Examining the Use of Hofstede’s Uncertainty Avoidance Construct in a Major Role in Ethics Research

This research focused on research in the Journal of Business Ethics (hereafter JoBE), International Journal of Value Based Management (hereafter IJoVBM) and Teaching Business Ethics (hereafter TBE) that used Hofstede’s uncertainty avoidance construct (hereafter UA). We identified research that used Hofstede’s UA to significantly contribute to the development of hypotheses, to construct an independent variable or as an independent variable. Understanding how UA has been used could provide opportunities for future research that increases our understanding of differences in international behavior and/or perceptions. The data indicate that of the 77 articles, 66.2% used UA strictly to construct hypotheses, 5.2% to construct alternative cultural indices and 28.6% as an independent variable.


Introduction
This research examines the use of Hofstede's cultural construct of UA in international ethics research; however, while we only report on Hofstede's UA construct, the majority of the research we examined in the course of this study included all of Hofstede's constructs.. UA is a surrogate for a society's relative willingness to tolerate ambiguous outcomes.Hofstede's (1991, 150) UA construct was calculated as the combined score for a series of three questions dealing with: rules orientation, employment stability, and nervousness or stress at work.Rules orientation examines the rigidity of an individual's beliefs about following rules.The employment stability factor reflects the anticipated tenure for the individual's current job.The stress-at-work factor reflects individuals' responses to statement about being nervous or tense at work.The uses of Hofstede's cultural dimension of UA as noted above were reviewed as they occurred in the JoBE as well as in the IJoVBM and TBE; in 2004, the latter two journals were merged back into the JoBE.These categories allow readers to quickly search and understand what information is available in these journals.A similar data compilation was performed for the Journal of International Business by Rapp et al. (2011) who suggested that future researchers examine other journals to compliment their study.This is the second in a two part series of articles; the first article (Authors, 20xx) examined the use of Hofstede's UA construct as a definition or in a comparison role.In the process of our review, we define the use of UA in a major role as significantly contributing to the development of research hypotheses, to construct an independent variable or as an independent variable.Consequently, we grouped articles into these three categories.Taken together with our first article and Rapp et al., the data in these articles provide an insight into how UA has been used in international research -both in general (Rapp et al.) and specifically in the area of ethics (Authors).

Methodology
The first step in the data gathering process was to identify articles to examine from the JoBE, which has been published since 1982.We began by using the journal's online search function for 'UA'.To ensure that we identified all articles, we also used Google's Advanced Scholar searching for 'UA' in the JoBE.We included only original journal articles; book reviews, comments, discussions and rejoinders were not included in our analysis.While this journal has been published since 1982, the first use of Hofstede's UA construct was in 1990; so our study actually includes articles from 1990 through 2011 (e.g., a 22-year period).While this journal has been published since 1982, the first use of Hofstede's UA construct was in 1990; so our study actually includes articles from 1990 through 2011 (e.g., a 22-year period).
While our primary aim was to include articles from the JoBE, the publisher of this journal also published the IJoVBM from 1988 through 2003 and TBE from 1997 through 2003.These journals were 'merged into the JoBE' at the beginning of 2004.We submit that article counts in the JoBE between 1988 and 2003 (i.e., when these journals were independently published) would be understated compared to other years if these journals were not considered in the article count.Consequently, our total article count and classifications include publications from the IJoVBM and TBE.After identifying the articles that included Hofstede's UA, the second step in the data gathering process was to determine how UA was used in each article.We classified the 77 articles as: significantly contributing to the development of research hypotheses, to construct an independent variable or as an independent variable.After classifying the articles, we subsequently reviewed the classifications for validation purposes and resolved any classification differences.
For the 22 years of this study, we initially identified 161 articles that used Hofstede's UA construct.Of the 77 articles that included Hofstede's UA, 70 were in the JoBE; five were in the IJoVBM; and, two were in TBE.The trend-line data in Figure 1 portray the growth in the use of Hofstede's UA in ethics research hypothesis development (dashed trend line) and as an independent variable (dotted trend line).

Overview
For the 22 years of this study, we found 77 articles.Figure 1 shows the increasing use of Hofstede's cultural dimension of UA in research shown in the JoBE, IJoVBM and TBE.Even though Hofstede's book first appeared in 1980 and cited four cultural dimensions including UA, the first article found in the JoBE that referred to Hofstede's cultural dimension of UA did not appear until 1990.There are multiple reasons for the time lag between Hofstede's first article and the first article referring to his cultural dimensions in this journal.One reason is that research at that time was not as readily accessible as it is today.In addition, there is the concern that articles before 1990 were scanned into online databases which disallow users to search articles for keywords.Therefore there could potentially be articles before 1990 that referred to Hofstede's cultural dimensions, but they did not appear in these searches.The 77 articles that included UA were categorized into three groups.The first group of 51 articles uses UA in hypothesis development in the literature review.The second group includes four articles that use UA in addition to an alternative cultural index.The third group includes 22 articles that use UA as a variable in the research.The division into three groups allows readers to better organize the different uses of UA.The number of articles appearing in our tables exceeds the number of total articles in the study because four of the articles (Voyer and Beamish, 2004;Peng and Beamish, 2008;Salter et al., 2001;Li, 2008) appear in more than one table.(Robertson, 2002;Rallapalli, 1999;Vega 1997;Goodwin & Goodwin, 1999;Thorne & Saunder, 2002;Hoffman, 1998;Yang et al., 2009).Four use UA to support research hypotheses.These articles suggest an association between UA and Japan's success and negotiating effectiveness (Parnell & Kedia, 1996), planning for the future (Lin & Yeh, 2009), accounting firms and behaviors (Cohen et al., 1992), trust levels (Goel et al., 2005) and Confucian work dynamism (Woodbine, 2004).The last article in Panel A (Volkema & Fleury, 2002) uses UA to describe the United States.

Using UA in the Literature Review to Construct Hypotheses
Panel B contains seven articles that use UA in ethics research.All suggested associations between UA and specific ethical behaviors or actions.Five found associations between UA and perceived ethical problems (Armstrong, 1996;Cherry et al., 2003), cheating (Salter et al., 2001), visible codes of ethics (Weaver, 2001), and validity of codes of ethics in Morocco (Oumlil & Balloum, 2009).The other two articles in Panel B suggested associations between UA and ethics without specifically mentioning any particular aspect of ethics (De Bock & Van Kenhove, 2010;Swaidan et al., 2006).
The articles in Panel C used UA in the literature review to contrast two countries.They compared Austria and Turkey (Smka et al., 2007), Japan and the Netherlands (Van Es & Pels, 2010), Japanese and Hispanics (Fadil, 1997) and North Americans and Asians (Caldwell & Clapham, 2003).Two articles include China in the two-country comparisons; Ge and Thomas (2008) compared China and Canada, while Tsui and Windsor (2001) compared China and Australia.Eight of the articles in Panel C compared the United States to: Brazil (Beekun et al., 2003), Croatia (Tavakoli et al., 2003), Egypt (Beekun et al., 2008), Jamaica (Sims & Keenan, 1999), Korea (Kim et al., 2010), Mexico (Daspro, 2009), Spain (Vitell & Hidalgo, 2006), Taiwan (Lu et al., 1999) and Turkey (Rawwas et al., 2005).The last article used UA to hypothesize about differences between the United States and Chinese based on manager incentives (Douglas & Wier, 2005).
The articles in Panel D used UA in studies that compared three or more countries.Ten of the fourteen included the United States as one of the countries researched.These ten articles used UA to compare the United States to Australia, Israel, South Africa and Turkey (Sims & Gegez, 2004), Canada, Colombia, Ecuador, Hong Kong and Ireland (Bernardi & Guptill, 2008), Canada, France and UK (Freeman & Hasnaoui, 2011), Canada, Hong Kong, Japan, Singapore and Thailand (Baker & Veit, 1998), Canada and Mexico (Sower et al., 1998), China, France, Germany and Hong Kong (Ramasamy & Yeung, 2009), Germany and Italy (Habisch et al., 2011), Hong Kong, Malaysia and Taiwan (Shafer et al., 2007), Thailand and Turkey (Burnaz et al., 2009) and Colombia, Ecuador and South Africa (Bernardi et al., 2009).The remaining four articles compared Australia, Canada and Sweden (Svensson et al., 2009), Gulf Cooperation Council Countries (At-Twaijri & Al-Muhaiza, 1996), Mexico, China, France, Japan and Germany (Gulbro & Herbig, 1998) and Latin American countries (Husted, 2002).
The Table 1 articles go into considerable detail concerning potential ethical differences as they relate to UA.All four panels include articles that used UA to support ethics research.There are nine articles that use one of Hofstede's stated connections between UA and behavioral contrasts (Table 1) in ethics research.They include: company rules should not be broken (Goodwin & Goodwin, 1999;Throne & Saunders, 2002;Cohen et al., 1992;Ge & Thomas, 2008;Lu et al., 1999); higher resistance to change (Vega, 1997); and, higher corruption in wealthy countries (Volkema & Fleury, 2002).Within Panel A, are five articles that highlight the associations between UA, behavioral contrasts and ethics.These articles describe that in high UA countries: company rules should be not broken (Goodwin & Goodwin, 1999;Thorne & Saunders, 2002;Cohen et al., 1992); there will be more resistance to change (Vega, 1997); and, corruption will be more prevalent in wealthy countries (Volkema & Fleury, 2002).One article explains that high UA associates with greater fear of the future (Armstrong, 1996) and another states that in high UA societies only known risks are taken (Swaidan et al., 2006).
The collection of articles in Panel C includes Ge & Thomas (2008) and Lu et al. (1999) which illustrate that, in high uncertainty avoidant societies, company rules should not be broken.In Panel D, one article (At-Twaijri & Al-Muhaiza, 1996) explained that in high uncertainty avoidant countries, experts are given key positions.In Panel B, there was an association between high UA and an increased level of importance of ethical problems (Armstrong, 1996), decreased in perception of ethical problems (Cherry et al., 2003), increased visible ethics initiatives (Weaver, 2001), decreased self-regulations (Salter et al., 2001), increased cheating (Salter et al., 2001), increased cheating for internal reasons (Salter et al., 2001) and increased view that an absence of rules is a license to do as one desires (Swaidan et al., 2006).

UA -Uncertainty Avoidance
The authors of the articles in Table 1 suggest differences other than those discussed by Hofstede.These include: high UA and less whistle-blowing (Zhang et al., 2009;Cohen et al., 1992); being less likely to perceive ethical problems (Cherry, 2006;Cherry et al., 2003); and referring to other professional codes of ethics when forming one's code (Vitell et al, 1993;Thorne & Saunders, 2002).In Panel A, results indicate that with high levels of uncertainty, people will be less likely to question rules (Rallapalli, 1999;Cohen et al., 1992) and have a greater intolerance of deviation from group norms (Vega, 1997).In Panel C, the following articles associated high UA with increased ethical reasoning (Tsui & Windsor, 2001), increased cheating (Srnka et al., 2007), increased level of importance of ethical problems (Tavakoli et al., 2003;Sims & Kennan, 1999), increased amount of employees placing their company's interests above their own interests (Vitell & Hidalgo, 2006) and increased probability that when faced with an ethical dilemma, one would follow the stated rules instead of making their own decision (Rawwas et al., 2005).Two articles in Panel D indicated that when there is high UA, there is a greater need for written rules (Sower et al., 1998) and concern about following the rules (Bernardi et al., 2009).
There is an overlap that occurs among panels; Panel A and C use UA to suggest differences in breaking company rules (Goodwin & Goodwin, 1999;Throne & Saunders, 2002;Cohen et al., 1992;Ge & Thomas, 2008;Lu et al., 1999).Panels B and C illustrate the potential differences using UA and the severity of perceptions of violations of ethics (Armstrong, 1996;Sims & Keenan, 1999) and suggest that UA may result in differences in the likelihood of cheating (Salter et al., 2001;Ge & Thomas, 2008).Panels C and D suggest a tendency towards rules orientation may differ with UA (Sims & Kennan, 1999;Sower et al., 1998).

Using UA to Construct an Alternative Cultural Indices
Kogut and Singh (1982) developed a calculation that uses Hofstede's cultural dimensions to construct a single score for each country rather than four individual scores (i.e., if one used Hofstede's dimensions directly).To compute their single index (Formula (1)), Kogut and Singh calculate differences in cultures by subtracting each target country's cultural dimensions (Hofstede, 1980) from the same cultural dimension of a control country, which is usually the United States.Once all of the cultural dimensions have been subtracted, the results are totaled for a single score.
(1) Kogut & Singh, 1988: 422.Where I ij : Index for the ith cultural dimension and jth country; (2) Uhlenbruck, 2004: 117. Where: I ia : Index for the ith cultural dimension for the ath country; I it : Index for the ith cultural dimension for the tth country; V i : Variance of the index of the ith cultural dimension; │D at │: Number of compared measures.
Two articles use Kogut and Singh's Index to compare it to UA. Li (2008) used Kogut and Singh to explain the likelihood of joint ventures, while Peng and Beamish (2008) related Kogut and Singh to the predictability of general economic data.The third article used Uhlenbruck to explain cultural differences in ethics (Gopalan & Thomson, 2003).The final article used Kogut and Singh's index as well as Uhlenbruck's.Voyer and Beamish (2004) used both cultural indices to explain differences between Japan and other countries.As mentioned briefly in the introduction, it is difficult to speculate on the association between alternative cultural indices, behavioral contrasts and ethics research due to the small number of articles that appear in Table 2. Of these four articles, none report associations between the alternative cultural index and ethics.This is not to say that these associations do not occur; it was just not evident in these four articles.The lack of research using these indices does, however, suggest a significant opportunity for future research.Such research could consider the general use of alternative cultural indices or examine potential associations between alternative cultural indices and ethics.

Using UA as an Independent Variable
Table 3 presents articles that use UA as an independent variable.As seen from this table, these hypotheses propose associations between cultures or behaviors.Similar to the previous tables, Table 3 is broken into three panels depending on article characteristics.Panel A presents the ten articles that used survey data for statistical analysis.Panel B contains six articles that used country data in the analysis.Panel C has six articles that used alternative sources and firm level data.Panel A includes articles with data from surveys of either students or adult college graduates in the workforce.Two articles use UA to suggest differences between cultures.Lin & Ho (2010) contrast the differences in UA between Taiwan and China, and Tan & Chow (2009) explain cultural differences between American Caucasians and people of Chinese descent who reside in either China or the United States.Five describe potential positive associations between UA and worker, labor and individual rights (Auger et al., 2007); good business, commitment, public relations role and corporate social responsibility (Kim & Kim, 2010); social desirability response bias (Bernardi, 2006); relativism (Forsyth et al., 2008) and, collectivism, long-term orientation, idealism and transformational leadership (Smith, 2011).In one article, Arnold et al. (2007) discovered a negative association between UA and measures of sensitivity.The last two articles found a relationship between UA and a behavior, but did not state whether this relationship was positive or negative.Peek et al. (2007) associated whistle-blowing and sexual harassment to UA; Christie et al. (2003) associated questionable ethical behavior with UA.
Panel B contains articles that utilize UA as an independent variable and publicly available database information from a variety of sources in the analysis.The sample sizes range from 30-to-133 countries.Four articles found positive associations with UA, one found a negative association and one found no association.UA was positively associated with OECD convention ratification (Baugh et al., 2010), corruption perception index (Li et al., 2008;Davis & Ruhe, 2003;Sanyal, 2005) and preference for bureaucratic structures, nationalism and unethical behavior (Davis & Ruhe, 2003).UA was negatively associated with individualism (Peng and Li, 2009).Husted (2000) found no association between UA and software piracy.
Panel C has six articles that used firm level alternative sample sources.Johan and Najar (2010) collected a sample from 123 firms which represented 23 countries.This article found that UA is positively associated with high efficiency and high performance fees.The next one included a sample from 528 firms in 20 countries (Cumming et al., 2010).This article found that UA is positively associated with Hofstede's masculinity dimension and the Corruption Perception Index as well as negatively associated with the power distance dimension.Clements et al. (2009) had a sample of 104 accounting firms; this article found that UA is positively associated with power distance and is negatively associated with individualism.Scholtens and Dam (2007) collected data from 2683 firms that came from 24 countries; these authors showed that UA is positively related with a firm's ethical policies.The third article in Panel C had a sample that utilized information from Estonian organizations (Alas, 2006).While Alas found that UA is positively associated with values of ethics, UA was negatively associated with the practice of ethics.The last article (Chen et al., 2008) had a sample from 2,018 firms representing 55 countries, which found no significant associations between UA and the variables examined.
The articles assigned to Table 3 incorporate an increased level of use of UA in ethics research, especially when creating, testing and discussing variables.These articles use both Hofstede's discussed associations as well as independent associations created by the authors.There are five articles that directly use Hofstede's suggested relationships between UA and ethics.In Panel A, two articles used Hofstede's potential connections (Table 3) between two factors.These articles found that, in high UA societies, only known risks are taken (Kim & Kim, 2010) and that there is preference for tasks with sure outcomes, no risks and following instructions (Peek et al., 2007).In Panel B, there is one article that uses Hofstede's potential differences.Davis and Ruhe (2003) found that high UA associates with an increased resistance to change.As seen in Panel C, Clements et al. (2000) found that as UA increases resistance to change also increases.Alas (2006) found that high UA societies believe that company rules should not be broken.
There were 15 articles with associations that were not suggested by Hofstede.The Panel A studies indicated that in high UA countries there is a preference for: a structured environment (Auger et al., 2007), more rules and standards (Kim & Kim, 2010), following rules and procedures (Peek et al., 2007), and leaders who have clear rules, procedures and guidelines (Smith, 2010).High UA countries also display increased: ethical sensitivity towards stakeholders (Kim & Kim, 2010), focus on legality instead of ethics (Christie et al., 2003), corruption (Arnold et al., 2007;Bernardi, 2006), social desirability response bias (Bernardi, 2006), and a lower tendency to whistle-blow (Peek et al., 2007).Panel A has more associations between UA and ethics than the other two panels of Table 3.In Panel B, there were three articles that included other potential differences.When UA is high: corruption is higher (Li et al., 2008); preference for structured order is higher (Davis & Ruhe, 2003); and, a decreased likelihood of recognizing ethical issues in the absence of formal rules (Husted, 2000).
The associations between high UA and high corruption can be found in Panels A and B (Bernardi, 2006;Li et al., 2008;Davis & Ruhe, 2003).Associations between high UA and more resistance to change are found in Panels B and C (Davis & Ruhe, 2003;Clements et al., 2009).The associations shown in this table between UA and ethics play a greater role in these articles than those in other tables because these variables were used directly rather than solely as support material.

Discussion and Conclusions
The data indicate that 66.2% of the articles used uncertainty avoidance strictly in the literature review to construct hypotheses, 5.2% to construct alternative cultural indices and 28.6% as an independent variable.However, an important difference between the data in this research and in Rapp et al. (2011) is that, while Rapp et al. found that 100 of the 118 (84.7%) articles they identified were in the three categories we examined, only 77 of the 161 (47.8%) articles we identified were in these three categories.Consequently, while the percentage comparison between studies may appear to be similar, the article counts represented are dramatically different.
In this research, only four of the 77 articles (5.2%) used Hofstede's cultural constructs in the computation of cultural distance indices; this compares to 29 of the 100 articles (29.0%) that Rapp et al. (2011) reported which used this methodology.This difference suggests that the use of Kogut and Singh's (1988) and Uhlenbruck's (2004) indices has not been adopted by ethics scholars, which presents an avenue for future research.The remaining 22 articles (28.6%) were the only ones that used UA as an independent or control variable compared to the 30.0%that  Bernardi, 2006) that could provide insights to differences in international behavior and/or perceptions.In wealthy countries, more corruption.
Table 4 summarizes associations related to Hofstede's UA index.Panel A illustrates seven of Hofstede's stated associations between UA and ethics.These include both low and high UA.The articles with findings related to low UA stated that, if necessary, employees may break rules (Goodwin & Goodwin, 1999;Cohen et al., 1992), there is less resistance to changes (Davis & Ruhe, 2003), willingness to take unknown risks (Swaidan et al., 2006) and there is less corruption in wealthy countries (Volkema & Fleury, 2002).The articles that had findings associated with high UA stated that company rules should not be broken (Thorne & Saunders, 2002;Ge & Thomas, 2008;Lu et al., 1999;Alas, 2006), there is more resistance to changes (Vega, 2007;Clements et al., 2009), only known risks are taken (Kim & Kim, 2010), there is fear of failure (Armstrong, 1996), preference is for tasks with certainty in outcomes, no risks and following instructions (Peek et al., 2007) and when experts are in key positions (At-Twaijri & Al-Muhaiza, 1996).
A limitation of our research is that it considered only articles in three ethics journals: the JoBE, the IJoVBM and TBE; of these journals, the last two have not been published since 2003.This limitation also provides the opportunity for future research which could examine the use of Hofstede's UA construct in other fields and topical areas to determine whether the results of this research are supported.

Figure 1 .
Figure 1.Number of articles using Hofstede's UA in ethics research by year Note: Hypothesis development (diamonds and dashed trend line) and independent variable (squares and dotted trend line).
V i : Variance of the index of the ith cultural dimension; u : Indicates the United States Uhlenbruck's (2004) cultural index (Formula (2)) is the same as Kogut and Singh's index; instead of using Hofstede's cultural dimensions, Uhlenbruck used Trompenaar and Hampden-Turner's (1998) dimensions.Uhlenbruck uses the same technique as Kogut and Singh, which results in a single score for each country.Uhlenbruck uses three of Trompenaar and Hampden-Turner's (1998) dimensions: individualism, specificity and attitudes towards organizations.Table 2 contains three articles that use either the Kogut and Singh index or Uhlenbruck index.

Table 1
Panel D includes articles that examine three or more countries.The 13 articles in Panel A of Table1include UA only in the literature reviews and can be subcategorized into three sections: definitions, hypotheses for research and one country studies.Seven simply describe UA

Table 1a .
Literature Review Only Articles

Table 1b .
Literature Review Only Articles

Table 1c .
Literature Review Only ArticlesItalicized data indicate publications in the IJoVBM and TBE.

Table 2 .
Articles Employing a Cultural Index Note: Italicized data indicate publications in the IJoVBM and TBE.UA -Uncertainty Avoidance.

Table 3a .
Articles Employing Uncertainty Avoidance as a Variable Note: Italicized data indicate publications in the IJoVBM and TBE.UA -Uncertainty Avoidance.
Rapp et al. (2011)The relative absence of the use ofKogut and Singh's and Uhlenbruck's  indices (5.2% versus 29.0%) between ethics research and research published in the Journal of International Business Studies is striking.Taken together, the current research andRapp et al. (2011)provide a useful indicator for future research areas in ethics research.There are ample opportunities for comparing samples from various countries using either Kogut and Sing's and Uhlenbruck's indices or UA as an independent variable (i.e.,

Table 4 .
Associations between Hofstede's Uncertainty Avoidance and Ethics Research