (3, 4, 5)
C 11.1.2 3 Eective process defined to secure the physical entry (3, 5) No quantification (3, 5)
C 11.1.3 2 Eectively managed to meet the requirements (5) No defined process (5))
C 11.1.4 3 Eective process defined to ensure compliance with regula-
tions (7)
No quantification (7)
C 11.1.5 0 Relevance of the control is not recognised (6)
C 11.1.6 3 Eective process defined for a secure delivery (3) No quantification (3)
C 12.6.1 4 Eective and quantified process defined for vulnerability
management (8)
No continuous improvement (8)
C 12.6.2 0 Relevance of the control is not recognised (6)
Scenario Maturity Level Numbers in brackets correspond to paragraphs of the scenario description in Appendix A
Furthermore, we stated that all answers are anonymous (e. g. no
saving of IP addresses), that all answers are stored on a German
university server and that by participating in the survey, partici-
pants agree that their answers are used exclusively for scientific
purposes within the scope of the presented research project. We
provided an open-text-field for feedback and a researcher’s e-
mail address for further questions and requests at the end of the
3.5. Data Collection and Data Sanitisation
The data collection strategy is twofold: the primary data
source is an online survey. In this survey the participants were
also asked about their willingness to be available for interviews.
These ex-post interviews form the second part of the data col-
lection. They help to better understand the participants’ ratio-
nale and to explain why they have performed well or badly.
3.5.1. Survey
The survey was distributed to security practitioners from var-
ious organisations between 15 April and 23 September 2019.
The online survey, created using the survey software LimeSur-
vey (version 2.63.1), was sent to (chief) information security
ocers but also to a number of professional mailing lists 1. Ad-
ditionally, the survey was also distributed in selected profes-
sional forums 2.
In total, 76 practitioners opened the questionnaire. 56 of
these data records fulfilled the quality criteria. If the control
question has not been answered correctly the data records have
been excluded from further analysis. Incomplete data records
1GI-SECMGT (security management working group of the German Infor-
matics Society),UP KRITIS BAKs (sector-specific working groups for critical
infrastructure protection), ERFA CyberSecurityAllianz (experience exchange
group of the cybsersecurity alliance), Teletrust, CAST (competence centre for
applied security technology), DCSO (german cybersecurity organisation), lo-
cal chapter of the (ISC)2(international information system security certifica-
tion consortium), newsletters (BSI alliance for cybersecurity (ACS), ISACA
(information systems audit and control association) working group Information
security and IT risk management)
2LinkedIn, XING and COBIT website
(e. g. with incomplete maturity level assessments) have been ex-
cluded. A few participants have intentionally de-anonymised
themselves (via the feedback form or via e-mail) and have noti-
fied the study conductors that they have not seriously answered
all questions. The de-anonymisation was possible since they
revealed the textual answers they gave. The corresponding data
records have been excluded as well.
3.5.2. Ex-Post Interviews
To further substantiate the results from the survey, ex-post in-
terviews were subsequently conducted with six participants of
the survey. The participants were interviewed on the basis of a
semi-structured interview guideline that was adapted after each
interview. The latest version is given in Appendix C. It consists
of two core aspects: the challenges of security maturity level as-
sessments and the kind of support that could be given in order to
face these challenges. The interviews were conducted between
14 May and 8 June 2020 ex-post interviews. The average dura-
tion was about 20-30 minutes, including an introduction to the
topic (a small recap), the main part with the two core topics and
a final question. Due to the Corona pandemic, these interviews
were conducted virtually (web conferences or telephone). Each
interview has been performed by two interviewers. One inter-
viewer conducted the interview and the second interviewer was
present from a quality assurance point of view., e g. to reduce
the risk of miscommunication between the interviewee and the
interviewer. The interview was recorded with the approval of
the participant. Based on the recordings a transcript was pro-
duced which was reviewed by both interviewers.
3.6. Demographics
The following two sections explain the demographic infor-
mation of the survey and interview participants.
3.6.1. Survey
We asked for the participants’ capabilities and skills. More
than half of them (55%) have been working in IT security for
more than 10 years. A little bit more than a quarter have 1-
5 years experience in the field of IT security (see Figure 2).
Figure 2: Participants’ IT security experience
Figure 3: Size of the participants’ organisations
Most participants (86%) indicated to be familiar with the basic
idea of maturity models. 58% stated they are familiar with CO-
BIT maturity levels, although only 40% indicated being famil-
iar with the rating scale for process attributes used in COBIT.
53% have more than 5 years of experience with the ISO/IEC
27002 controls. The certifications obtained by the participants
draw a similar picture. 46% have obtained an ISO/IEC 27002
certification, followed by 35% who have a CISM or a CISA
certification, 25% with a general ISMS certification, 23% with
a CISSP certification (multiple selections were possible).
It is noticeable that almost than two thirds (64%) of the par-
ticipants work in companies with more than 1,000 employees
regarding the size of the organisations (in terms of the number
of employees) where the participants work. 39% of the prac-
titioners work in companies with more than 10,000 employees
(see Figure 3). But still, only 19% of the companies have been
certified against ISO/IEC 27002. It is also worth mentioning
that 25% of the organisations are critical infrastructures.
3.6.2. Ex-Post Interviews
All six interviewees have more than 10 years of work experi-
ence, two have even more than 20 years of experience. Half of
them work in large companies with more than 5,000 employ-
ees. In addition, all of them are experienced in working with
ISO/IEC 27002 controls and have - at least basic - knowledge
of maturity models. Most of them also have practical experi-
ence with maturity models.
3.7. Coding the Qualitative Data
For the qualitative evaluation, we coded the textual answers
of the participants. This coding was done for two dierent types
of qualitative data, for responses from participants in the survey
and for responses from participants in a follow-up interview.
3.7.1. Coding the Practitioners’ Rationale
The practitioners have been asked to indicate the security
measures needed to reach the next maturity level in order to bet-
ter understand the practitioners’ rationale behind their maturity
level assessments (see Question F1). To keep the practition-
ers’ eort reasonable, we have only asked for the three security
controls 5.1.1, 11.1.1, and 12.6.1 (see Table 3).
The practitioners’ open-ended responses were coded using
an iterative process to identify recurring themes. To build an
initial set of codes, two coders disjointly coded a random sam-
ple of the practitioners’ responses, discussed the resulting codes
and consolidated them. Then, the two coders re-coded all an-
swers to the open-ended questions.
The suggested security measures to achieve the next maturity
levels have been coded using the codes described down below.
The codes were only assigned to obviously wrong answers.
Scenario Misinterpreted This code indicates that the practi-
tioner has not considered any detail described in the sce-
nario. For example, when the practitioner suggested the
implementation of a specific security measure to achieve
the next maturity level, although it has been also men-
tioned as already existing in the scenario description.
Control Misinterpreted This code describes statements
where the practitioner has commented on another control.
The latter one can happen, for instance, with related
security controls. However, each security control must
be assessed independently. A statement is labelled with
this code if the suggested measures are not related to the
respective control.
Security Measures Exaggerated Statements suggesting
exaggerated security measures that clearly do not ad-
dress the scenario’s scope, are labelled with this code.
Borderline cases have not been labelled. Practitioners
that do not fully get into the scenario might assess the
controls with a dierent mindset. This is of relevance
because, for instance, large companies might require more
rigorous measures than the small company described in
the scenario.
Furthermore, the practitioners also commented on general
challenges with maturity level assessments. This information
serves a broader understanding of the challenges for the practi-
tioners. Several answers did not provide any useful information
or did not identify any challenge, so they were not considered in
the analysis (e. g. I don’t have time to write this down). The
remaining answers were categorised using the codes below.
Control Dependencies This codes reflect challenges with re-
gard to the dependencies between controls. It is assigned
when controls are not assessed independently due to con-
trol dependencies.
Dierentiation Between Maturity Levels Diculties with
maturity level assessments can also arise from challenges
to dierentiate the maturity levels due to unclear defini-
tions. This code has been assigned when the maturity
level definitions caused a deviation from the predefined
scenario maturity levels.
Scope for Interpretation Another challenge is the degree of
subjectivity inherent for qualitative assessments. This
code has been assigned in cases of general critics regard-
ing the subjectivity of maturity level assessments.
Mapping Controls to Processes This code was assigned for
diculties to map the given security controls to actual pro-
Lack of Skills A lack of skills can mean that practitioners do
not have enough background knowledge to evaluate con-
trols with confidence. This code has been assigned for an-
swers that explicitly refer to a lack of skills.
3.7.2. Coding the Ex-Post Interviews
The transcribed answers of the participants were provided
with codes following an iterative process. Two coders carried
out the coding independently of each other. The individual re-
sults of the coding were then discussed and consolidated. The
codes for the ex-post interviews have been divided into three
code categories and were described as follows.
Reasons for Diculty in Maturity Level Assessment This
code category describes the possible reasons that may
have led to the survey participants’ having diculty with
maturity assessment in the scenario. The participants,
for example, could have a tendency to be less critical in
self-evaluation or stricter in external evaluations. This
can be challenging for the assessment. The code category
was divided into four sub-categories (see Section 4.2.2):
dierences in internal and external evaluations, distinction
in degrees of maturity, the dierence between evaluators
and not all controls represent processes.
Reasons for Suggesting Exaggerated Measures This code
category describes possible indications and reasons
that may have led to the suggestion of exaggerated
measures. The code category was divided into two
sub-categories (see Section 4.2.4): in and neglecting
economic considerations.
Support for Maturity Levels Assessments This code cate-
gory indicates whether the participants can be supported
in their maturity assessment. In order to guarantee this,
various supports are conceivable, e.g. a more detailed de-
scription of the maturity levels or suitable examples. The
code category was divided into six sub-categories (see
Section 5.3): discuss maturity assessments, more detailed
description of maturity not necessary, provide examples,
use a catalogue of measures, orientation towards standards
and training courses for maturity assessments.
3.8. Statistical Analysis
We first test whether the variable in question, the linear de-
viation of practitioners’ assessments to the scenario maturity
levels, is normally distributed. This is required to decide which
statistical test is needed for the group comparisons. Thus, we
conduct the Shapiro-Wilk test for normality and find that the
variable is normally distributed. Consequently, we apply a
standard t-test to assess whether there are statistically signifi-
cant dierences between groups of practitioners (with regard to
their professional experiences and the certificates they have ob-
tained). A sample size of 30 or more is a rule of thumb for sam-
ple sizes to apply parametric statistical tests like a t-test. Since
we are close to that threshold for most group comparisons, we
chose the t-test, while considering that the sample size might
limit the validity of our results. We also use standard t-tests and
Spearman’s rank correlation to analyse the relationship between
practitioners’ correct assessments and their self-perception.
4. General Results of the Survey and the Interviews
The following section presents the quantitative results of the
practitioners’ maturity level assessments in Section 4.1 and the
results of the qualitative analysis in Section 4.2 which com-
prises the textual statements provided in the survey and the
statements given in the interviews.
4.1. Practitioners’ Maturity Level Assessments
Figure 4a illustrates the assessment results for each control
in the form of boxplots. The red circle shows the median of
the practitioner assessments. The upper and lower hinges of the
boxes show the interquartile range (from 25th percentile to 75th
percentile). The adjacent values are illustrated by the whiskers
which indicate the minimum and maximum of the practitioner
assessments. Furthermore, we added the scenario maturity lev-
els to the graph (indicated by green rectangles) in order to en-
able a fast comparison between the assessments of the practi-
tioners and the predefined scenario maturity levels. We ordered
the controls in ascending order following the scenario maturity
The boxplots show that the interquartile ranges are relatively
large for all controls except for 12.6.2 and 12.6.1 indicating a
large variance in the assessments of the practitioners. A similar
picture is drawn by the large range between the lower and the
upper whiskers which ranges for many controls from level 0 to
(a) Boxplots illustrating the practitioner assessment for each control; red
circle =median of practitioners’ answers ; green rectangles =scenario
maturity levels
(b) Spider diagram showing the median and mean practitioner assessments
compared to the scenario maturity levels
Figure 4: Practitioners’ assessments compared to the scenario matu-
rity levels
However, for seven (out of ten) controls the median of the
assessments still equals the scenario maturity level. For the
remaining controls there is a deviation of one to the scenario
maturity level.
To visualise the results in a more detailed manner, we created
a spider graph plotting the scenario maturity levels, the median
and the mean of the practitioner assessments (ordered by the
scenario maturity level). The diagram is depicted in Figure 4b.
Comparing the average of the practitioners’ assessments with
the scenario maturity levels controls, three controls had a devia-
tion greater than 1 (5.1.2, 11.1.4 and 11.1.5), most of the others
had a deviation of less than 0.5.
Examining the median of the practitioner responses, the con-
trols 5.1.2 11.1.4 and 11.1.6 had a deviation of one. The median
values for all the other controls have been rated similarly to the
scenario maturity levels, so their deviation is zero.
4.2. Practitioners’ Qualitative Feedback
The quantitative results described in the previous section in-
dicate that many practitioners struggled with the maturity level
assessments. To better understand the reason why this is the
case an extensive qualitative analysis has been conducted. The
results are presented in the following paragraphs. They are
based on statements given by the practitioners in both the sur-
vey and the ex-post interviews. The qualitative analysis investi-
gates questions like why they have assessed the maturity levels
as they did, how confident the practitioners have been with the
quality of their assessments, or what have been the self-reported
and the actual challenges for the practitioners.
4.2.1. Feedback on Improving the Maturity Levels
To understand why the practitioners have assessed the ma-
turity levels as they did, they have been asked in the survey to
explain for three controls (5.1.1, 11.1.1 and 12.6.1) by which
additional measures the next maturity level could be achieved
(see Question [F1, H1 and J1]).
With 56 participants explaining the possible improvements
for three controls a maximum number of 168 explanations was
possible. However, in many cases the practitioners have already
assigned the highest maturity level of 5 so no answer could be
given. Besides that, several other answers had to be excluded
from the analysis since no meaningful evaluation was possi-
ble (e. g. To describe this I would need too much time). The
remaining 66 answers provide insights about the potential dif-
ficulties in assessing the scenario’s maturity level. To enable a
systematic qualitative analysis, we have coded the answers (see
Section 3.7).
Control 5.1.1. 28 answers refer to control 5.1.1 on policies for
information security (see Question [F1]):
Scenario misinterpreted (19): the practitioners suggested,
for instance, a regular confirmation of knowledge from
the employeesas a further measure to achieve the next
higher level of maturity which indicates that the scenario
was misinterpreted because this concrete issue was already
addressed in the scenario description (in paragraph 7), “an
employee is obliged to check compliance with the regula-
tions on a regular basis during the year. Another example
demanded that the relevance [of the policy] must be com-
municated more clearlyto achieve the next higher ma-
turity level, although it has been explicitly mentioned that
the policy has not only be published but that the employees
have also been committed to it.
Control misinterpreted (9): many answers refer to con-
trols with a very dierent focus, i. e., they ask for regular
trainings of the employees(defined in 7.2.2), for guide-
lines for working in the server room(defined in 11.1.5),
or that the access to the server room and installation of
software must be defined and verified by a change man-
agement process(defined in 11.1.3 , 12.6.2 and 12.1.2).
However, other answers refer to the much more related
control 5.1.2 by demanding that the information security
policy must be continuously checked.
Control 11.1.1. 20 answers refer to control 11.1.1 on the phys-
ical security perimeter (see Question [H1]):
Security measures exaggerated (13): an example for an ex-
aggerated security measure mentioned several times is the
implementation of a turnstile for the server room to only
allow one person to pass at a time. Other than for larger
organisations, this is not feasible for the small company
described in the scenario.
Scenario misinterpreted (5): a few comments addressed
the video surveillance and requested, for instance, a video
surveillance of the access area or a continuous video
surveillance. However, according to the description, the
entire company premises are permanently monitored. The
same holds for the suggested security measure to define
and to apply security perimeterswhich is explicitly de-
scribed in the scenario.
Control misinterpreted (2): two answers from the practi-
tioners’ indicate a misinterpretation of the control. One
refers to the definition of working guidelines (which is de-
fined in 11.1.5) and the other one refers to the implemen-
tation of an information security risk management(which
is defined in the non-normative ISMS requirements)
Control 12.6.1. 18 answers refer to control 12.6.1 on the man-
agement of technical vulnerabilities (see Question [J1]):
Control misinterpreted (13): almost all the misinterpre-
tations with regard to control 12.6.1 concern the policy
for new software installations which directly covered in
control 12.6.2 (“Restrictions on software installation”) and
12.5.1 (“Installation of software on operational systems”).
Security measures exaggerated (5): an exemplary exag-
gerated measure is the automatic analysis and prediction
of neuralgic points by artificial intelligencewhich goes
beyond what maturity level 5 strictly requires.
A similar picture is drawn by the participant’s self-reported
challenges. The participants have been asked in the survey for
the reason of their uncertainty when evaluating the controls (see
Question [M1] in the questionnaire): Given three answer op-
tions, 56% of them have stated that the COBIT standard was
unclear, 44% had diculties with the scenario, and for 26% the
controls were not clear enough.
Summary. The answers of the practitioners show that dierent
challenges exist for each of the controls. Some challenges did
not appear for certain controls but every control was accompa-
nied with at least two of the challenges. A misinterpretation
of either the scenario or a control happened in 49 cases. Be-
sides that, the suggestion of exaggerated measures was a cen-
tral pattern. These findings are supported by the participant’s
self-reported challenges.
4.2.2. Reasons for Suggesting Exaggerated Measures
The results of the previous section show that one root cause
leading to dierent maturity levels lies in the suggestion of ex-
aggerated measures. For this reason, the study participants were
asked in the interviews for possible explanations for suggesting
such measures. The answers were categorised using the follow-
ing code categories:
Individual background (6): the largest number of explana-
tions argue that dierent individual backgrounds of partic-
ipants and how they are professionally socialisedfrom
their experience in a server farm or a data centre oper-
ationor in the critical infrastructure sector like in in-
surance companies or energy providers [might have the
eect] that they are a little stricter than others from less
regulated sectors. They might have regulatory require-
ments that we, for example, do not even know about or
do not have in this form. Accordingly, they might tend
to (unconsciously) expect more sophisticated measures
when you take [such highly regulated] companies and in-
frastructures that you know well as a reference.
Besides that, also the size of organisations someone has
worked for might have an eect since in a larger com-
pany, however, I naturally need higher requirementsbe-
cause, for instance, I do not know all the employees
Neglecting economic considerations (5): another explana-
tion deals with the economic characteristics regarding the
eciency of measures. When asked about the turnstile,
it was stated that in smaller organisations one would typi-
cally implement another equivalent measure that requires
fewer resources but is equally eective.Another inter-
viewee explained that in such a survey situation when
there is a larger distance to the target of evaluationand
when you don’t have to look at it economically [as in your
own company] you can fulfil your “dreams”. However,
one interviewee, who works as an auditor, argued that ef-
ficiency is not subject of my evaluation.
Summary. The answers of the interviewees show that the rea-
sons for suggesting exaggerated measures are twofold: they can
be explained by someone’s individual background which can
make it dicult to fully get into the described scenario of the
study. Apart from that, the individual viewpoint can also in-
fluence whether the cost eciency of measures is taken into
account in the assessments for the study similarly as one would
expect it in case of internal real-world assessments.
4.2.3. Feedback on Challenges with Maturity Level Assess-
The practitioners have also been asked in the survey for gen-
eral challenges with assessing security controls using the CO-
BIT maturity levels (see Question [L1]): these self-reported
challenges complement the picture drawn in the previous sec-
tion. The answers were grouped into four categories:
The scope for interpretation (8): the scope for interpreta-
tion was also perceived as challenge: Like all standards,
the COBIT maturity level is partly subject to subjective
assessments and oers a lot of room for interpretation.
This would hold for the subjective perception depending
on organisations /organisational areabut also for the
perspectives of dierent stakeholders (internal or exter-
Dierentiation between maturity levels (15): to distin-
guish certain maturity levels has been perceived as the
major challenge in assessing the controls’ with COBIT
maturity levels. The practitioners stated that the bor-
ders between maturity levels are often very softly de-
finedand the borders are dicult to distinguishas rea-
sons. The answers also indicate that not all maturity lev-
els are equally challenging to assess. This would hold in
particular for the the dierentiation of the maturity lev-
els 3-5respectively for the the dierentiation in real life
between the maturity levels 2, 3 and 4. Further answers
suggest that diculties can also be caused by mixing up
dierent maturity level models: the COBIT 5 maturity
levels are not congruent with the familiar CMMI levels.
Therefore, some uncertainties..
Control dependencies (3): another self-reported challenge
were the dependencies between the controls which makes
the assessment more dicult.
Mapping Controls to processes (3): a few practitioners ex-
pressed the diculty of mapping controls to the under-
lying processes since some controls would be performed
only once and would not necessarily reflect a process.
Lack of skills (3): the last group subsumes a lack of skills
by the practitioners. This can involve, for instance, the
assignment of the security measures to the individual ma-
turity levelsbut also a lack of knowledge or experience
with the COBIT standard (e. g. lack of familiarity with
the COBIT standard, or only basic understanding of
COBIT maturity levels).
Summary. Over half of the statements (29 out of 56) indicate
that the dierentiation between maturity levels is a predominant
challenge. This especially holds for border cases since there is
not always a sharply defined border. This is where the second
challenge, the scope of interpretation, comes into place which
goes in line with the results in Section 4.2.1 referring to partic-
ipants misinterpreting the scenario or suggesting exaggerated
security measures.
4.2.4. Reasons for Diculty in Maturity Level Assessments
To further investigate the challenges and to complement the
already drawn picture, the interviewees were asked for the chal-
lenges of maturity level assessments (see Question [B1 and B2]
of Appendix C): their statements have been structured into three
sub-categories. The first two categories serve a better under-
standing of the challenge “scope for interpretation” described
before; the last category corresponds to the challenge to distin-
guish between maturity levels.
Dierences in internal and external assessments (9): one
category of explanations deals with the dierences be-
tween internal and external assessments. The majority of
the explanations were about a tendency towards mild as-
sessment in self-assessments and more rigid external as-
sessments. This attitude is explained by some participants
as follows: it is ”easier” to be objective in the evalua-
tion of an external company where you are not directly in-
volvedor mostly people judge themselves less critically
than they would be judged by their peers. One intervie-
wee also highlighted the problem of informal information
channels (as an external person and as a consultant, this
is, of course, a little more dicult, because the informal
information channels, which somehow should be disre-
garded, influence the result). This statement highlights
the degree of tension the evaluators are under.
Not all controls represent processes (3): the next cate-
gory addresses that not all controls are seen as process-
oriented. The participants described this area of tension,
for instance, as follows: I would rather see controls as
triggers of processes, others do not really fall into the
process-oriented approach.
Selectivity for maturity level (4): the interviewees agreed
to the statements of the survey that in reality one is of-
ten exactly on the threshold between two levels of maturity
and the evaluation of these grey areas, these transitions
are always dicult. Besides that, it is often dicult to
separate personal opinion from hard requirements.
Summary. Major challenges in the assessment of maturity lev-
els are seen in particular in the subjective bias in internal as-
sessments (in contrast to more objective external assessments).
Further diculties would arise in the dierentiation of matu-
rity levels, especially at the gray areas between two levels, as
well as the assessment of controls that are not entirely process-
5. Analysis of Individual Assessment Capabilities
The previous section demonstrates that the practitioners as a
whole had only a mediocre performance in assessing maturity
levels of security controls. The results showed disperse and
deviant practitioner ratings for some of the controls.
While we already have gained insights on why this might be
the case, in this section, we aim to investigate if there are cer-
tain subgroups that performed significantly better than others.
This way, we aim to identify general challenges of the assess-
ment, areas or groups which are in particular need of further
support, and professional characteristics (e. g. work experience,
academic background, certificates) which have an eect on the
assessment capability. With these insights, we aim to get a bet-
ter understanding of how to improve the quality of maturity
level assessments.
For this purpose, we use a mixed-methods approach. First,
the practitioners’ maturity level assessments are analysed quan-
titatively, followed by a qualitative analysis of textual state-
ments that the participants have entered in the survey. Finally,
ex-post interviews have been performed with some of the par-
ticipants to better understand their rationales and their chal-
5.1. Quantitative Analysis of the Practitioners’ Assessments
The boxplots in Figure 5 show the deviation ƒweper practi-
tioner, sorted by average deviation from the scenario maturity
level. The red points represent the median deviation. It illus-
trates that while for many practitioners the median deviation
for all their assessments is zero or close to it, that’s mostly be-
cause they had roughly the same number of over- and under-
assessments. Not surprisingly, the range of deviations for each
practitioner increases from the left to the right.
Figure 5: Boxplots illustrating each practitioner’s deviation in the as-
It has also been examined whether the participants were ei-
ther too strict or too mild in their assessments. The t-test did
not show any significant results in this respect.
To measure the practitioners’ performance, we had a look at
both metrics, the average linear deviation from the scenario ma-
turity levels and the number of matches with scenario maturity
levels. The histogram depicted in Figure 6a presents the distri-
bution of the practitioners’ performance based on the linear de-
viation from the scenario maturity levels. The linear deviation
ranges from from 0.3 to 1.9 (with an average of 1.07). Figure 6b
shows the number of matches with the scenario maturity levels.
It demonstrates that the best practitioners have 7-8 (out of 10
possible) matches and that there are substantially more practi-
tioners that struggled with the majority of the assessments.
Not surprisingly, a Spearman’s rank correlation revealed a ρ
of 0.79 (with the p-value <4.2·1013) which means that with
high probability there is a strong correlation between the aver-
age linear deviation from the scenario maturity levels and the
number of practitioners’ matches with the scenario’s maturity
levels. Thus, for our further analysis, we focused on the aver-
age linear deviation from the scenario maturity levels.
Table 5: Analysis of the professional characteristics for the top and
bottom 25% practitioners
Professional Characteristics Number of Occur. for
25th Perc. 75th Perc.
Longtime work exp. 11 (79%) 5 (36%)
Longtime ISO/IEC 27002 exp. 7 (50%) 3 (21%)
CMM/CMMI/SSE-CMM exp. 9 (64%) 4 (28%)
CISM/CISA certificate 7 (50%) 2 (14%)
IT-Grundschutz certificate 5 (35%) 1 ( 7%)
ISMS certificate 9 (64%) 0 ( 0%)
ISO/IEC 27001 certificate 10 (71%) 4 (28%)
Without certificate 1 ( 7%) 4 (28%)
Table 6: T-tests analysing dierences between certain groups for the
deviation of the practitioners’ assessments and the scenario maturity
Independent Variables Group Size t-value
yes no
Longtime work exp. 18 38 n.s.
Longtime ISO/IEC 27002 exp. 16 40 n.s.
CMM/CMMI/SSE-CMM exp. 26 30 n.s.
CISM/CISA certificate 20 36 2.1056
IT-Grundschutz certificate 10 46 2.1482
ISMS certificate 14 42 3.4833∗∗
ISO/IEC 27001 certificate 26 30 2.6762∗∗
Without certificate 12 44 n.s.
* and ** asterisks indicate statistical significance at 5%-level and 1%-level
In order to identify the decisive characteristics that aect
the practitioners’ assessment capabilities, we first had a look
at the 25th and the 75th percentile (14 practitioners each) and
have compared their professional characteristics. As shown
in Table 5 large dierences become evident for eight at-
tributes: longtime work experience and longtime experience
with ISO/IEC 27002 (both with more than 10 years), experi-
ence with the maturity models CMM, CMMI or SSE-CMM,
and whether the participant has obtained a specific security cer-
tificate (CISM/CISA, IT-Grundschutz, ISMS, ISO/IEC 27001)
or not. For example, 9 out of the top 14 (25th percentile) have
obtained an ISMS certification, whereas none of the bottom 14
(75th percentile) has such certification.
In the following, the eects of these attributes are statisti-
cally analysed. Table 6 provides the results of t-tests assessing
whether there are statistically significant dierences between
these groups of practitioners in their average deviation per con-
trol. The overall group size is 56 (N=56).
The t-tests indicate that having obtained any of the listed
security certificates has a positive eect on the assessment
capabilities. Practitioners with such certifications show a
lower deviation to the scenario maturity levels and perform
significantly better than others. This especially holds for
ISMS and ISO/IEC 27001 certifications with a strong eect at
1%-level , and IT-Grundschutz certifications and CISM/CISA
at 5%-level. For the other variables there is no evidence for
significant eects.
(a) Histogram for the linear deviation from the scenario maturity levels (b) Histogram for the matches with the scenario maturity levels
Figure 6: Practitioner assessments compared to the scenario maturity level
0 1 2 3 4 5 6 7 8 9 10
Number of assessments perceived as incorrect
Number of incorrect assessments
Figure 7: Perveived self-assessment capabilities
Furthermore, we analysed the practitioners’ self-assessment
capabilities to investigate how the practitioners perceived the
quality of their assessments. The practitioners were asked with
how many (out of the ten) assessments they felt uncertain. (see
Question [K1]). Their self-perception was then compared with
their actual assessments. The results are shown in Figure 7. Re-
sults with the same characteristics were stacked on each other.
Figure 7 suggests that most practitioners were overly confi-
dent, so we validated that impression with a t-test which showed
that there is a statistically significant dierence between their
perceived and their actual performance. The t-test yields a t-
value of 2.574074 with a p-value of <0.3454 ·106which
means that the practitioners have on average correctly assessed
roughly 2.5 controls less than they thought (6.5 incorrect as-
sessments vs. 4 assessments perceived as incorrect).
Moreover, the Spearman’s rank correlation coecient has
been calculated. There is no significant correlation between the
perceived and the real assessment capabilities for the practition-
ers as a whole. This is dierent for both professional character-
istics “longtime work experience”, “longtime ISO/IEC 27002
experience” and “CMM/CMMI/SSE-CMM experience”. For
these characteristics a weak (negative) correlation could be
found (see Table 7). So for any of these practitioner groups,
more confident practitioners tend to perform worse than other
practitioners. We could neither find a similar nor an opposite
eect for the dierent groups of certificate holders.
Table 7: Spearman’s rank correlation indicating statistically signifi-
cant correlations between certain groups for the number of assessments
perceived as incorrect and the actual number of incorrect assessments.
Independent Variables Group Size ρ
Longtime work exp. 18 -0.3911*
Longtime ISO/IEC 27002 exp. 16 -0.5717*
CMM/CMMI/SSE-CMM exp. 26 -0.4981*
CISM/CISA certificate 20 n.s.
IT-Grundschutz certificate 10 n.s.
ISMS certificate 14 n.s.
ISO/IEC 27001 certificate 26 n.s.
Without certificate 12 n.s.
All participants 56 n.s.
* and ** asterisks indicate statistical significance at 5%-level and 1%-level
5.2. Qualitative Analysis of the Practitioners’ Rationale
To get an idea how the dierent challenges in the assessment
(see Section 4.2) were applying to dierent groups of partici-
pants, the occurrence and the distribution of codes have been
analysed with respect to dierent groups of practitioners. We
decided to investigate the four groups with significant dier-
ences in the performance (see Table 6): holder of an ISO/IEC
27001 certificates (26 yes, 30 no), holder of a CISM/CISA cer-
tificate (20 yes, 36 no), holder of an ISMS certificate (14 yes,
42 no) and holder of an IT-Grundschutz certificate (10 yes, 46
no). Additionally, since we chose a small company in our sce-
nario, we assumed that the code ’security measures exagger-
ated’ might be more relevant for participants in a large com-
pany. Therefore, we also analysed this group. For that purpose,
we defined large company by companies with more than 5,000
employees since this divided the participants into two groups of
almost the same size (27 in a small or medium vs. 29 in a large
company). These five groups are now compared with respect to
the occurrence of the codes.
Not all of these five groups split their members and non-
members into equal-sized groups. Therefore, for each group,
we calculated the relative number of codes per group and com-
pared it with the percentage of members in the group as shown
in Fig. 8. The Y-axis represents the proportion of group mem-
bers versus non-members (e. g. 0.52 (29/56) ratio for partici-
pants from large companies), the X-axis describes the propor-
tion of participants for whom a code has been assigned (e. g.
0.94 (15/16) ratio for exaggerated security measures from the
participants of large companies). To foster a better visualisa-
tion, we also added the 45-degree line. Points located above
this line represent more codes in the non-member group while
points below the line represent more codes in the member
Regarding the distribution of the respective groups of the
code ’control misinterpreted’ (see Figure 8a), all groups are rea-
sonable close to an equal distribution of codes for members and
non-members with the group of participants with an ITG cer-
tificate slightly standing out.
Similarly, the distribution of the respective groups of the code
’scenario misinterpreted’ (see Figure 8b) shows no larger devi-
ations and all groups are reasonable close to the equal distribu-
tion of codes for members and non-members.
For the code ’security measures exaggerated’ (see Figure 8c),
two groups stand out. Holders of an ISMS certificate did not
propose any exaggerated security measure. On the other hand
most of the exaggerated security measures were proposed by
participants from large companies.
Overall, the analysis indicates that there are no huge dif-
ferences between the groups regarding the interpretation of
the controls and the scenario. In contrast, exaggerated secu-
rity measures were mostly proposed by participants from large
companies which may have had problems to adapt to the sce-
nario of a somewhat small company.
5.3. Support for Maturity Levels Assessments
The previous results indicate that the assessment of security
maturity levels is hard for most of the practitioners. There-
fore, we investigated in the ex-post interviews how the assessors
could be supported. The key statements have been structured
into the following categories:
Discussion of maturity level assessments (3): this sugges-
tion aims to improve the assessment quality by the dis-
cussion on maturity levels between two or more assessors.
This is underlined by statements from participants like it
makes sense to discuss this with a consultantor to gain
experience in an audit team.
Provide examples (1): another type of supports points out
that examples for the dierent maturity levels would help
the participants (examples would help you to find out
what to look for or what are the typical characteristics
when you want to assess the degree of maturity).
Present “must” and “should” requirements (1): one in-
terviewee referred to the TISAX standards which is a
de-facto standard in information security management in
the German automotive industry. Its VDA-ISA presents
a number of “must” and “should” requirements for each
control (in order to obtain more consistent results, one
would have to add some more information such as the
VDA-ISA or TISAX standards).
Use catalogue of measures (1): another suggestion goes
one step further and asks for more additional information
per control and maturity level: I definitely consider a cat-
alogue of measures per control/maturity level as helpful.
Training courses for maturity level assessments (1): an-
other option could be a specific a training which could help
to improve the assessments quality. Pure trainings for
maturity assessment are not known to meindicates that
the participants are not aware of any corresponding oer
but would find it useful.
Additionally, two practitioners also mentioned that a more
detailed description of the maturity levels would not be help-
ful for them. However, there is still a wide range of ideas that
can be applied in order to support assessors. This can also be
achieved by combining dierent types of support.
6. Discussion
Both the quantitative and the qualitative analysis indicate
that most practitioners struggled with the maturity level assess-
6.1. Quality of Maturity Level Assessments
In this section, we discuss findings related to the quality of
the assessments.
Assessing a Scenario Versus Assessing the Own Company
One possible explanation for the poor performance of many
practitioners is that the scenario was built for another context
(size, industry domain, etc.) than the participants’ company. A
study by El Emam et al. showed that the assessment capability
is higher when practitioners assess their own organisation [23].
So the practitioners might have performed better if they had
evaluated their own company instead of a hypothetical scenario.
This hypothesis is also supported by the finding that some of
the practitioners did not suciently get into the scenario. This
is especially the case for practitioners from larger organisations
(a) Code ’control misinterpreted’ (b) Code ’scenario misinterpreted’ (c) Code ’security measure exaggerated’
Figure 8: Distribution of codes for certain groups
who have suggested exaggerated security measures more fre-
quently (see Figure 8c). Although the size of an organisation
does not determine per se its protection requirements, larger or-
ganisations, in general, can implement and maintain complex
security measures more economically, e.g. through economies
of scale and more financial and personnel resources.
Neglecting Economic Considerations and Wiggle Room
Quite a few practitioners proposed exaggerated measures to im-
prove a certain maturity security level. The reason for that was
partly not considering the preconditions in the scenario (small
company), partly that there is some room for interpretation and
last but not least a disagreement of the consideration of eco-
nomic factors. Building on economic factors and restrictions
is also connected to the previous finding that assessors tend to
assess their own company better. It might also be the reason
for the findings of El Emam et al. [23] who found that internal
assessors typically know their organisation’s processes better
hence a dierent perception.
In contrast to the other interviewees one of them, working as
an external auditor, denied that economic restrictions should be
considered in the assessment. However, it is explicitly stated in
ISO/IEC 27000 that the implementation of controls should take
into account economic considerations like “the cost of imple-
mentation and operation in relation to the risk being reduced,
and remaining proportional to the organisation’s requirements
and constraints”. Also, other related standards emphasise the
need to balance security and costs in the selection and imple-
mentation of security controls, e. g. ISO/IEC 27005 and, in par-
ticular, ISO/IEC 27016 which focuses on economic aspects in
information security management.
Assessors with Certificate Perform Better
It could also be seen that practitioners with security certificates
(especially ISMS and ISO/IEC 27001 certificates) performed
significantly better than others. This indicates that assessing
maturity levels is not a trivial task and requires specific expert
Quality Assurance: Practitioners’ are Overconfident
Regarding the practitioners’ overconfidence, one possible ex-
planation is that they had experience with other maturity mod-
els which lead to the observed overconfidence.
Furthermore, the statistical analyses have shown that practition-
ers with longtime ISO/IEC 27002 experience or with experi-
ence in maturity level models with higher confidence tend to
perform worse than others. This cognitive bias is quite common
in other areas and also known as Dunning-Kruger eect [27].
Mixing Up Maturity Level Models
Some practitioners might have mixed up dierent maturity level
models or dierent versions of the same model that they are
more familiar with. One practitioner has explicitly stated that
it is challenging to work with a new maturity level model other
than the one they are familiar with. To mitigate this possible is-
sue, we explicitly described the COBIT maturity model and its
characteristics in the questionnaire and put the relevant tables
on each page with assessments again. Participants also had the
possibility to open a portable document format file in another
window to always have the descriptions at hand.
6.2. Impact for Real-World Assessments
In this section, we discuss how our findings could be used
to improve the quality of real-world assessments. Since most
practitioners struggled with the maturity level assessments, in
practice there is the need to provide additional support. There
are numerous ways to provide additional support to practition-
ers. This became evident due to the practitioner’s answers in the
survey but also, and more extensively, in the ex-post interviews
where the practitioners have been explicitly asked what kind of
assistance would have helped them (see Question [M1]).
Detailed Definitions of the Security Maturity Level
For example, it could be helpful to provide a more detailed def-
inition for each maturity level, as suggested by some practi-
tioners in the survey. However, the interviewees did not see a
value in more detailed definitions. So probably those descrip-
tions only help less experienced assessors. In addition to hav-
ing only definitions, one could, for instance, better explain them
based on concrete examples.
Support for Connecting Measures and Controls
Some practitioners explicitly referred to the VDA-ISA ques-
tionnaire that comes up with more detailed definitions. It also
defines for each question (that in most cases covers several se-
curity controls according to ISO/IEC 27002) set security mea-
sures that (a) must or (b) should be implemented in order to
fulfil the respective controls. This additional support could also
be extended to a catalogue of security measures that defines
for each control which (exemplary) measures are to be imple-
mented in order to achieve each maturity level. This approach
is defined in the COBIT framework.
Assessment in Teams
Another factor that might inhibit the assessment quality is the
number of practitioners involved in an assessment. In the ex-
periment, the practitioners conducted the assessments indepen-
dently. However, the results might have been better if two
practitioners had discussed and assessed the maturity levels to-
gether, similar to the concept of pair programming in agile soft-
ware development.
Training Courses for Maturity Level Assessments
Besides the already mentioned measures, training courses for
COBIT maturity level assessments could be valuable that pro-
vide sucient time to practice the learned know-how inten-
sively. The interviewees were not aware of any dedicated train-
ing course. One participant of the survey also asked if he could
use the experiment of the survey because of this.
Mapping Controls to Processes
Assessing maturity levels of ISO/IEC 27002 controls is a com-
mon practise in industry and is also supported by standard GRC
tools like risk2value which is used by major companies [8].
However, it was mentioned that not all ISO/IEC 27002 con-
trols map equally well to processes. Some of them would need
to be merged accordingly to be assessed in a process-oriented
manner. A prominent example following this procedure is the
VDA-ISA questionnaire (a de facto standard in the German au-
tomotive industry described earlier), where maturity level as-
sessments of controls are conducted in this way. Alternatively,
the ISO/IEC 27002 controls can be mapped to (COBIT) pro-
cesses as shown by Sheikhpour et al. [28].
6.3. Limitations and Threats to Validity
Besides the already mentioned limitations, i. e. a hypothetical
scenario instead of the practitioners’ organisations, we discuss
further limitations here.
In our experiment, all relevant information was given in the sce-
nario description. In practice, it can be much more complex to
collect all relevant information that is needed to assess the ma-
turity levels, especially for larger organisations. Thus, it is not
clear to which degree our experiment reflects reality. However,
given that one of the key points of our experiment was a the-
oretically sound elaboration of the security maturity levels, we
could not ask the participants for evaluations within their or-
ganisation. Furthermore, this would also have lead to problems
regarding the comparability and confidentiality of the security
maturity levels.
El Emam et al. [23] found in their study that the degree of re-
liability strongly varied between dierent processes. That bears
the risk that our selection of processes/controls is not represen-
tative for all ISO/IEC 27002 controls.
Due to the participants’ time limitations, we could only ask
for the assessment of a limited amount of controls. Having the
full list of controls to assess may have led to less misinterpreta-
tions of controls because the dierences between the mixed-up
controls would have been more obvious if both of them would
have been presented.
A general limitation for this kind of study is the self-selection
bias since the practitioners who decided to participate in the
study might not be a representative sample for all practitioners.
This can lead to biased results.
Another limitation is the number of participants. A higher
number of participants would make it possible to measure
smaller eects, influencing the practitioners’ ability to assess
the maturity levels as well. Besides the total number of partici-
pants, also the high number of participants from larger compa-
nies (22) could also have led to non-representative results.
Apart from that, the results are only valid for Germany since
the study was conducted there. With practitioners from other
countries, who are more or less familiar with the COBIT stan-
dard and with the ISO/IEC 27002 controls, the results might be
7. Conclusion
“Measuring security is hard” as Pfleeger et al. state [29]. This
is particularly true at a high-level organisational level that typ-
ically deals with complex targets of evaluation and has a large
scope. For that purpose, maturity models play an important role
in assessing information security. However, these assessments
are carried out by humans, and thus the ’human factor’ of the
evaluation may lead to possible uncertainties and subjectivity,
e. g. because a pool of evaluators with dierent physical and
mental conditions of the same evaluator on dierent days were
The main goal of our paper was to examine the practitioners’
capabilities to accurately assess the maturity level for security
controls. In contrast to other studies, we did not compare the
level of agreement between the participants but constructed a
scenario where we defined the maturity level for each security
control following the involved standards, i. e. ISO/IEC 27002
and COBIT. We show that many practitioners had a large de-
viation from the predefined maturity levels. The result con-
firms our design decision and the need to compare the practi-
tioners’ assessments to a baseline and not only to each other.
Due to this design decision, we were also able to identify dif-
ferences in the evaluation of practitioners for certain groups.
We showed that practitioners with security certificates had bet-
ter assessment results. Moreover, the practitioners’ perceived
performance seemed not to be related to their actual perfor-
mance and for some groups practitioners (e. g. more than 10
years work experience), we even found a weak inverted corre-
lation, also known as Dunning-Kruger eect [27].
We conclude that some form of assistance for the evaluation
or classification into maturity levels is needed and elaborate
on possible support such as assessment in teams or specialised
documentation. As a consequence, this can be seen as another
change in the profile of and thus the requirements for the ed-
ucation of security managers (cf. Virtanen [30]), where some
dedicated training for this kind of task is needed.
In future work, a systematic analysis of how quality can be
improved concerning the maturity models could reduce the un-
certainty of these assessments. It would also be useful to iden-
tify more individual and organisational characteristics that im-
prove the quality of assessing controls’ maturity levels. Further-
more, it would be interesting to investigate the intra-assessor
reliability, to learn how the assessments of the same assessor
dier when assessing multiple times.
This work was partially supported by the European Union’s
Horizon 2020 research and innovation program (grant agree-
ment number: 830929).
[1] Accenture security, Ninth annual cost of cybercrime study,
2019-Cost- of-Cybercrime- Study- Final.pdf, 2019.
[2] IBM security, Cost of a data breach report 2019, https://www.all-
2019_Cost_of_a_Data_Breach_Report_final.pdf”, 2019.
[3] Council of European Union, General data protection regulation, https:
//”, 2016.
[4] T. DeMarco, Controlling Software Projects: Management, Measurement,
and Estimates: Management, Measurement and Estimation, Pearson Ed-
ucation (US), 1982.
[5] R. B¨
ohme, Security metrics and security investment models, in: Lec-
ture Notes in Computer Science (including subseries Lecture Notes in
Artificial Intelligence and Lecture Notes in Bioinformatics), volume 6434
LNCS, pp. 10–24.
[6] M. Rudolph, R. Schwarz, Security Indicators - A State of the Art Survey
Public Report, FhG IESE VII (2012).
[7] Verband der Automobilindustrie (VDA), Information security as-
security-assessment.html, 2020.
[8] Avedos,, 2020.
[9] C. Schmitz, S. Pape, Lisra: Lightweight security risk assessment for
decision support in information security, Computers & Security 90 (2020)
[10] M. Schmid, S. Pape, A structured comparison of the corporate infor-
mation security maturity level, in: G. Dhillon, F. Karlsson, K. Hedstr¨
A. Z´
uquete (Eds.), ICT Systems Security and Privacy Protection, Springer
International Publishing, Cham, 2019, pp. 223–237.
[11] C. Schmitz, A. Sekulla, S. Pape, V. Pipek, K. Rannenberg, Easing the
burden of security self-assessments, in: N. L. Clarke, S. Furnell (Eds.),
Twelfth International Symposium on Human Aspects of Information Se-
curity & Assurance, HAISA 2018, Dundee, Scotland, UK, August 29-31,
2018, Proceedings, University of Plymouth, 2018, pp. 272–281.
[12] P. Bowen, J. Hash, M. Wilson, C. M. Gutierrez, W. Jerey, Information
Security Handbook: A Guide for Managers, NIST Special Publication
800-100 (2006).
[13] NIST, NIST Releases Version 1.1 of its Popular Cybersecurity Frame-
work, National Institute of Standards and Technology (2018).
[14] Bundesamt f¨
ur Sicherheit in der Informationstechnik, BSI - IT-
Grundschutz, 2019.
[15] P. Kusumah, S. Sutikno, Y. Rosmansyah, Model design of information
security governance assessment with collaborative integration of cobit 5
and itil (case study: Intrac), in: 2014 International Conference on ICT
For Smart Society (ICISS), IEEE, pp. 1–6.
[16] D. Proenc¸ a, J. Borbinha, Maturity models for information systems
- a state of the art, Procedia Computer Science 100 (2016) 1042
1049. International Conference on ENTERprise Information Sys-
tems/International Conference on Project MANagement/International
Conference on Health and Social Care Information Systems and Tech-
nologies, CENTERIS/ProjMAN /HCist 2016.
[17] J. D. Christopher, D. Gonzalez, D. W. White, J. Stevens, J. Grundman,
N. Mehravari, T. Dolan, Cybersecurity capability maturity model (c2m2),
Department of Homeland Security (2014) 1–76.
[18] M. Saleh, Information security maturity model, International Journal of
Computer Science and Security (IJCSS) 5 (2011) 21.
[19] M. B. Chrissis, M. Konrad, S. Shrum, CMMI for Development: Guide-
lines for Process Integration and Product Improvement, Addison-Wesley
Professional, 3rd edition, 2011.
[20] International Organization for Standardization (ISO), International Elec-
trotechnical Commission (IEC), ISO/IEC 15504-5:2012, information
technology process assessment part 5: An exemplar software life
cycle process assessment model, 2012.
[21] Information Systems Audit and Control Association (ISACA), CobiT 5:
A Business Framework for the Governance and Management of Enter-
prise IT, Rolling Meadows, 2012.
[22] S. Zhang, F. H. Le, et al., An examination of the practicability of cobit
framework and the proposal of a cobit-bsc model, Journal of Economics
1 (2013) 5.
[23] K. E. Emam, L. Briand, R. Smith, Assessor agreement in rating spice
processes, Software Process: Improvement and Practice 2 (1996) 291–
[24] H.-y. Lee, H.-W. Jung, C.-S. Chung, J. M. Lee, K. W. Lee, H. J. Jeong,
Analysis of interrater agreement in iso/iec 15504-based software process
assessment, in: Proceedings Second Asia-Pacific Conference on Quality
Software, IEEE, pp. 341–348.
[25] Bundesamt f¨
ur Sicherheit in der Informationstechnik, Zuordnungstabelle
ISO zum modernisierten IT-Grundschutz, 2018.
[26] F. Gleich, Tisax participant handbook, 2019.
[27] J. Kruger, D. Dunning, Unskilled and unaware of it: How diculties
in recognizing one’s own incompetence lead to inflated self-assessments,
Journal of Personality and Social Psychology 77 (2000) 1121–34.
[28] R. Sheikhpour, N. Modiri, An approach to map cobit processes to iso/iec
27001 information security management controls, International Journal
of Security and its Applications 6 (2012) 13–28.
[29] S. Pfleeger, R. Cunningham, Why measuring security is hard, IEEE
Security Privacy 8 (2010) 46–54.
[30] T. Virtanen, Changes in the profile of security managers, in: Security
education and critical infrastructures, Springer, 2003, pp. 41–49.
All sources were last accessed on the 7th of January 2021.
Appendix A. Scenario Description
General: CloudSec is an IT service provider that primarily of-
fers cloud services (IaaS, PaaS and Saas) for other compa-
nies. For this reason physical security is very important.
Fig. 1 schematically shows the scenario described below.
(1) Security assessment: to systematically improve their in-
formation security CloudSec use an ISMS (Information
Security Management System) and evaluate the maturity
levels of their company on a quarterly basis using a tool
specifically provided for this purpose. They are guided by
the security controls of ISO/IEC 27002. In addition, an
internal control system has been implemented which pre-
scribes binding inspection activities at various intervals to
ensure, among other things, that the controls are carried
out properly.
(2) Information security policy: CloudSec’s management re-
cently approved and published its first information security
policy and committed its employees to it. For this purpose,
a security team commissioned by the management first de-
fined the security goals and strategies. As the process took
a long time no further resources will be invested in the re-
view of this policy in the future. However, according to
internal guidelines, compliance with the information secu-
rity policy must be regularly reviewed by the security team
on a random basis.
(3) Access control by gatekeeper: to get access to the build-
ing all persons have to prove their identity by showing
their employee ID card to a gatekeeper. The security team
is also responsible for ensuring that the gatekeeper can see
the updated access rights at any time. These are created
based on an organisation-wide process and are regularly
updated. Since couriers do not have access rights to the
building, all packages are delivered directly to the gate-
keeper, who forwards them to the recipient. Furthermore,
all visits are documented and external persons are also su-
(4) Video surveillance: moreover, the entire company
premises are permanently monitored by a video system
and checked for suspicious activities.
(5) Protection measures for server room: since the server
room has been classified a critical security zone, it is pro-
tected by a burglar-resistant door and by a fingerprint scan-
ner. For security reasons, it is also located in the rear part
of the building.
(6) Working guidelines: however, there are no guidelines for
working in the server room. Due to a lack of information
security risk management, the relevance of this topic is
currently not recognised in the company. The same applies
to guidelines for the installation of new software.
(7) Compliance with regulations: the relevant regulations for
protection against other external threats (e. g. fire protec-
tion regulations) were systematically identified, analysed
and evaluated within the company. In addition, an em-
ployee is obliged to check compliance with the regulations
on a regular basis during the year in order to take action if
(8) Vulnerability management: in order to guarantee the se-
curity of Cloudsec’s own server applications, these are au-
tomatically checked with vulnerability scanners on a daily
basis (identification). In order to continuously increase the
security level, the identified vulnerabilities and their criti-
cality (CVSS value) are stored centrally, checked directly
by the security team (analysis) and if necessary depend-
ing on the risk for the company mitigated immediately
(evaluation and handling). In addition, the security of the
technical infrastructure is continuously documented using
KPIs and checked at least once a year by external experts
using penetration tests.
(9) Backups: all servers are backed up daily. The backups are
stored in a fire compartment other than the server room.
Restore tests are carried out only occasionally.
Appendix B. Questionnaire
Section A: Organisation’s Demographics
A1 What industry sector does your organisation operate in? If
you are an IT consultant please state your industry experi-
ence under “Other”.
Automobile industry
Chemical and raw in-
Services and crafts
E-commerce and
Energy sector
Food sector
Finance and
Healthcare sector
technology and
Media and culture
Metal and electronics
State and administra-
Tourism and gastron-
Transport and trac
Water management
A2 Is your organisation classified as critical infrastructure?
I don’t
A3 Is your organisation certified against the ISO/IEC 27001
I don’t
A4 How many employees does your organisation have?
Less than 100
More than 10,000
3Text box for additional comments
Section B: Person’s Demographics
B1 In which area do you work in your organisation?
IT security or infor-
mation security
General IT
B2 How many years of experience do you have in IT security
or information security?
Less than 1
More than 20
B3 How many years of experience do you have in Critical In-
Less than 1
More than 20
B4 Are you familiar with the basic idea of maturity models?
Yes No
B5 What specific maturity models are you familiar with?4
COBIT 5 maturity
CMM, CMMI oder
B6 How many years of experience do you have with COBIT 5
maturity levels?5
Less than 1
More than 20
B7 How many years of experience do you have with CMM,
Less than 1
More than 20
B8 What other maturity models are you familiar with?5
B9 How many years of experience do you have with other ma-
turity models?5
Less than 1
More than 20
B10 Are you familiar with the rating scale for process at-
tributes used in COBIT (see below)? Evaluation Achieve-
ment of goals:5
Not achieved (N): 0% - 15%
Partially achieved (P): 15% - 50%
Largely achieved (L): 50% - 85%
Fully achieved (F): 85% -100%
4Only asked if familiarity indicated in B4
5Only asked if familiarity indicated in B5
B11 How many years of experience do you have with ISO/IEC
27002 controls?
Less than 1
More than 20
B12 Which certifications have you obtained in the field of in-
formation security so far?
ISO/IEC 27001 (e. g.
ISO/IEC 27001 Lead
IT basic security
B13 What other trainings (trainings/courses) have you com-
pleted in the field of information security?
None Other3
B14 What is your highest level of education?
No vocational quali-
Vocational training
Bachelor’s degree
Master’s degree
B15 Which subject did you study at university?6
Computer Science
IT Security
Section C: Description of Maturity Levels
On this page, the term ’security control’ and the concept of
’maturity levels’, which are used in the following, are intro-
duced. They are important for answering further questions. So
please read the following paragraphs carefully.
Evaluation of security controls based on maturity levels
A security control describes security requirements that
can be fulfilled by performing the corresponding security
measures. So a control is associated with one or more
security measures. A security control can be assessed by
its maturity level. The higher a control’s maturity level,
the higher are the chances that it is performed in a cor-
rect and eective way. For the following questions, the
COBIT maturity levels are used. The levels are defined
in the table below.
Section D: Description of Scenario
This page describes a scenario. The described security con-
trols should be assessed on the following pages. So please read
the text carefully.7
6Only asked if university background indicated in B14
7Here the description of the scenario as sketched in Section Appendix A
Section E: Scenario’s Maturity levels
On the following pages, the COBIT maturity levels for the
scenario presented are to be determined for selected controls.
Additional questions are then asked for three of the maturity
E1 Please assess the COBIT maturity levels for each security
control on the left side according to the described scenario.
The security controls are defined in Section 5 ’information
security policies’, sub-section 5.1 ’management direction
for information security’ of the ISO/IEC 27002.8
0 1 2 3 4 5
5.1.1 - Policies for information security:
a set of policies for information security
should be defined, approved by manage-
ment, published and communicated to em-
ployees and relevant external parties.
5.1.2 - Review of the policies for informa-
tion security: the policies for information
security should be reviewed at planned inter-
vals or if significant changes occur to ensure
their continuing suitability, adequacy and ef-
Section F: 5.1.1 Next Level
You have just assessed the maturity level for the following
control: “5.1.1 - Policies for information security: a set of poli-
cies for information security should be defined, approved by
management, published and communicated to employees and
relevant external parties.
F1 What additional activities would be necessary in your opin-
ion so that CloudSec reach the next level of maturity for
this control?
Section G: Assessment 11.1
G1 Please assess the COBIT maturity levels for the security
controls on the left side according to the described sce-
nario. You can also open the previous descriptions by
clicking on the links.
The security controls are defined in Section 11 ’physical
and environmental security’, sub-section 11.1 ’secure ar-
eas’ of the ISO/IEC 27002. 8
8Tab. 1 included.
11.1.1 - Physical security perimeter: secu-
rity perimeters should be defined and used
to protect areas that contain either sensitive
or critical information and information pro-
cessing facilities.
11.1.2 - Physical entry controls: secure ar-
eas should be protected by appropriate entry
controls to ensure that only authorised per-
sonnel are allowed access.
11.1.3 - Securing oces, rooms and facili-
ties: physical security for oces, rooms and
facilities should be designed and applied.
11.1.4 - Protecting against external and
environmental threats: physical protection
against natural disasters, malicious attack or
accidents should be designed and applied.
11.1.5 - Working in secure areas: procedures
for working in secure areas should be de-
signed and applied.
11.1.6 - Delivery and loading areas: ac-
cess points such as delivery and loading ar-
eas and other points where unauthorised per-
sons could enter the premises should be con-
trolled and, if possible, isolated from infor-
mation processing facilities to avoid unau-
thorised access.
Section H: 11.1.1 Next Level
You have just assessed the maturity level for the following
control: “11.1.1 - Physical security perimeter: security perime-
ters should be defined and used to protect areas that contain
either sensitive or critical information and information process-
ing facilities.
H1 What additional activities would be necessary in your opin-
ion so that CloudSec reach the next level of maturity for
this control?
Section I: Assessment 12.6
I1 Please assess the COBIT maturity levels for the security
controls on the left side according to the described sce-
nario. You can also open the previous descriptions by
clicking on the links.
The security controls are defined in Section 12 ’operations
security’, subsection 12.6 ’technical vulnerability manage-
ment’ of the ISO/IEC 27002. 8
12.6.1 - Management of technical vulner-
abilities: information about technical vul-
nerabilities of information systems being
used should be obtained in a timely fashion,
the organisation’s exposure to such vulner-
abilities evaluated and appropriate measures
taken to address the associated risk.
12.6.0 - This is a control question: please
rate it with maturity level “5 - Optimizing”
to show that you have read the question.
12.6.2 - Restrictions on software installa-
tion: rules governing the installation of soft-
ware by users should be established and im-
Section J: 12.6.1 Next Level You have just assessed the ma-
turity level for the following control: “12.6.1 - Management of
technical vulnerabilities: information about technical vulnera-
bilities of information systems being used should be obtained
in a timely fashion, the organisation’s exposure to such vulner-
abilities evaluated and appropriate measures taken to address
the associated risk.”
J1 What additional activities would be necessary in your opin-
ion so that CloudSec reach the next level of maturity for
this control?
Section K: Confidence
K1 In total, you have assessed the maturity levels for ten secu-
rity controls. For how many of them have you been uncer-
Section L: Challenges
L1 Where do you see possible challenges in the evaluation of
controls based on the COBIT maturity level?
Section M: Uncertainty
M1 What was the reason for your uncertainty when evaluating
the controls?9
The scenario was not quite clear to me.
The controls from ISO/IEC 27002 were not quite clear to
The COBIT maturity levels were not quite clear to me.
Section N: Evaluation Interest
N1 Are you interested in the evaluation of the study?
Yes No
Section O: Email Address You can send us your e-mail ad-
dress under the following link. By entering the e-mail address
separately, no conclusions can be drawn about your answers:
Section P: Contact
P1 Are you available for further contact to discuss your an-
swers (e. g. a short telephone or Skype interview)?
Yes No
9Only asked if uncertainty indicated in K1
Section Q: E-Mail Address
Q1 Please let us know your e-mail address. In order to be able
to discuss your answers afterwards, the given e-mail ad-
dress will be linked to your answers.10
Section R: Feedback
R1 If you have any comments or suggestions about the survey,
you can let us know now:
Appendix C. Interview guide
Section A: Opening Questions
A1 Do you agree that the following conversation is recorded?
A2 Do you agree that the results may be used in scientific pub-
Section B: Assessment of the Challenge of Scenario Evalua-
B1 How did you perceive the maturity assessment in the sce-
B1.a To what extent did you have problems assessing the sce-
B2 How do you generally see the assessment of such hypothet-
ical scenarios?
B2.a How does the assessment of the scenario dier from an
assessment in practice?
B2.b We noticed that some participants had problems assess-
ing the scenario. What factors might have led to this?
B3 How do you see the relevance of the dierent maturity mod-
els compared to each other?
B3.a To what extent do you see dierences in qual-
ity/dissemination of he approaches?
Section C: Possible Assistance for the Scenario Evaluation
C1 What would help you to assess maturity levels?
C1.a Discuss in a team?
C1.b Additional documentation?
C1.c More trainings for assessing maturity levels?
Section D: Closing Questions
D1 Are there any other aspects that are important to you but
have not been addressed by us?
D2 May we contact you for further questions?
10Only asked if availability indicated in P1
... GRC in the context of cybersecurity refers to a systematic approach for managing and mitigating risks, adhering to legal and regulatory requirements, and ensuring sound decision-making and oversight within an organization (Bachlechner et al., 2014;Gale et al., 2022). The three main functions of cybersecurity GRC include governance, risk assessment, and compliance (Bachlechner et al., 2014;Schmitz et al., 2021). In governance, organizations may establish a cybersecurity governance framework to oversee the management of cyber risks, such as the case of a major bank implementing a robust cybersecurity policy (Schmitz et al., 2021). ...
... The three main functions of cybersecurity GRC include governance, risk assessment, and compliance (Bachlechner et al., 2014;Schmitz et al., 2021). In governance, organizations may establish a cybersecurity governance framework to oversee the management of cyber risks, such as the case of a major bank implementing a robust cybersecurity policy (Schmitz et al., 2021). In risk assessment, companies can identify and prioritize risks, like when an energy company conducts a risk assessment to identify vulnerabilities in its critical infrastructure (Wang et al., 2020). ...
Full-text available
This study investigated the potential of Generative Pre-trained Transformers (GPTs), a state-of-the-art large language model, in generating cybersecurity policies to deter and mitigate ransomware attacks that perform data exfiltration. We compared the effectiveness, efficiency, completeness, and ethical compliance of GPT-generated Governance, Risk and Compliance (GRC) policies, with those from established security vendors and government cybersecurity agencies, using game theory, cost-benefit analysis, coverage ratio, and multi-objective optimization. Our findings demonstrated that GPT-generated policies could outperform human-generated policies in certain contexts, particularly when provided with tailored input prompts. To address the limitations of our study, we conducted our analysis with thorough human moderation, tailored input prompts, and the inclusion of legal and ethical experts. Based on these results, we made recommendations for corporates considering the incorporation of GPT in their GRC policy making.
... However, there is no common understanding of ISA, its maturity or its decisive factors. Information security maturity is discussed in previous literature by several researchers (Greene et al., 2022;Lopez-Leyva et al., 2020;Nasir et al., 2019;Schmitz et al., 2021), but ISA maturity requires further study (Fertig et al., 2020). We found only one article and one conference paper in Scopus with the "information security awareness maturity" keyword search. ...
... The literature on ISA maturity models remains scarce. Information security maturity is discussed by several researchers (Greene et al., 2022;Lopez-Leyva et al., 2020;Nasir et al., 2019;Schmitz et al., 2021) but ISA maturity is not. Scopus search using "information security awareness" and "maturity models" resulted in two publications, only one article and one conference paper. ...
Purpose This paper aims to provide a maturity model for information security awareness (MMISA), based on the literature, expert interviews and feedback. In addition to developing the MMISA, the authors investigate the role of the three decisive factors that affect ISA maturity level: risk management mechanism, organizational structure and ISA. Design/methodology/approach The research methodology is a combined one; qualitative and quantitative methods were applied, including surveying the literature, interviews and developing a survey to collect quantitative data about decisive factors that affect ISA maturity level. The authors perform a variance-based partial least squares-structural equation modeling (PLS-SEM) investigation of the relationships between these factors. Findings The investigation of decisive factors of ISA maturity levels revealed that if the authors identify a strong risk assessment mechanism (through a documented methodology and reliable results), the authors can expect a high level of ISA. If there is a well-defined organizational structure with clear responsibilities, this supports the linking of a risk management mechanism with the level of ISA. The connection between organizational structure and ISA maturity level is supported by ISA activities: an increased level of awareness actions strengthens an organizational structure via the best practices learned by the staff. Originality/value The main contribution of the proposed MMISA model is that the model offers controls and audit evidence for maturity levels. Beyond that, the authors distinguish in the MMISA model controls supporting knowledge and controls supporting attitude, emphasizing that this is not enough to know what to do, but the proper attitude is required too. The authors didn't find any other ISA maturity model which has a similar feature. The contribution of the authors' work is that the authors provide a method for solving this complex measurement problem via the MMISA, which also offers direct guidance for the daily practices of organizations.
... BIS standard [52,56,57] Developed by the German government, focusing on the security product; three distinguished wings: BIS 100-1, 2, and 3. ...
... BSI 100-1 describes a few mandatory requirements that are integrated with ISO 27001 to ensure the security of IT infrastructure [56]. The BSI 100-2 standard concentrates on IT security management, step-by-step task implementation for security, and integrating the best practices, while BSI 100-3 is developed for risk analysis based on IT-Grundschutz principles [57]. ...
Full-text available
Cyber attackers are becoming smarter, and at the end of the day, many novel attacks are hosted in the cyber world. Security issues become more complex and critical when the number of services and subscribers increases due to advanced technologies. To ensure a secure environment, cyber professionals suggest reviewing the information security posture of the organization regularly via security experts, which is known as penetration testing. A pen tester executes a penetration test of an organization according to the frameworks and standardization guidelines. Security breaches of the system, loopholes in OS or applications, network vulnerabilities, and breaking data integration scopes are identified, and appropriate remediation is suggested by a pen tester team. The main aim of a penetration process is to fix the vulnerabilities prior to the attack in tangible and intangible resources. Firstly, this review work clarifies the penetration conception and is followed by the taxonomy of penetration domains, frameworks, standards, tools, and scoring methods. It performs a comparison study on the aforementioned items that develops guidelines for selecting an appropriate item set for the penetration process according to the demand of the organization. This paper ends with a constructive observation along with a discussion on recent penetration trends and the scope of future research.
... The notion of maturity is commonly perceived as a gradual progression from lower to higher levels of maturity [18]. The maturity assessment model builds criteria that are dynamic and continuously evolving, whose scope of application is gradually expanding and whose application domains are increasingly becoming broader, including enterprise digital transformation [19], Industry 5.0 [20], and information security [21], etc., and are proving to be instrumental in advancing the continued development of the field. BIM development is likewise a dynamic process for achieving specific goals and capabilities. ...
The phase of operation and maintenance (O&M) is the most time-consuming and cost-intensive phase in the project life cycle. However, the potential benefits of Building Information Modeling (BIM) in this phase have not been fully explored, unlike in the design and construction phases. This is particularly evident in the absence of a comprehensive assessment of its application capabilities. In light of this setting, we develop a BIM maturity assessment model (BIM MAM) for the project's O&M phase. The proposed model comprises of an assessment indicator system that facilitates experts in providing individual assessment results, and a collective opinion aggregation method in a probabilistic context based on a multi-objective optimization model that is employed to generate the ultimate collective assessment results. Our multi-objective optimization model incorporates the influence of human behavior factors on the final results by introducing the fairness concern utility level as an objective. Finally, we take Wuhan Jiangxia Sewage Treatment Plant as a practical case to illustrate the effectiveness and feasibility of the proposed BIM MAM.
... Security governance is an important topic in IT management [4]. IT management in the process of poor information management causes several problems such as information security vulnerabilities that cause threats such as information loss, destruction, theft and interception of important information of agencies or organizations [5]. Continuous improvement steps in IT governance, especially in the process of managing data, must be able to reduce the risk of such threats [6]. ...
Full-text available
Data security is very important in the digital era, this makes it an important issue for companies, organizations and educational institutions. Such as the Open Journal System (OJS) security which is useful for managing online journals. Abdurrab University has utilized the Open Journal System (OJS) to manage published journals. However, the management of the Open Journal System (OJS) security has not been managed properly, such as Abdurrab's Open Journal System (OJS) which has been hacked, there is still a lack of documentation and system security procedures and has never been assessed the level of capability of the information system so that it is not known exactly the level of security management of the Open Journal System (OJS) at Abdurrab University. This study aims to assess the capability level of Abdurrab University's Open Journal System (OJS) using the COBIT 5 domain DSS05 framework and Process Assessment Model (PAM) to assess system capabilities and guide IT management and provide effective and efficient recommendations. The results of the study obtained a questionnaire calculation value of 2.40 with Process Attribute Level domain DSS05 PA 2.1 Performance Management and PA 2.2 Work Product Management until an achievement score of 63.33% was obtained which was included in the Largely Achieved category with achievements at level 2 (Managed Process). The conclusion of this study is that the level of security capabilities of the Open Journal System (OJS) of Abdurrab University needs to be increased to level 3 (Established Process) because it has not been fully met, so that this research can be used as a reference for improvement by Abdurrab University.
... A quantitative assessment, obtained using expert methods, of the potential risk associated with a systematic or onetime violation of medical secrecy plays a critical role in minimizing risks [30,31]. Such indicators help to objectively assess the danger of potential threats, vulnerabilities, size of the damage, and sources of threats and calculate one integral indicator for the entire protection system. ...
Full-text available
The purpose of this investigation is to develop a method for quantitative assessment of the uniqueness of personal medical data (PMD) to improve their protection in medical information systems (MIS). The relevance of the goal is due to the fact that impersonal PMD can form unique combinations that are potentially of interest to intruders and threaten to reveal the patient's identity and medical confidentiality. Existing approaches were analyzed, and a new method for quantifying the degree of uniqueness of PMD was proposed. A weakness in existing approaches is the assumption that an attacker will use exact matching to identify people. The novelty of the method proposed in this paper lies in the fact that it is not limited to this hypothesis, although it has its limitations: it is not applicable to small samples. The developed method for determining the PMD uniqueness coefficient is based on the assumption of a multidimensional distribution of features, characterized by a covariance matrix, and a normal distribution, which provides the most reliable reflection of the existing relationships between features when analyzing large data samples. The results obtained in computational experiments show that efficiency is no worse than that of focus groups of specialized experts. Doi: 10.28991/HIJ-2023-04-01-09 Full Text: PDF
... The extent of the content in catalogues has also been discussed in the literature [24]. Quinn, Ivy, Barrett, Witte and Gardner [25] point out the difficulty in timing between when to perform an activity in relation to how complete the catalogues are. ...
Full-text available
Information Security Risk Management (ISRM) is fundamental in most organisations today. The literature describes ISRM as a complex activity, and one way of addressing this is to enable knowledge reuse in the shape of catalogues. Catalogues in the ISRM domain can contain lists of, e.g. assets, threats and security controls. In this paper, we focus on three aspects of catalogue use. Why we need catalogues, how catalogue granularity is perceived, and how catalogues help novices in practice. As catalogue use is not yet a widespread practice in the ISRM, we have selected a domain where catalogues are a part of the ISRM work. In this case, the Air Traffic Management (ATM) domain uses a methodology that includes catalogues and is built on ISO/IEC 27005. The results are based on data collected from 19 interviews with ATM professionals that are either experts or novices in ISRM. With this paper, we nuance the view on what catalogues can contribute with. For example, consistency, coherency, a starting point and new viewpoints. At the same time, we identify the need to inform about the aim of the catalogues and the limitations that come with catalogue use in order to leverage the use – especially from a novice perspective.KeywordsInformation Security Risk ManagementCataloguesRisk management practice
... It introduces the concept of a maturity model and briefly reviews existing models. 45 Maturity level assessments of information security controls: An empirical analysis of practitioners assessment capabilities [91] The use of maturity models to measure information security is widespread, but the quality of maturity level assessments has not been adequately investigated. This study aimed to analyze the accuracy of security managers' ability to assess maturity levels of security controls using the COBIT maturity levels. ...
Full-text available
In today’s world, private and government organizations are legally obligated to prioritize their information security. They need to provide proof that they are continually improving their cybersecurity compliance. One approach that can help organizations achieve this goal is implementing information security maturity models. These models provide a structured framework for measuring performance and implementing best practices. However, choosing a suitable model can be challenging, requiring cultural, process, and work practice changes. Implementing multiple models can be overwhelming, if possible. This article proposes a prioritization strategy for public institutions that want to improve their information security maturity. We thoroughly analyzed various sources through systematic mapping to identify critical similarities in information security maturity models. Our research led us to create the AIM (Awareness, Infrastructure, and Management) Triad. This triad is a practical guide for organizations to achieve maturity in information security practices.
Full-text available
The continued dependence on information technology applications has led to the adoption of electronic channels and software applications to support