Information Ethics Among College Students – Selected Countries Analysis

ABSTRACT Information ethics (IE) influences how people use and produce information and affects information services, information technology (IT), and professional practices. Lack of IE severely affects individuals, teams, organizations, or the community. This research bridges a gap in understanding IE perceptions and IE-declared behavior among different sociodemographic profiles in the global communicative space. We devised an extended version of the IT ethics survey to measure attitudes toward IE issues. We surveyed a sample of 1,648 college students in seven countries. We identified four clusters, uncovering groups of people with very high (Legalists), high (Moralists), fair (Pragmatists), and low (Anarchists) perceptions of the IE issues. Clusters allow the exploration of patterns in IE perceptions that would remain hidden at an individual level of analysis and can help to predict IE-related behavior. The findings imply that teachers should adapt IE curricula, and managers should customize the policies and interventions to these groups. Furthermore, managers should consider group behavior and account for factors that affect it in the corporate world, such as lower computer knowledge. International managers should be aware that certain groups behave more unethically with IT and are overrepresented in certain countries.


Introduction
Information ethics (IE) has been essential to the global digital economy and information systems research for over three decades.IE influences people's behavior in producing and using information and principles for business information services and professional practices (Lin, Wang, & Yueh, 2022).IE is often used as an approach or framework for analyzing human behavior, decision making, and ethical problems related to information and technology (Stahl, Eden, Jirotka, & Coeckelbergh, 2014).Thus, understanding the influence of perceptions of IE issues on decision making, and behavior is urgently needed due to the unprecedented development and adoption of emerging technologies on a global scale.Individuals' lack of perception of IE issues results in unethical behavior with information technology (IT) that severely affects teams and organizations (Streiner, Burkey, Dahm, Cimino, & Pascal, 2022;Tarafdar, Gupta, & Turel, 2015) and society (Richardson, Petter, & Carter, 2021).
Current students in higher education programs worldwide will enter the workforce and become future employees of organizations.Therefore, it is important to understand how they perceive IE issues and behave in these situations.IE will influence not only their way of using information but also the development of information services (Streiner, Burkey, Dahm, Cimino, & Pascal, 2022) and professional practices (Lin, Wang, & Yueh, 2022).IE directly affects future employees' adherence to information system security policies, a key problem in corporations (Karjalainen, Siponen, Puhakainen, & Sarker, 2020).At the same time, following high ethical standards at work positively impacts job satisfaction and prevents unethical behavior (Kowal & Roztocki, 2015).
This research bridges a gap in understanding IE perceptions and IE-declared behavior among students' profiles in the global communicative space.IE research with student samples typically examines factors influencing perceptions and behavior, such as culture, nationality, and demographics, as we document in a systematic literature review of IE studies in students.This results in a limited understanding of factors moderating IE behavior.There might be other uncovered groups within cohorts of subjects that share common behavioral and attitudinal patterns but remain undiscovered if analyzed with traditional cause-effect approaches such as regression analysis.We address and reveal the value of analyzing IE perceptions and IE-declared behavior among different sociodemographic profiles from a global perspective.Therefore, the main objective of this research is to identify common patterns in perceptions of IE issues and IE-declared behavior.
Our main original contribution is going beyond the individual level of analysis by using clustering analysis from a global perspective.We identified four significant clusters in the dataset that provide insights into the profiles of individuals as to their perceptions of IE dilemmas and declared ethical behavior.The analysis was done on a relatively robust dataset of 1,648 students from seven countries which overcomes the limited sample size of other similar studies in the field.Both academia and practitioners can benefit from our findings.Teachers could consider highlighting certain IE topics in their curricula.At the same time, managers could use the findings to revisit organizational and security policies and standards to mitigate the impacts of IT misuse by employees.Hence, different profiles of target groups should be considered when preparing teaching materials or policies.IT and business researchers might wish to follow up on our study and conduct a survey with large samples addressing a broader range of emerging IE issues such as remote work, mass surveillance, disinformation, and fake news.Finally, the research addresses a global IT issue using data from seven countries, contributing to the development of global IT.
The paper is structured as follows: After this introduction, the second section presents the recent research on IE with a focus on student populations and identifies a lack of studies working with large samples of subjects.The third section describes the underlying theory and presents the research model used in this paper.The fourth section contains descriptive statistics and results of cluster analysis.The fifth section discusses identified clusters of subjects with common patterns of IE perceptions and declared behavior and summarizes implications for theory and practice.The sixth section concludes the work.

Literature Review
The term "computer ethics" (CE) was coined in the 1970s (Bynum, 2008) and followed by the start of broader discourse and research in the 1980s and 1990s (Himma & Tavani, 2008).CE analyzes the nature and social impact of computer technology and formulates and justifies policies for the ethical use of computers (Moor, 1985).However, the advance of social media, e-commerce, and the ubiquitous presence of computers in consumer devices since the 2000s have raised the need to go further than assessing the impacts of a particular technology.The pervasiveness and ubiquity of computing pushed new political and social debates about surveillance, intellectual property of digital content, and the social impacts of social media and big data (Stahl, 2021).
Information ethics (IE) has been conceptualized as a theory that allows the analysis of moral issues and values behind human actions with technology at different levels of abstraction (Floridi, 1999(Floridi, , 2008).Floridi's conceptualization of IE proved powerful in analyzing the current digitalized world (Dodig-Crnkovic, 2012).The IE concept revisits and adds new dimensions to old ethical problems (Niederman & Baker, 2022).Yet, Floridi was criticized by several ethicists for logical inconsistencies in separating informational entities from humans who create and use them (Ess, 2008;Stahl, 2008).
Information systems (IS) as a research field has always been at the intersection of management, organization, and computer science.Therefore, consideration of ethics in IS practice is often seen to be close to business ethics and professional codes of conduct such as the ACM Code (Herwix et al., 2022).Business ethics deals with the ethical dimensions of the exchange of goods and services and the involved entities (Moriarty, 2008).As business ethics, IE can examine stakeholders' actions, consequences, and morality by applying general ethics theories.
The ethics perceptions of business students are one of the fast-growing research topics in the empirical business ethics literature (Brand, 2008;Smith, Soderberg, Netchaeva, & Okhuysen, 2022).Current students in universities face ethical issues as soon as they start working.They are mostly millennials with a very strong relationship with technology and, nowadays, make up the majority of the workforce being hired.Therefore, it is relevant for educators and business managers to know how they decide on ethical issues and act upon these decisions in general and with IT.The influence of cross-cultural ethical issues on decision-making is important for international business managers (Ermasova, 2021).

Past IE Studies in Students
We made an overview of past studies on IE in students after searching papers in Scopus and Web of Science databases using "inf* ethics" or "comp* ethics" or "IT ethics" in the article title, abstract, or keywords, and the following terms in abstract: "student" and "college" or "university" and "survey" or "study".After removing duplicates, conference papers, and papers that were not available in full text and the ones that were out of the scope (e.g., professional ethics of nursing students) or did not contain any empirical analysis, a total of 29 papers were used.In addition, twenty-two relevant papers cited by the selected studies were reviewed and added to the list, increasing the number to 51.The papers were scrutinized for theoretical frameworks, used methods, measurement scales, samples, and main findings.The systematic review of the papers is presented in Table A1 in Appendix 1.Using IE as a field of inquiry to examine students' perceptions and behavior toward ethical issues, we considered factors empirically examined in the IE literature.A brief overview of the individual factors with references to the studies is presented in the following sections.

Gender
Gender is the most used factor in the ethical decision-making (EDM) studies and is recommended as a control variable (Casali & Perano, 2021).
Most studies dealing with ethical perceptions and behavior among university students concluded that there are gender differences in the perception of IE issues and ethical behavioral intentions toward IT.Females were more sensitive to IT ethics issues than their male counterparts (Chiang, Chen, Teng, & Gu, 2008;Kim & An, 2017;Kini, Rominger, & Vijayaraman, 2000;Peslak, 2006).However, certain studies did not confirm differences in attitudes toward access-related issues (Mohamud, Zeki, & Saidin, 2016) or IE issues in general (Khan, 2012).Females were more consistent in their IE attitudes across various study programs (Akbulut, Uysal, Odabasi, & Kuzu, 2008).

Age and Degree Year
Less unanimous evidence is available regarding the influence of age and degree year on IE perceptions and behavioral intentions.While some studies suggest that older students are more mature and more inclined to ethical behavioral intentions with IT (Almasri & Tahat, 2018;Gan & Koh, 2006;Sendag, Duran, & Fraser, 2012), others did not report significant differences (Ozdogan & Eser, 2007).Age also belongs to the extensively researched factors influencing EDM that are recommended to use as a control variable in more complex studies involving less validated factors (Casali & Perano, 2021).

Marital Status and Religion
Unlike other demographic variables, marital status and religious beliefs have not played a significant role in how college students perceive IE (Cha & Song, 2020;Hsieh, Yeh, & Chuan, 2012;Mohamud, Zeki, & Saidin, 2016).In business research focused on EDM, marital status was not reported either, but religion was used and showed moderate influence on decision making (Casali & Perano, 2021).

Work Experience
Within the context of IE of university students, work experience is closely related to age, as students who study part-time already work or have work experience from the past.Students with work experience report more ethical behavioral intentions and more aware attitudes toward IE issues (Gan & Koh, 2006).This finding is supported by studies focusing on IE attitudes of working professionals (Tahat, Elian, Sawalha, & Al-Shaikh, 2014).In the review of EDM studies in business research, work experience was measured jointly with education and was found to be highly researched and moderately influential (Casali & Perano, 2021).

Other Factors Influencing Information Ethics
The influence of culture is a popular stream of research among international studies about students' IE.Many studies attempted to identify the influence of nationality and culture and reported differences in perceptions based on nationality (Al-Rafee & Dashti, 2012;Almasri & Tahat, 2018;Chang & Chen, 2016;Chiang & Lee, 2011;Kim & An, 2017;Martin & Woodward, 2011;Park, Oh, & Kang, 2016;Ulman et al., 2021;Yates & Harris, 2011).EDM studies in business report nationality as a highly influential factor that should be used as a control variable (Casali & Perano, 2021).
In EDM studies, group pressure is reported as a highly influential factor, although less often researched.Personality was a frequently researched factor that was also highly influential same as behavior.The intention was not reported as a standalone factor but should be included under the behavior variable (Casali & Perano, 2021).
Since the above-cited studies vary greatly in sample size, type of instruments, used theoretical frameworks, and are mostly based on convenience samples, it is impossible to generalize their results, let alone identify common patterns.Most of the analyzed studies recommend teachers pay more attention to teaching about IE and practical computer ethics issues in classes and managers to improve organizational policy and provide more training on appropriate ethical behavior at work.Teaching IE using a simulation game proved effective (Lin, Wang, & Yueh, 2022).

Research Model
In this section, we introduce our research model drawing from the theoretical foundations of Mason's PAPA framework (Mason, 1986).The PAPA framework is the basis for the research design and the empirical study and guides the discussion of the results.This framework sorts IE issues into four dimensions: Privacy, Accuracy, Property, and Accessibility (PAPA).

Privacy
The original concept of privacy deals with maintaining personal space and restricting the physical access of others to people or information about them (Westin, 1966).In the current context, privacy represents conditions and safeguards under which people disclose their personal information or some sort of control over access to one's personal information (Belanger, Hiller, & Smith, 2002;Stone, Gueutal, Gardner, & McClure, 1983).
Mason described an environment where individuals and organizations could share data in a rather limited way.However, he rightly pointed out the danger of the cumulative effect of multiple revelations of personal data.The current level of data sharing and use of big data analytics in facial recognition, real-time data collection, social networks, and predictive analytics challenge the traditional assumption that individuals should have control over their data and the right to decide about it.Such unprecedented development could not have been expected in the past.On the other hand, some individuals prefer a trade-off between their privacy and convenience.These users make conscious decisions and accept the associated risks (Niederman & Baker, 2022).

Accuracy
Accuracy stands for accountability, authenticity, constancy, and accuracy of information.Lack of accuracy may lead to misinformation that negatively affects people's lives, especially when inaccurate information is used by an authority (Mason, 1986).Since organizations collect data from multiple sources in many contexts, and often without human interventions during data collection and processing, the challenge to accuracy is much more serious.The issue of accuracy is a concern not only for industry actors who develop digital products and services and rely on big data but also for government organizations, researchers, and, most importantly, individuals who are either subject of data collection or rely on data in their decision making (Gordon et al., 2021).

Property
Property is related to intellectual property (IP) rights, data ownership, ways of data exchange, and evaluation of data (Mason, 1986).Property problems include software piracy and digital piracy; however, they go beyond these issues and make this category the most complex and challenging within the PAPA framework.The complexity of intellectual property rights is given by different national laws, their interpretation, and enforcement (e.g., Gallini & Scotchmer, 2002;Lemley, 2004).
Since data has become a commodity, the property issue has become more complex.Individuals may give consent to a third party to collect and use their data as input to a big data analytics system (e.g., giving consent to use location data in apps such as Waze, Google Maps, or Facebook) without knowing how the organization will use their personal data and for what purpose (Richardson, Petter, & Carter, 2021).These situations raise questions about data ownership and knowledge between individuals and organizations.

Accessibility
Accessibility covers information access and personal data protection, authorization of access to data, and its conditions.The preconditions for accessibility are intellectual skills to work with information, access to the connection, and access to the information itself (Mason, 1986).The first two prerequisites depend on economic and psychological investment, while the last relates to the ability to protect the information against illegitimate access (Niederman & Baker, 2022).
The principles of accessibility can be in practice in conflict with other PAPA dimensions.For example, IP rights protection would hinder accessibility, while access to inaccurate health-related information on the Internet may have severe consequences (Mathiesen, 2004); in healthcare, a patient's privacy might conflict with the need for fast access to the patient's electronic health records (Shen et al., 2019).
The constructs proposed by Mason proved to be correct, especially for privacy (Bélanger & Crossler, 2011).The four dimensions were reaffirmed to be valid in an empirical study made twenty years later (Peslak, 2006).Another study by Woodward, Martin, and Imboden (2011) suggested a refinement of the property component due to different levels of personal risk, while the remaining dimensions provided reliable results.
Although the PAPA framework identifies categories of ethical concerns in IT, it does not address the issues of what is comprised in each category.As mentioned in the description of the PAPA categories above, new instances of privacy, accuracy, property, and accessibility concerns are generated by new information and communication technologies (Niederman & Baker, 2022).However, the categorization is robust enough to describe the ethical issues (Stahl, 2021).It provides a solid starting point for categorizing the issues arising from emerging technologies such as artificial intelligence (AI) (Niederman & Baker, 2022).

Research Questions
Following the PAPA framework and past IE studies (Aasheim, Kaleta, & Rutner, 2021;North, Richardson, & North, 2017;Ulman, Marreiros, Quaresma, & Harris, 2021;Yates & Harris, 2011;Zhang, Simon, & Lee, 2016), we assume that people who behave unethically will likely do so even in situations when IE is in question.In addition, we can state that individual perceptions of what is ethical or unethical may vary between different PAPA dimensions.That is, an individual might have strong ethical views on property and not so strong on privacy -since they reflect different valuesand, consequently, behave differently in what concerns these issues.
Since there is no consensus in the literature on the influence of various sociodemographic variables on IT ethical perceptions and behavior, introducing a typology concerning IT ethical patterns is important.Measuring and analyzing perceptions and behavior at the individual level could enable group (cluster) users according to identified patterns and independent of their sociodemographic profiles.Each group or cluster would represent a common type of IE perception.
Segmentation is a basic concept that has been part of modern marketing for more than 40 years (Morgan, Whitler, Feng, & Chari, 2019).Segmentation is important not only in consumer and business-to-business marketing (Mora Cortez, Højbjerg Clarke, & Freytag, 2021) but also in healthcare (Chong, Lim, & Matchar, 2019), tourism (Quer & Peng, 2021), and other domains.However, studies on the segmentation of university students by their perceptions, behavior, or behavioral intentions toward IE issues are scarce.They deal only with software piracy (Gan & Koh, 2006;Hsu & Shiue, 2008), just one of the IE issues discussed in previous sections.
Therefore, we formulate the first research question: RQ1: Are there clusters of students with similar patterns of IE perceptions?
Additionally, it is important to see if groups with different perceptions have different behaviors.This analysis will help, on the one hand, to understand the different facets of IE perceptions and, on the other hand, help to understand how different IE perceptions impact behavior.Therefore, we formulate the second research question: RQ2: What differences exist between clusters in terms of IE-declared behavior?
Once the clusters are identified and their IE perceptions and behavioral intentions understood, it is crucial to analyze the sociodemographic characteristics of these groups.The characteristics may then guide teachers and managers in working with these groups.The basic sociodemographic variables identified as influential in EDM and IE studies will be used for cluster profiling: gender, age, degree year, type of study (full-time or part-time study), computer knowledge, and perceived importance of IE.Therefore, the third research question is as follows: RQ3: What differences exist in the sociodemographic profiles of the clusters?
In summary, our purpose is to investigate individuals' unethical use of IT and identify common patterns between different groups of individuals.The resulting research model is depicted in Figure 1.

Methodology
This section discusses the methodological approach, data collection, and analysis methods chosen to answer our research questions.

Sampling
As stated previously, the population in the analysis was college students enrolled in graduate and undergraduate programs at various business schools in seven countries.The selection of the respondents was based on a convenience sampling method, comprising 1,648 participants.We can argue that college students are an adequate subject to answer IE-related research questions.Students are frequently used in studies interested in the decision-making of managers, where a student sample serves for comparison (Leonard, Cronan, & Kreie, 2004) or even substitutes a managerial sample without a major threat to generalizability (Randall & Gibson, 1990).However, some authors argue that although students are highly appropriate research subjects, they should not be used as proxies for practicing managers (Campbell & Cowton, 2015).Nevertheless, using student samples for research of relationships between IT-related behavior and IE is relevant since school is where students form their cognition of IE, which they later bring to the workplace (Lawson, 2004;Nonis & Swift, 2001).The sensitivity to ethical issues does not differ significantly between, for example, undergraduate accounting students and practicing industry accountants (Fiolleau & Kaplan, 2017).
We have built a sample from seven countries, including the United States of America, one African country (Cape Verde), two Eastern European countries (Czech Republic and Hungary), and three Western European countries (Italy, Portugal, and Spain).The selection of countries was broad to address different levels of development, political systems, history, religion, and culture.While Africa is represented by at least one country, Asia and its non-christian countries remain uncovered.

Data Collection
We extended the original IT ethics survey developed by Harris (2000) to measure attitudes toward IE issues.This IT ethics survey was validated in multiple international studies with samples in languages other than English (Harris, Yates, Quaresma, & Harris, 2010;Ulman, Marreiros, Quaresma, & Harris, 2021;Woodward, Martin, & Imboden, 2011).
Observing ethical dilemmas in real life, particularly on a large scale, is infeasible since most people encounter these situations surreptitiously and often alone or in private.Even if observation were possible, social desirability bias would likely skew the results.Therefore, we elected to adopt an anonymous survey methodology.Further, we employed specific scenarios about ethical attitudes and practices to avoid the generalization bias encountered with hypothetical survey questions (Alexander & Becker, 1978;Reis & Gable, 2000).Potential responses ranged from completely legal to illegal (although no indication of legality was given to the respondent).The scenario approach is widely used in business ethics research (Fiolleau & Kaplan, 2017) and higher education (Streiner, Burkey, Dahm, Cimino, & Pascal, 2022), corporate and government ethics training (Ben-Jacob, 2005;Fleischmann, Robbins, & Wallace, 2009).
The scenarios presented in narratives to reflect IT use portray specific behaviors of fictional people familiar to the respondents.According to the PAPA framework, these scenarios represent different aspects of IE: privacy, accuracy, property, and accessibility (Mason, 1986).Following the PAPA dimensions, we designed the vignettes.Privacy-related questions included monitoring employees' e-mail and company web traffic, using work e-mail for private purposes, accessing payroll data, and collecting personal data without subjects' consent.Accuracy was represented with scenarios where data was intentionally manipulated, or software malfunction was not reported to receive a benefit or avoid penalty.The property consisted of cases where copyright or terms of use were breached in various ways.Accessibility examined scenarios that described software, hardware, and data usage violating rules and law.These categories seem to include the many ethical areas of concern to IS and business professionals today.
The survey also included a question about the personal belief of the respondent on the importance of IE, a question about their perceived computer knowledge, and questions on personal behavior with IT misuse.These personal behavior questions asked respondents if they had performed any unethical actions described in the scenarios (Appendix 3).However, since the responses about behavior are self-reported, they must be seen as subjective assessments rather than actual behavior.Sociodemographic questions were also included about respondents' age, gender, and year in school.
The original survey was conceived in English and then translated into the native language of each country in the study, i.e., Czech, Hungarian, Italian, Portuguese, and Spanish.We translated and backtranslated the instrument and had it reviewed by a native-speaker faculty.This ensured that the survey was meaningful to the respondents of all participant countries and consistent with the original version.We tested the translated versions of the survey on a small group of students to verify that they understood the phrasing and meaning of the scenarios.
Students were asked to participate voluntarily, and their anonymity and confidentiality were assured.By request of the study's authors, participating faculty members from each country administered the online questionnaire to their students.The estimated time to complete the survey was 30-50 minutes.
We collected data from business schools in the seven selected countries between October 2017 and January 2018.The breadth of countries represented in the sample is unique since most published papers focus only on one or two countries.

Data Analysis
Since individual perceptions of ethical behavior vary, it is important to explore common patterns of IE perceptions across PAPA dimensions and their group characteristics, as stated by our research questions.Also, it is important to see if groups with different perceptions have different ethical declared behaviors.For that, we calculated a summative scale for each PAPA dimension, adding the respective scenarios' scores in each dimension for each individual.Additionally, a total score for declared behavior was calculated, with seven as the maximum score meaning that the respondent declared he or she engaged in all the presented actions, and zero, meaning that the respondents declared they had not engaged in any of the actions.
Afterward, that data was subjected to cluster analysis of the PAPA perceptions dimensions' scores with a two-stage process.Hierarchical analysis was employed in the first stage to indicate the appropriate number of clusters.Hair, Black, Babin, and Anderson (2014) suggest a procedure based on inspecting the distance information from the collected responses.The appropriate number of clusters is found at the stage where a "large" increase in the distance measure indicates that a further merger would decrease homogeneity.Hair, Black, Babin, and Anderson (2014) also point out that selecting the final cluster solution requires substantial researcher judgment.This analysis showed that a four-cluster solution is ideal since a further increase in the number of clusters would not result in a considerable increase in the homogeneity of the clusters.
Following the hierarchical analysis and considering relative cluster size and the desire for parsimony, we employed the K-Means optimization procedure to generate a four-cluster solution.Information about cluster membership, a nominal cluster identity variable, and distance to the cluster center were saved for posterior analysis.We performed F-tests with the cluster variables.These are based upon differences between clusters where the null hypothesis states that average variable scores for each cluster are equal against an alternative hypothesis that they are not.

Results
In this section, we first describe the sample's overall demographics across the seven countries.Second, we analyze the groups from the cluster analysis of perceptions on ethical dilemma scenarios.We also report the significant differences among clusters in what respects individual IE-declared behavior and sociodemographic characteristics.

Sample Description
Table 1 summarizes the sample demographics by gender, age, degree year, and country.The sample (N = 1,648) was distributed as follows: 187 respondents from Cape Verde, 507 from the Czech Republic, 154 from Hungary, 379 from Italy, 115 from Portugal, 148 from Spain, and 158 from the USA.
Females were slightly overrepresented in the sample, especially in Hungary, Italy, Cape Verde, and Spain.The largest proportion of males originated from Portugal and then the United States.Only the Czech Republic presented a similar representation of women and men.The age structure was also significantly different across countries.The proportion of students under 20 was highest in Italy (44.9%), while students between 20 and 24 were the majority in the USA (75.9%), the Czech Republic (70.6%), and again in Italy (52%).
Undergraduate students prevailed except for the Czech Republic, where graduate students were the majority (62.1%).It should be noted that European countries have a three-year undergraduate program, while Cape Verde and the United States require a four-year program for a bachelor's degree.

Cluster Analysis
As explained in the methodology section, the two-step clustering procedure resulted in a four-cluster solution.Results from the F-test indicate that the four cluster variables have significantly different patterns between groups (p < .05).Therefore, the criteria used to cluster respondents can be considered meaningful.The Games-Howell (G-H) post-hoc test was applied to determine which clusters differ significantly from each other (Table 2).
The next step of the analysis was to interpret and profile the clusters.Each cluster is examined in terms of the cluster variate to assign a label accurately describing the nature of the clusters (Hair, Black, Babin, & Anderson, 2014, p. 448).The profile analysis involves the description of the clusters in terms of variables that were not used for clustering.These may include demographic, psychographic, and other measured variables (Aaker, Kumar, Leone, & Day, 2016).Moreover, the emphasis is on the characteristics that differ significantly across the clusters (Hair, Black, Babin, & Anderson, 2014, p. 449).The profile of each group was established from the mean of the four PAPA dimensions and from the identification of sociodemographic and behavioral variables for which there are significant differences between groups at a 5% level of significance based on an F-test for metric variables (Table 2) and a χ2 contingency test for nominal variables (Table 3).The tables show the means and frequencies of each cluster, and their comparison allows for the analysis of the variability among clusters and the total population.
The means, in Table 2, for the four PAPA dimensions and "Behavior" reflect the "ethics tendency"; a lower mean suggests a more tolerant attitude to IT misuse, and a higher mean reflects a stricter interpretation of IT actions that might be unethical.From the results in Table 2, it can be said that despite significant differences between clusters, respondents tend to perceive IE issues as unethical and declare not to engage in unethical behavior.Moreover, a self-reporting question about respondents' knowledge of computers revealed that, in the whole sample, students evaluated themselves, on average, as "knowledgeable" (mean 2.72).Additionally, most students believed that IE is an important issue for students and industry professionals (mean 3.62).
Out of the four identified clusters, Cluster 2 is the biggest cluster, with more than a third of the respondents (36%), followed by Cluster 3 (31%), Cluster 1 (22%), and finally, Cluster 4 (11%).We found significant differences in all sociodemographic (Table 3) and behavioral variables at the 5% level, except for gender and age.So, clusters differ from each other by how strongly they perceive the described behaviors as unethical.The four clusters score in the same position for all four dimensions.Therefore, people who tend to be more "ethical" (higher mean) are so in all four dimensions of PAPA.The opposite is also true.
Cluster 1 is more different from Cluster 3 and Cluster 4 and more similar to Cluster 2 in terms of sociodemographic characteristics, declared behavior, and PAPA perceptions scores.Cluster 1, which has the highest score in all PAPA variables, has a relatively larger proportion of Czech students (42.9% vs. 30.8% in the total sample) and a relatively smaller proportion of Spanish (4.2% vs. 9.0%), Hungarians (4.7% vs. 9.3%), Portuguese (4.7% vs. 7.0%), and Cape Verdeans (7.5% vs. 11.3%).This cluster has more first-year students (46.3% vs. 41.4% of the total sample), fourth-year students (34.1% vs. 28.1%),and relatively fewer students from other degree years.
In the importance of the IE variable, Cluster 1 members tend to believe, significantly more than respondents of Cluster 3 and 4, that IE is important both for college students and industry professionals.They also tend to think, more than respondents of Cluster 2, that they are knowledgeable about computers.Regarding ethical behavior, Cluster 1 differs significantly from Clusters 3 and 4, with the lowest mean on this variable, meaning that they tend to declare, less than other clusters, unethical behavior connected with IT.Because of their ethical perceptions and declared behavior, we can label Cluster 1 as "legalists," that is, "people who have an excessive adherence to law or formula, giving too much attention to the rules and details" (CUP, 2022a).
Cluster 2, the biggest group, follows Cluster 1 but differs significantly from it on PAPA dimensions' scores.They tend to judge the scenarios less strictly than Cluster 1 respondents but still consider the different IT scenarios presented mostly unethical.Therefore, we name Cluster 2 as "moralists," "people who are concerned with the code of interpersonal behavior that is considered right or acceptable in a particular society and have strong ideas about moral principles" (CUP, 2022b).
Respondents from Portugal and Italy are slightly overrepresented in Cluster 2, and students from Hungary are underrepresented.The degree year distribution of respondents in this cluster is similar to the total sample.Cluster 2 members, as members of Cluster 1, differ significantly from Clusters 3 and 4 in the perceived importance of IE.However, they differ from Cluster 1 in their declared computer knowledge, judging themselves, on average, as less knowledgeable.Cluster 2 members also report behaving more ethically in declared IE behavior than in Clusters 3 and 4.
Cluster 3, the second biggest group, follows the previous two groups regarding the mean scores of the four PAPA dimensions.This means that, on average, respondents in this group tend to condemn the unethical actions described in the various scenarios less strictly.As such, we label Cluster 3 as "pragmatists," that is, "they are people who deal with things sensibly and realistically in a way that is based on practical rather than theoretical considerations" (CUP, 2022c).Also, these respondents tend to declare more often to behave unethically, making them significantly different from those in Clusters 1 and 2.
Respondents from Spain, Hungary, and Cape Verde are somewhat overrepresented in Cluster 3, and there is a lower proportion of Czech students in this group compared to the total sample.As in Cluster 2, the degree year distribution in Cluster 3 is quite similar to the total sample.However, this cluster differs significantly from the previous two in the importance of IE, tending to think it is more important for professionals than students.
The smallest group is Cluster 4, which is also the group with the lowest mean scores on all four PAPA dimensions.Hence, we label Cluster 4 as "anarchists," who are "people who do not recognize the authority; they believe that government and laws are not necessary" (Merriam-Webster, 2022).
Cluster 4 is not significantly different from Cluster 3 in the importance of IE, IE-declared behavior, and perceived computer knowledge.This group has a relatively higher proportion of Spanish, Hungarian, and Cape Verdean respondents and a relatively lower proportion of Czech, Portuguese, and Italians.Students from the first and fourth years are underrepresented in this cluster, opposing students from the second and third years, who are overrepresented.

Discussion
This section will first discuss the main results of our study, framing them with previous research.We identify the implications of our results for practice and research and the limitations of the present study and propose directions for future research.

Main Results
Based on our analysis, we can reply to all our research questions.First, we found four groups of students with different perceptions of IE.Second, these groups do differ in IE-declared behavior.Finally, the identified groups, indeed, have different sociodemographic profiles.Even though students can recognize good or bad behavior, their ability to tell the difference is not very strong.This could be evidenced especially with the group labeled as anarchists since those students do not shy away from acting sometimes unethically and tend to see certain unethical behavior of others as less problematic.
Nevertheless, the "anarchist" cluster is the smallest, reinforcing the idea that most university students of the countries in the study do recognize IT-related unethical behaviors.This conclusion is supported by the fact that the "moralist" cluster accounts for more than one-third of the sample and that, together with the "legalists," represent almost 60% of respondents.It can be stated that university students tend to be more central on the ethical thinking scale -not so squeaky clean as "legalists," but certainly not what we call "anarchists".These results might be explained by students' computer knowledge and the ever-increasing importance of ethics subjects in university curricula.Our results are in line with research on IE with students from different countries (e.g., Beycioglu, 2009;Cilliers, 2017;Masrom & Ismail, 2008;North, Richardson, & North, 2017).
As in other studies (e.g., Almasri & Tahat, 2018;Chang & Chen, 2017;Ulman, Marreiros, Quaresma, & Harris, 2021), it could be concluded that the dimension of clusters varies among countries.Czechs tend to be more "legalists," while Portuguese and Italians act more like "moralists," and Spanish, Hungarian, and Cape Verdean turn out to be more "pragmatists" or "anarchists."American students seem to represent the population since they are not over or underrepresented in any of the clusters.
Even though a clear pattern could not be found in cluster country representation, these results might have a twofold explanation: Firstly, the Czech Republic is a country that stands out in the legalist cluster and, at the same time, is also the country where graduate students are the majority of the sample.Therefore, it might be argued that the higher the level of education, the more knowledgeable and responsible students are regarding ethical issues (Almasri & Tahat, 2018;Masrom & Ismail, 2008;Mohamud, Zeki, & Saidin, 2016).
Second, except for Italy, the samples with a higher percentage of women (Cape Verde, Italy, Spain, and Hungary) are the ones who act more like pragmatists or anarchists.Therefore, this confirms that women are more pragmatic and carry their pragmatism to questions of ethics and IT (Tahat, Elian, Sawalha, & Al-Shaikh, 2014).Some studies reported no effect of gender on IE perceptions (Khan, 2012;Mohamud, Zeki, & Saidin, 2016).Others reported that women perceive IE issues more sensitively in certain PAPA dimensions (Peslak, 2006) or are generally more aware of IE issues than men (Kuo, Lin, & Hsu, 2007;Masrom & Ismail, 2008).
Opposite to what was proposed in the research framework, people who tend to classify the described scenarios as unethical or computer crime do so in issues related to all four PAPA dimensions.This means that if more ethical judgments are made on privacy, they are also made on property, accessibility, and accuracy issues.The reverse also applies.The assumption that one person could perceive ethical issues from different PAPA dimensions differently could not be corroborated.Therefore, ethical views on IT-related issues seem unidimensional (Williamson, Clow, Walker, & Selwyn Ellis, 2011).
Moreover, we found that people with stronger ethical perceptions declare to engage less in unethical behavior, which shows that perceptions and behavior are related.This result follows the findings of similar studies that changing ethical perceptions is a path to improving behavior (Aasheim et al., 2021;Chiang, Chen, Teng, & Gu, 2008;Zhang, Simon, & Lee, 2016).
Nevertheless, our results also suggest that there are mediating factors that weaken the relationship between perceptions and behavior.Some of those factors might be the IE importance, and perceived computer knowledge since differences in these variables among clusters do not follow the same pattern of perceptions.For example, legalists and moralists have the same view about the importance of IE for organizations and students, and their behavior is not significantly different.This might mean it is not enough to teach people about IE to reduce IT misuse.Promoting ethical values and increasing users' computer knowledge is also necessary.This conclusion is in line with past research that concludes that unethical IT-related behavior is associated with various individual or circumstantial factors As we mentioned, perceptions and behaviors are related, but cluster members' declared behavior differences are weaker than those in perceptions.This may also be because the application of sanctions for IT misuse is rather small.Although there is a perception of what is wrong behavior, in practice, engaging in unethical behavior is not a problem if there is no perceived risk of sanctions (Fitriasih, Hati, & Achyar, 2019;Rybina, 2011;Williamson, Clow, Walker, & Selwyn Ellis, 2011).On the other hand, people tend to follow trends and be recognized by their peers (Tang & Farn, 2005).

Implications for Practice
We can highlight four managerial implications based on our findings.
First, the survey instrument we used can be adapted and applied to students enrolling in information literacy or information technology classes and as an entrance quiz for newly hired employees.The instrument would then allow to quickly indicate potential problems in students or employees with IE perceptions and serve as input for developing relevant interventions.
Second, it is important to consider that individual behavior can project into group behavior which affects the whole organization.At the same time, managers should not overestimate the perceived importance of IT ethics by current or prospective employees since other factors affect their behavior, such as lower computer knowledge.
Third, managers should know that some groups behave more unethically with IT.These groups with less strict perceptions of IE issues are characteristic of certain countries, ages, and work experience.Managers should consider this when revising organization and security policies and customizing interventions.
Fourth, information systems and computer science educators should consider including IE topics in the curricula or as an independent course.Combined with the indicative survey tool and profiles proposed in this research, the teachers should be able to better understand differences in IE perceptions and IE behavior among their students and adjust the curricula accordingly.

Implications for Research
We would like to highlight the following two implications for research based on our findings.
This study is the first to date that attempted to segment a cohort of university students based on their perceptions of IE issues.We identified four clusters, uncovering groups of people with very high (Legalists), high (Moralists), fair (Pragmatists), and low (Anarchists) sensitivity to the IE issues.The cluster analysis results proved that the cluster profiles could be used for predicting behavior concerning IE issues.Profiling can monitor progress in training programs focused on information literacy, data and privacy management, and other relevant topics at educational institutions and business or government organizations.
Second, this paper is the first to present a systematic literature review of IE studies with students, which is an addition to the body of literature and can serve as a starting point for future studies.

Limitations of the Study
The presented study has several limitations that we describe in the following text.
First, respondents came mainly from one university in most countries, which may weaken the external validity of the findings.The sample sizes were not balanced because of the smaller number of selected respondents from the USA, Portugal, Spain, and Hungary.Therefore, we recommend cautiousness before generalizing the present results since the research was based on a convenience sample of students.However, the high number of observations and the variety of respondents on what concerns various sociodemographic traits tend to be representative of the population (Bolthausen & Wüthrich, 2013;Sedlmeier & Gigerenzer, 1997).
Second, perceptions of IE issues were measured through several items for each PAPA dimension and as third-person scenarios to avoid socially biased responses.However, the behavior was measured with fewer items, not evenly distributed across the PAPA dimensions, using a dichotomous yes/no scale, which might limit its adequacy in comparing perceptions and declared behavior results.We made this methodological option to avoid a lengthy questionnaire.We propose redesigning the behavioral intention items into first-person perspective vignettes with the same scale as IE perception vignettes.This would allow us to test the relationships between IE perceptions, declared behavior, PAPA dimensions, and other variables, for example, from the Theory of Planned Behavior.
Third, the presented survey is a cross-sectional study design that cannot capture changes in behavioral intentions over time.A longitudinal study tracking the subjects over long periods would help better understand how different factors influence IE perception and behavior over time.
Fourth, the subjects used in this survey were university students.Despite the limitations of using student samples, as discussed in the literature review, students are at the forefront of technology trends and represent the future workforce entering the labor market.Therefore, the findings have face validity both for the educational and business environment.

Conclusion
Students' lack of sensitivity to IE issues is a problem for educators and business managers.Quantitative studies of student convenience samples focusing on influencing factors of IE currently prevail in the literature.This paper contributes to the research by segmenting students according to their perceptions of IE issues.Clusters allow the exploration of patterns in IE perceptions that would remain hidden at an individual level of analysis and can be the basis for predicting IE-related behavior.Students can understand and evaluate the various dimensions of IE issues.We argue that there is a positive relationship between IE and unethical behavior in IT use and that those vary among people with different sociodemographic profiles and levels of computer knowledge.Strengthening perceptions of IE issues can reduce unethical behavior and lower threats.
For research, this paper suggests extending the IT ethics survey instrument with first-person vignettes representing all PAPA dimensions to measure IE-declared behavior.Further, a longitudinal study monitoring the evolution of IE perceptions and IE-related behavior in students and working professionals should be conducted.This would allow researchers to explore variances between populations.Finally, future IE studies modeling PAPA dimensions and testing mediating variables' influence on IE perceptions and IE-related behavior would expand theoretical knowledge.For practice, the theory suggests that different groups require different interventions that should be customized to the sociodemographic profiles of students and employees.The study validates the PAPA issues and finds that all topics were considered important ethical issues among college students.Privacy was viewed as the most important, followed by accessibility and accuracy, which are viewed equally, and property which is viewed lowest but still very important.Gender played a significant role in determining the recognition of privacy and accuracy as important ethical issues.(Tang & Farn, 2005) Behavioral Model of Softlifting Taiwan 206 Experiment, exploratory factor analysis A classroom experiment supplemented with a survey.Both group pressure and financial gains are significant determinants of software piracy.The factors interact so that financial gain is not a significant factor when group pressure to software pirating dominates.In contrast, financial gains will support soft lifting intention when group pressure is toward purchasing.(Kini et al., 2000) Kohlberg's CMD USA 841 ANOVA Women tend less to hold software piracy intentions, but generalized stages of moral development may not influence the moral intensity of software piracy.The propensity toward moral intensity is affected by demographic variables.(Wong, 1995) -Hong Kong 46 Interview, focus groups Students who had attended a six-week sub-course on ethical and social matters related to computer use showed different attitudes than a control group that did not attend the course.
(14) Do you know anyone who has ever made an illegal copy of the software?(15) Have you ever downloaded songs or movies without paying for them when required?(16) Do you know anyone who has ever downloaded songs or movies without paying for them when required?

Importance of Ethics
In general, you believe ethics in information systems/computers is: Not an important issue for anyone An important issue for college students but not for industry professionals Not an important issue for college students but an important issue for industry professionals An important issue for both college students and industry professionals
#Values with the same letters across clusters are not significantly different for p < .05.

Table A1 .
Namlu and Odabasi (2007)2008)survey instrument was used.Male students have significantly less ethical behavioral intentions in three factors (selfishness, academic cheating, and computer ethics) than female students.The sample was taken only from one university.theunethicalcomputer using behavior scale (UECUBS) developed byNamlu and Odabasi (2007).Males and females did not show similar behavioral patterns across different programs of study.Females demonstrated consistent ethical judgments across different study programs, while males had different judgments according to the field of study.Triggered Academic Dishonesty Scale (ITADS) was developed and used in the survey.Fraudulence, plagiarism, falsification, delinquency, and unauthorized help were measuring e-dishonesty.Individual factors, institutional policies, and peer pressure influenced e-dishonesty.The effect of IE cognition on IE behavioral intentions was proved but was not the most important factor.Other influencing factors were gender, computer availability, computer experience length, and home internet access.Samples were taken from two junior high schools.Intellectual property violations seem to be more acceptable with using IT; students also report that for themselves, it is less problematic to cheat with IT than without it, despite the field of study.The study aimed to develop a scale (UECUBS) to determine unethical computer use behavioral intentions.A factor analysis of the related items revealed that the factors could be divided into intellectual property, social impact, safety and quality, net integrity, and information integrity.Students demonstrate no significant difference in ethical sensitivity by age, school grade, family income, or school ownership status.Having an ethics course has led to no significant difference in ethical sensitivity.Further, students with higher ethical sensitivity prefer including a compulsory ethics course in their curriculum.In contrast, those with lower ethical sensitivity tend to keep ethics courses away from the curriculum.Age was negatively related to software piracy, and computer experience or computer usage demonstrated a direct relationship to software piracy.Moreover, older respondents who used university software mainly at their workplace tended to pirate less frequently, while students tended to be pirates more often than university employees.