Journal of Business Ethics

, Volume 114, Issue 2, pp 325–339

An Empirical Analysis of the Ethical Reasoning of Tax Practitioners

Authors

    • Kemmy Business SchoolUniversity of Limerick
  • Jane Frecknall Hughes
    • The Open University Business School
  • Barbara Summers
    • Leeds University Business SchoolUniversity of Leeds
Article

DOI: 10.1007/s10551-012-1347-x

Cite this article as:
Doyle, E., Frecknall Hughes, J. & Summers, B. J Bus Ethics (2013) 114: 325. doi:10.1007/s10551-012-1347-x
  • 1.1k Views

Abstract

How tax practitioners approach ethical dilemmas remains generally unexplored in academic literature. We use here Rest’s original Defining Issues Test (Development in judging moral issues. Minneapolis: University of Minnesota Press, 1979; Moral development. Advances in research and theory. New York: Praeger Publishers, 1986), combined with a tax context-specific test and in conjunction with a control group of non-tax specialists, to examine tax practitioners’ moral reasoning in a social and tax context. We investigate: (i) the effect of a tax context on issues raised (finding that practitioners generally reason at lower levels than in social scenarios); (ii) whether the profession attracts people who reason at certain levels (finding that it does not); and (iii) whether practitioners are affected by training/socialization in their professional context (finding that that they are).

Keywords

DITMoral reasoningRevenue practitionerTax practiceTax practitioner

Introduction

In recent years, there has been a growing concern regarding the ethical behavior of tax practitioners (Shafer and Simmons 2008). Many firms in the United States, have been investigated for facilitating aggressive tax avoidance through the marketing of questionable tax shelters (Herman 2004; Johnston 2004; Scannell 2005) and companies, driven by their top executives, are often accused of using “tax havens” or tax shelters for the primary purpose of avoiding or, indeed, evading their tax obligations (Dyreng et al. 2007, 2010; Godar et al. 2005; Wilson 2009). Shafer and Simmons (2008) suggest that some tax advisers have abandoned concern for the public interest or social welfare in favor of commercialism and client advocacy, and go so far as to suggest that tax practitioners do not believe strongly in the value of ethical or socially responsible corporate behavior.

While ethics have been identified as a significant variable influencing tax practitioners (Milliron 1988) and some studies have identified the particular ethical issues they face (see, in particular, Marshall et al. 1998, in Australia and Yetmar et al. 1998, in the United States), to date little work has been done to investigate the manner in which tax practitioners approach ethical dilemmas, in other words, their ethical reasoning. We look to address this gap in the literature here. If action is to be taken to encourage an improvement in the ethical behavior of tax practitioners, a better understanding of the factors contributing to unethical behavior is needed. It may be that the tax profession, for example, attracts individuals who are more prone to low-level ethical reasoning, which is one of the components of ethical behavior according to Rest’s (1983) four component model. On the other hand, considering issues in the tax context may lead to forms of thinking that make unethical action less problematic for both tax practitioners and non-specialists, such as seeing tax evasion as a “victimless crime,” or making it easier to focus on the benefits to the individual of saving tax, rather than the loss to society in aggregate from taxes not collected. Alternatively, being involved professionally in the tax industry may affect ethical reasoning. For example, does the need to be aware of the intricate details of regulations lead people to think more in terms of following the letter of the law rather than its spirit or does the culture in the organizations in which tax practitioners’ work promote certain ways of viewing a situation? Different causes of unethical behavior, or combinations thereof, would need to be addressed using different approaches.

In this study, we look to unpick the potential influences of the individual decision maker and the context in which ethical dilemmas arise, by assessing moral reasoning using a well-known and validated psychometric instrument, the Defining Issues Test (DIT), which has been used in numerous studies investigating moral reasoning in a variety of business contexts. We combine the original DIT with a context-specific adaptation of the instrument which poses similar ethical dilemmas, but within the tax domain.

We use a 2 × 2 quasi-experimental design to compare ethical reasoning across different contexts (social vs. tax) and different groups of participants (tax professionals vs. non-specialists) in Ireland.1 Control groups of non-specialists have not previously been included in context-specific DIT research, but the use of a control group is important here in separating the characteristics of the individual from those of the context. We also further sub-divide tax professionals into those who work for private firms (whose focus tends toward minimizing tax liabilities for their clients) and those who work for the Revenue authority (and are therefore focused on collecting taxes to be used for the public good) to investigate further the impact of socialization in the tax profession. Our investigation addresses the following research questions:
  1. 1.

    Is the level of moral reasoning of those in the tax profession generally lower than in the population as a whole?

     
  2. 2.

    Does locating an ethical dilemma in a tax context alter the way in which people reason about it?

     
  3. 3.

    Does the professional involvement of tax professionals with tax affect their reasoning in the tax context?

     
  4. 4.

    Does the socialization/training of tax professionals affect their reasoning in the tax context?

     

The remainder of this article is set out as follows. “Measuring Levels of Moral Reasoning: The Defining Issues Test (DIT)” section discusses the development of the moral reasoning test used in this article, the DIT, and outlines the need for context to be considered in examining moral reasoning. “Research Questions and Hypotheses” section develops the research questions into testable hypotheses, while “Research Method” section describes the methodology used. “Results” section presents the findings which are then discussed in “Discussion” section. The final section provides conclusions to the article.

Measuring Levels of Moral Reasoning: The Defining Issues Test (DIT)

Cognitive developmental psychologists believe that before an individual reaches a decision about how and whether to behave ethically in a specific situation, ethical or moral reasoning takes place at a cognitive level.2 The psychology of moral reasoning aims to understand how people think about moral dilemmas and the processes they use in approaching them. The considerations used by individuals as a basis for their ethical reasoning develop as they grow older and there is empirical evidence to support the contention that moral reasoning ability develops sequentially, progressively considering broader issues (Kohlberg 1973; Rest 1979a).

Kohlberg (1969) developed a model of ethical cognition based on interview studies in which participants responded in an open-ended way to ethical dilemmas. He sets out three developmental levels: “pre-conventional morality,” “conventional morality,” and “post-conventional morality.” At the pre-conventional level, an individual is focused entirely on him/herself. He/she considers him/herself removed from the normal rules and expectations of society. Behavior appearing ethical to a person at this level is motivated solely by the person’s desire to avoid punishment or because the outcome of the behavior is in his/her best interests. At the conventional level of reasoning, an individual is concerned about family, society, the welfare of others, and the perception that others have of him or her. Individuals at this level see themselves in relation to others and as part of, and loyal to, the wider community. At the post-conventional level, also known as the principled level of reasoning, an individual is concerned about others in society and will act on behalf of others even if that means breaking established rules of law. Within each of the three levels, there are two developmental steps, resulting in a total of six stages, with the second stage in each level being a more advanced and organized form of the first. Each successive stage in Kohlberg’s model is considered qualitatively higher both cognitively and ethically. Individuals move upwards through the stages during their development, although not all individuals reach post-conventional morality.

Kohlberg’s perspective revolves around the idea that our moral judgements are the product of a conscious decision process during which individuals move directly from conscious reasoning to moral verdict (Kohlberg 1981). It therefore ignores the importance of moral emotions (Greene and Haidt 2002). Many researchers argue that both conscious reasoning and intuition play a part in judgement (Greene and Haidt 2002; Cushman et al. 2006). The moral dilemma itself (Greene et al. 2001), the cognitive focus of the judgement maker (Lewicka 1997), and the moral principle being triggered (Cushman et al. 2006) may all impact on the extent to which cognitive reasoning dominates intuition or vice versa. It is acknowledged by Haidt (2007) that the use of conscious verbal reasoning, the re-framing of the scenario, and discussing the situation with others serve to override an individual’s immediate intuitive response, particularly in a professional context. Given that tax professionals would have to communicate their rationale to the client, and potentially the Revenue authority, and that the situation is not personal to them so reasoning may play a more important role (as suggested by Greene et al. 2001), Kohlberg’s cognitive moral reasoning approach can provide useful insight and remains a valid construct in the context of examining tax practitioners.

Rest developed the DIT in 1979 (Rest 1979a) using Kohlberg’s cognitive development theory as a basis. The DIT tests moral reasoning using social dilemmas. It is a self-administered, multiple-choice instrument, making use of the same ethical dilemmas used by Kohlberg in his original analysis. Rest (1979b) developed the items for the instrument based on an interpretation of the stages in Kohlberg’s stage-sequence theory (see Table 1). The test assumes that a person can operate at many stages at once, and, rather than attempting to assess the stage to which a person “belongs,” it instead measures the comprehension and preference for the principled level of reasoning (i.e., post-conventional reasoning at stages five and six) (Rest et al. 1999).
Table 1

Six stages of moral reasoninga

Pre-conventional: Focuses on the individual

Stage one

The morality of obedience: do what you are told

Stage two

The morality of instrumental egoism and simple exchange: let’s make a deal

Conventional: Focuses on the group and relationships

Stage three

The morality of interpersonal concordance: be considerate, nice and kind: you’ll make friends

Stage four

The morality of law and duty to the social order: everyone in society is obligated to and protected by the law

Post-conventional: Focuses on the inner self and personally held principles

Stage five

The morality of consensus-building procedures: you are obligated by the arrangements that are agreed to by due process procedures

Stage six

The morality of non-arbitrary social cooperation: morality is defined by how rational and impartial people would ideally organize cooperation

aAdapted from Rest (1994)

Participants taking the DIT are presented with either six (long-form DIT) or three (short-form DIT) ethical dilemmas stated in a third person form. The dilemmas are presented as narratives describing the circumstances of a third party faced with making a decision on how to act in a given scenario. After reviewing the dilemmas, participants choose what the actor should do in the circumstances from three options offered: “take the action,” “do not take the action,” or “cannot decide.” They are then asked to rate the importance of 12 considerations relating to the particular dilemma, indicating how important each is (in their opinion) in making the decision described in the scenario using a five level scale (great importance, much importance, some, little, or no importance). The 12 statements were constructed by Rest to include considerations that would be prevalent at particular stages of ethical reasoning development in each situation. Once the 12 items have been rated, the participant is asked to select the four items that he/she considers to be of most importance to the decision and to rank these in order of importance. The first of the DIT scenarios, “Heinz and the Drug,” is set out in Appendix 1 as an example.

In scoring the DIT, weighted points are allocated to the considerations chosen as the four most important in each scenario. The points corresponding to the highest modes of ethical reasoning (stages five and six) are used to construct a single measure known as the “P” score (standing for “principled moral thinking”) for each participant (Rest 1994). Since the Rest (1979b) model is developmental and sequential, a higher P score implies a lower percentage of reasoning at lower levels. Thus, the P score measures the percentage of a participant’s thinking that is at a principled level. Based on hundreds of studies carried out in the United States, Rest and Narvaez (1998) report that junior high students generally average P scores in the 20s, senior high students in the 30s, college students in the 40s, graduate students in the 50s, and moral philosophers in the 60s.

The Need for Context-Specific Tests of Moral Reasoning

While the DIT has been described as “a broad, general measure of moral reasoning” (Fisher 1997, p. 143), concern has been expressed that it does not, and cannot, fairly represent the reasoning used when facing ethical dilemmas in a business context (see, for example, Dellaportas et al. 2006; Elm and Nichols 1993; Fraedrich et al. 1994; Trevino 1986, 1992; Weber 1990; Welton et al. 1994). There are several reasons why individuals may behave differently in a business versus a social context. Jackall (1988) suggests, for example, that what is ethically acceptable at work may not be acceptable at home or outside the corporation. He posits that the finding in many studies that managers use lower level ethical reasoning to resolve business problems is consistent with an understanding of human behavior based on cultural anthropology, where individuals play different roles that allow them to accept different values, norms, and behaviors in different life domains (e.g., home and work). This suggests that socialization at work might be expected to affect attitudes to particular issues encountered in the work context.

Becoming a professional involves adopting certain values and being induced into a vast array of both formal and informal norms, which are both taught and learned, whether consciously or not (Anderson-Gough et al. 2002, p. 41). Socialization theory leads us to expect an eventual convergence of personal values with those of the organization. Individuals are molded to fit the organizational environment, or selected and promoted on the basis that they fit into the prevailing culture, while those who do not fit in leave the organization (Douglas et al. 2001). O’Fallon and Butterfield (2005) report widespread support for the notion that ethical climates and cultures exist within organizations. While socialization has never been explicitly examined in the literature on tax practitioners and ethics, the existence of a socialization effect within the accounting profession is widely accepted (see, for example, Abdolmohammadi et al. 2003; Collins 1978; Jones and Hiltebeitel 1995; Ponemon 1992). Abdolmohammadi et al. (2003) found that the selection-socialization effect in the accounting profession results in the disproportionate hiring of individuals with a cognitive style associated with relatively low levels of ethical reasoning. They advance two reasons for this. First, individuals with a cognitive style linked to low-level moral reasoning self-select into the accounting profession and second, once there, they tend to promote those individuals with the same cognitive style as their own. While many tax practitioners do work in large accounting firms, which have been the subject of socialization research, many others work in legal practices, in industry, with a Revenue authority or as sole practitioners.

Training and professional knowledge may also have an impact, particularly in areas such as tax, where the legislation is not always clear about the action required in a particular situation, leaving a range of options for which a case can be argued (Hume et al. 1999). Awareness of this ambiguity may make it easier for, or indeed lead, practitioners to decide on the basis of other considerations (such as benefit to someone in the role of a client in the scenario, even if this implies a lower level of moral reasoning than considering society as a whole). Without detailed knowledge of the possible options that a tax practitioner might have, an inexperienced person might have to reason from first principles, leading to more weight being given to society as a whole. The business context may also affect individuals’ reasoning regardless of professional involvement. For example, in the tax context, the anonymity of those who are losers if tax is not paid could lead to perceptions that avoidance or evasion of tax is a victimless crime.

These concerns about the effect of a specific context have led to the development, in a number of areas, of instruments based on the DIT, but containing context-specific scenarios. For example, in business-related areas, there are accounting specific tests (Thorne 2000, 2001; Welton et al. 1994); broad business dilemmas and auditing scenarios (Massey 2002; Weber 1991); management vignettes (Loviscky et al. 2007); and a test of taxpayer compliance (Fisher 1997). We also use a context-specific version of the DIT, as discussed below.

DIT Studies Most Relevant to the Tax Context

While Kohlberg’s theory on moral reasoning and Rest’s DIT have not previously been used to examine tax practitioners, some work has been done in related areas, notably in the accounting domain and among taxpayers. This literature is used to motivate some of our hypotheses, while noting that tax practitioners are not always accountants (lawyers, for example, often work as tax practitioners) and that issues such as ambiguity and the “victimless crime” perception, which are present in the tax context, may be less prevalent in other domains. Indeed, it has been recognized that one of the deficiencies of ethics research in accounting has been the failure to separate the accounting profession into major functional areas when analyzing ethical issues (Marshall et al. 1998).

The DIT has been used in numerous studies to investigate the impact of different variables on the ethical reasoning of accountants, such as age (Etherington and Hill 1998; Ponemon and Gabhart 1993); gender (Etherington and Hill 1998; Eynon et al. 1997; Jones and Hiltebeitel 1995); ethical intervention3 (Armstrong 1993; Bay and Greenberg 2001; Hiltebeitel and Jones 1992; Ponemon and Gabhart 1993; Shaub 1994; St. Pierre et al. 1990); objectivity (Ponemon 1995); and locus of control (Tsui and Gul 1996). Many empirical studies suggest a lower level of moral reasoning in accounting students and accountants than would be expected, given their age and education (Armstrong 1987; Arnold and Ponemon 1991; Bernardi and Arnold 1997; Fisher and Ott 1996; Lampe and Finn 1992; Ponemon 1990, 1992; Shaub 1994; Sweeney 1995). There is a caveat to these findings, however, as comparisons are made with previously published results, rather than with comparable adults in the study context.

Work with context-specific DIT instruments sheds some light on the impact of context. Work with accounting students (Dellaportas et al. 2006; Thorne 2001) and auditors and accounting students4 (Massey 2002) indicates that participants apply a more principled level of reasoning to resolve social dilemmas than to resolve moral dilemmas in accounting or auditing. These studies, however, do not include control groups of non-specialists and thus the cause of the difference is ambiguous: is it the context, specialist knowledge, or the sort of individuals attracted to these areas? Fisher’s (1997) study of taxpayers found that they exhibited lower levels of reasoning in dilemmas in a tax paying context, but whether the impact of context is the same for tax professionals and taxpayers, given the difference in knowledge and socialization (and, indeed, the different roles they play in the context) needs clarification.

Research Questions and Hypotheses

Our research questions seek primarily to unpick the issues of the individual, the context, and professional training/socialization, as follows.
  1. 1.

    Is the level of moral reasoning of those in the tax profession generally lower than in the population as a whole (i.e., does the tax profession attract people prone to low levels of moral reasoning)?

     
  2. 2.

    Does locating an ethical dilemma in a tax context alter the way in which people reason about it?

     
  3. 3.

    Does the professional involvement of tax professionals with tax affect their reasoning in the tax context?

     
  4. 4.

    Does the socialization/training of tax professionals affect their reasoning in a tax context?

     

The literature discussed above suggests that accounting practitioners have lower levels of moral reasoning than is expected, given their age and education. Given this, our initial hypothesis for Research Question 1 will be:

H1

When considering social context ethical dilemmas, tax practitioners will employ lower levels of moral reasoning than non-specialists.

As noted above, however, comparisons in previous studies have either been based on prior research focusing on a similar professional cohort but often from a different time period and jurisdiction or on group averages compiled by the Centre for the Study of Ethical Development (CSED) in Alabama (formerly Minnesota), which were published in 1986 (Rest 1986b). Here we make use of a control group of adults in the same jurisdiction as the practitioner participants, thus allowing a more meaningful comparison, given that we have no extant information on the average P score for the population in our jurisdiction and time period. This will allow a more accurate determination of whether individuals with intrinsically different cognitive moral development from that of their population of origin are self-selecting into the tax profession.

The control group is also beneficial in addressing Research Questions 2 and 3. Although previous literature in related areas indicates lower levels of reasoning taking place in business contexts, the lack of a control group leaves unclear the issue of whether this is an effect of the context or the individual’s role in that context. A comparison of the results for the practitioner and non-specialist control groups in both contexts lets us separate the impact of professional socialization/training and context. Based on previous work with accounting professionals, our initial hypothesis will be that the lower levels of moral reasoning found in professional versus social contexts arise from the professional involvement of the practitioners rather than the context itself.

H2

When considering ethical dilemmas set in a tax context, the moral reasoning level of practitioners will be lower than it was in a social context, but there will be no difference in the levels of moral reasoning involved for non-specialists.

If a context effect is found for practitioners alone, or the context effect is different for practitioners from that for non-specialists, then we can proceed to investigate the source of the effect for professionals indicated in Research Question 4. Professionals differ in their training, tax experience, and socialization. Training and socialization can be unpicked from tax experience by separately examining tax practitioners from the private sector and practitioners working with the Revenue authority. Both categories will have experience doing tax work. Private sector tax practitioners, however, will have received professional training in tax through the Irish Tax Institute, while the Revenue practitioners are likely to have been trained internally. The socialization of these two categories of practitioner also has a different emphasis. Private sector tax practitioners will have a client focus, looking to manage (and often minimize) their clients’ tax liability. Those working for the Revenue authority, however, given its aim to collect tax to fund activities for the common good, might be expected to have a focus on issues closer to those found in the higher levels of moral reasoning in Kohlberg’s cognitive moral development model. Our hypothesis is therefore:

H3

When considering tax context ethical dilemmas, private sector tax practitioners will employ a lower level of moral reasoning than Revenue employees.

This kind of examination has not been done in the literature before now, but the make-up of the practitioner sample will facilitate this analysis.

Finally, the literature suggests that there is a need to control for certain demographic variables, which have previously been found to impact on moral reasoning. The DIT literature has identified education as being an important influence on moral reasoning. Rest and Narvaez (1994, pp. 13–14) indicate that ethical reasoning increases with age and education and suggest that 38 % of P score variation is explained by these two variables. As age and level of education increase, P scores also increase. Gilligan (1982) contended that cognitive moral development research contained a gender bias, drawing particular attention to the fact that Kohlberg’s research involved interviewing male subjects only. On the basis of 500 studies using the DIT, Rest (1986a) notes that the overwhelming majority of moral reasoning research has found no statistical difference between men and women. Where differences have been found, women are as likely to achieve higher moral reasoning scores as men. These results have been supported in later studies (see, for example, Rest and Narvaez 1994). Given these prior findings, we will consider gender, age, and education as potential control variables in the development of our analysis models, but expect that the level of moral reasoning will not differ based on gender.

Research Method

This study uses a 2 × 2 quasi-experimental design comparing the moral reasoning of tax practitioners with that of non-specialists in the context of social and tax-based ethical dilemmas. The test of reasoning in social dilemmas uses the short-form (three scenarios) DIT.5 In 1998, an updated version of the DIT was developed, named DIT-2 (Rest and Narvaez 1998; Rest et al. 1999), together with an additional scoring metric which is calculated using specific computer software designed and maintained by the CSED at the Universities of Minnesota and Alabama (Rest et al. 1997). The short form of the original DIT was used in this study primarily for comparability with other studies in this and other professional areas. The aim was also to produce social and tax dilemma scores that had the same basis to address Research Question 1 more fully by allowing a comparison of the scores across the two contexts. It would not have been possible to obtain DIT-2-based measures for the tax context, and so the use of the original instrument is more appropriate for the purposes of this study.

For the tax context, we use a tax-specific version of the DIT, the TPDIT, also with three scenarios, the development of which is described in Doyle et al. (2009). The TPDIT was developed to preserve the psychometric characteristics of the original test and to match it as closely as possible to the three scenario version of the DIT. The difference in the TPDIT, as compared with the DIT, lies in the nature of the dilemmas presented to participants and the related “items for consideration” following each dilemma, all of which are tax practice related. An example of one of the dilemmas included in the TPDIT is set out in Appendix 2.

Two counterbalanced versions of the research instrument were produced, both containing a demographic questionnaire at the end, to allow any order effects to be identified and controlled for. The order of scenarios within the DIT is not manipulated in studies, so this approach was used for the TPDIT also, but the order in which the DIT and TPDIT were presented was counterbalanced.

The research instrument was administered to 384 tax practitioners and 306 non-specialists in Ireland in 2009 using a combination of random, convenience, and snowball sampling techniques. The practitioners worked in a range of tax-related roles in Ireland, including in private practice and in the Revenue authority. The non-specialist sample had no professional involvement in taxation.

There was a 39 % response rate from tax practitioners (150 completed instruments) and a 45 % response rate from non-specialists (137 completed instruments). Following checks for full completion of the scenario-based questions and the subject reliability checks described in the DIT manual (Rest 1986b), a useable sample of 201 instruments was available for analysis (tax practitioners n = 101 and non-specialists n = 100).

Results

Overview of Samples and Preliminary Checks

As indicated above, two groups of participants took part in the study, tax practitioners and non-specialists. Demographic information on the two groups is given in Table 2. Potential relationships between age, gender, and education, and the two dependent variables (P score on the DIT, PSCOREDIT, and P score on the TPDIT, PSCORETAX) were explored using multiple regression models (see Tables 3, 4). The regression models included an indicator of the participant group (TAXPRACTITIONER, set to 1 for tax practitioners and 0 for non-specialists) and an indicator of the order of the two contexts in the instrument to control for any order effects (TAXFIRST, set to 1 if the tax based dilemmas were presented first and 0 when the social dilemmas were presented first). Education was classified in terms of whether the individual’s highest qualification was below degree level, at degree level, or at postgraduate level, and dummy variables for below degree level (EDNODEGREE) and postgraduate education (EDPG) were included in the models.
Table 2

Demographic information on participants

 

Tax practitioners

Non-specialists

Educated to degree level or above

80 %

76 %

Gender

M—44 %, F—56 %

M—39 %, F—60 %

Mean age (SD)

34.44 (9.92)

40.39 (14.14)

Mean years of tax experience (SD)

11.77 (10.31)

Range 1–38

Percentages do not add to 100 where data are missing

Table 3

Predictors of PSCOREDIT

Means for PSCOREDIT by group

Group

PSCOREDIT

Non-specialists

31.168

Tax practitioners

31.600

Of tax practitioners:

 

Revenue practitioners

28.056

Private sector practitioners

32.858

Independent variables

Coefficient

T ratio

p value

TAXPRACTITIONER

0.741

0.319

0.750

TAXFIRST

−5.513

−2.480

0.014**

Age

0.012

0.127

0.899

Gender

2.223

0.995

0.321

EDNODEGREE

−9.346

−2.973

0.003***

EDPG

0.830

0.322

0.748

Significance levels: *** .01, ** .05, * <.1

Table 4

Predictors of PSCORETAX

Means for PSCORETAX by group

Group

PSCORETAX

Non-specialists

28.522

Tax practitioners

21.266

Of tax practitioners:

 

Revenue practitioners

26.944

Private sector practitioners

19.234

Independent variables

Coefficient

T ratio

p value

TAXPRACTITIONER

−6.681

−2.995

0.003***

TAXFIRST

−4.875

−2.285

0.023**

Age

0.068

0.731

0.466

Gender

2.699

1.259

0.210

EDNODEGREE

−4.161

−1.379

0.170

EDPG

−0.262

−0.106

0.916

Significance levels: *** .01, ** .05, * <.1

These models indicated that the P scores of those without a degree were significantly lower than for graduates (the base category) for social dilemmas, although possession of a higher degree did not make a significant difference over and above the effect of a first degree. An indicator of non-graduate status was therefore carried forward into the main models as a covariate. Order effects were also identified in both regressions, and order was therefore controlled for in later analysis.

Hypothesis Testing

Research Questions 1–3 were explored using a GLM Repeated Measures analysis, with the two dependent variables captured by a within-subjects measure CONTEXT and TAXPRACTITIONER as a between-subjects measure. TAXFIRST was also included as a between-subjects measure to control for order effects and EDNODEGREE (indicating education ending below degree level) was included as a covariate. The results of this analysis are shown in Table 5.
Table 5

GLM repeated measures model looking at the impact of context and participant group (non-specialists vs. tax practitioners)

 

Type III sum of squares

df

Mean square

F

Sig.

Within-subjects effects and interactions

 CONTEXT

4848.809

1

4848.809

25.734

0.000***

 CONTEXT * EDNODEGREE

638.951

1

638.951

3.391

0.067*

 CONTEXT * TAXPRACTITIONER

1557.927

1

1557.927

8.268

0.004***

 CONTEXT * TAXFIRST

45.969

1

45.969

0.244

0.622

 CONTEXT * TAXFIRST * TAXPRACTITIONER

499.872

1

499.872

2.653

0.105

Between-subjects effects and interactions

 EDNODEGREE

2916.443

1

2916.443

11.625

0.001***

 TAXPRACTITIONER

1010.463

1

1010.463

4.028

0.046**

 TAXFIRST

2179.934

1

2179.934

8.689

0.004***

 TAXFIRST * TAXPRACTITIONER

157.116

1

157.116

0.626

0.430

Significance levels: *** .01, ** .05, * <.1

The intuition of our results is clear in the interaction graph in Fig. 1 below. CONTEXT, the within-subjects measure reflecting the two P Scores PSCOREDIT and PSCORETAX, is significant (p < 0.001) as is TAXPRACTITIONER (p < 0.05) and there is a significant interaction (p < 0.01). The marginal means suggest that the significant main effect for context is driven by the tax practitioners, giving rise to the significant interaction. Separate GLMs for practitioners and non-specialists confirmed this, with no significant effect of CONTEXT for non-specialists (p > 0.1 for non-specialists, p < 0.001 for practitioners). A MANOVA on both scores with TAXPRACTITIONER as a between-subjects effect confirmed that TAXPRACTITIONER is not significant for PSCOREDIT (p > 0.1 for PSCOREDIT, p < 0.01 for PSCORETAX).
https://static-content.springer.com/image/art%3A10.1007%2Fs10551-012-1347-x/MediaObjects/10551_2012_1347_Fig1_HTML.gif
Fig. 1

Interaction graph for PSCOREDIT and PSCORETAX for non-specialists and tax practitioners (marginal means)

Overall these results provide a coherent picture, and one which is also supported by the regression results in the preliminary checks. There is no support for the position that tax practitioners have a generally lower level of moral reasoning than the population as a whole in their area, and H1 is thus rejected. The value of having a control group, and thus comparing with ordinary people in the same geographic and temporal context, is clear here. Although previous studies have found that accounting professionals have lower levels of moral reasoning on the social dilemmas in the DIT, these have been compared with previously published results. Our average P scores in the social context are also lower than published results might lead us to expect, suggesting that geographic location and/or time period may lead to a different general level of moral reasoning in populations.

Although the results indicate an impact of context, this is driven by the tax practitioners, and our results are consistent with H2, with tax practitioners being affected by context, whereas non-specialists do not reason at a significantly different level in social as opposed to tax contexts. As predicted, the tax practitioners show a lower level of moral reasoning in their professional context.

To investigate Research Question 4, a further GLM Repeated Measures analysis was undertaken. As before, the two dependent variables were captured by a within-subjects measure CONTEXT, and TAXFIRST was included as a between-subjects measure to control for order effects, with EDNODEGREE being included as a covariate. In this model, a different between-subjects factor, PARTICIPANT, was used, which indicated whether the individual concerned was a non-specialist, a tax practitioner working for the Revenue or a tax practitioner working in the private sector (75 practitioners could be identified as private sector and 24 as working for the Revenue).6 The results of this analysis are shown in Table 6. Again there is a significant interaction between CONTEXT and the type of participant (p < 0.01). The intuition of the results is clear in the interaction graph (Fig. 2) and supports our hypothesis that private sector practitioners will employ lower levels of moral reasoning in the tax context. Indeed, the result for Revenue practitioners is much closer to non-specialists than to the other practitioners.
Table 6

GLM repeated measures model looking at the impact of context and participant group (non-specialists vs. Revenue practitioner versus private sector practitioner)

 

Type III sum of squares

df

Mean square

F

Sig.

Within-subjects effects and interactions

 CONTEXT

2520.463

1

2520.463

13.581

0.000***

 CONTEXT * EDNODEGREE

261.017

1

261.017

1.406

0.237

 CONTEXT * PARTICIPANT

2701.649

2

1350.825

7.279

0.001***

 CONTEXT * TAXFIRST

258.978

1

258.978

1.395

0.239

 CONTEXT * TAXFIRST * PARTICIPANT

604.562

2

302.281

1.629

0.199

Between-subjects effects and interactions

 EDNODEGREE

3065.364

1

3065.364

12.147

0.001***

 TAXFIRST

2313.226

1

2313.226

9.166

0.003***

 PARTICIPANT

1460.529

2

730.265

2.894

0.058*

 TAXFIRST * PARTICIPANT

264.604

2

132.302

0.524

0.593

Significance levels: *** .01, ** .05, * <.1

https://static-content.springer.com/image/art%3A10.1007%2Fs10551-012-1347-x/MediaObjects/10551_2012_1347_Fig2_HTML.gif
Fig. 2

Interaction graph for PSCOREDIT and PSCORETAX for non-specialists, Revenue Tax practitioners, and private sector tax practitioners (marginal means)

A MANOVA on both scores with PARTICIPANT as a between-subjects effect confirmed this situation. PARTICIPANT only had a significant main effect for PSCORETAX (p < 0.001). Pairwise comparisons (Bonferroni adjusted) confirmed that there was no significant difference between participant groups in terms of PSCOREDIT (> 0.1 in all comparisons), but that the private sector practitioners had significantly lower scores for PSCORETAX than the Revenue practitioners (p < 0.05) and non-specialists (p < 0.001).

Discussion

The dominant message emerging from this analysis of moral reasoning scores is that tax practitioners’ moral reasoning is less principled when they are presented with dilemmas in a tax context than when considering dilemmas in a social context. This finding is consistent with previous studies in the accounting, auditing, and management domains utilizing context-specific instruments to assess the moral reasoning of professionals (see, for example, Loviscky et al. 2007; Massey 2002; Thorne 2000). However, the use of a control group of non-specialists in this study facilitates a more comprehensive assessment of the reasoning of tax practitioners in comparison with the general public.

The fact that tax practitioners do not reason significantly differently from non-specialists in the social context suggests that individuals whose reasoning is less principled than the norm (as measured by the non-specialist control group) are not self-selecting into the tax profession. All participants reason at a similar level in the social domain, with the difference in means being below one percentage point. Although having a similar level of principled reasoning does not mean that identical principled items for consideration were selected (given the maximum score possible is around three times that achieved by the participants in this study), principled items all have a similar focus and it thus indicates similar levels of concern for more global considerations. Once the context changed to tax, however, differences in moral reasoning were evident, with tax practitioners utilizing significantly lower level moral reasoning than non-specialists who remained consistent in their reasoning across both contexts. This difference was substantial in size, with the level of principled moral reasoning being 34 % higher in non-specialists (average PSCORETAX of 28.522 vs. 21.267). This suggests that something connected with working in the tax practice environment causes principled reasoning to be used less by tax practitioners in this situation. The tax domain of itself does not cause this change, as evidenced by the fact that the reasoning of non-specialists does not alter significantly when the context changes. The P scores achieved by tax practitioners in a tax context (mean PSCORETAX 21.267) were very low compared with their P scores in a social context (mean PSCOREDIT 31.600). This may be driven by the weight tax practitioners give to legal rules in the tax context, of which non-specialists are unaware, but further analysis is needed before any such conclusions could be reached.

Further investigation of the drivers of differences in reasoning was facilitated by a comparison of private sector and Revenue practitioners. Revenue practitioners show a pattern of reasoning that is very similar to non-specialists and their reasoning is not at a significantly different level in either the social or tax contexts. On the basis that Revenue practitioners fulfill a public service role with an emphasis on collecting the maximum tax revenue in accordance with legislation, in order to fund government spending and support society as a whole, this finding is, perhaps, not surprising. The fact that Revenue practitioners reason differently from private sector practitioners, however, indicates that tax knowledge and experience are not what is driving the difference between reasoning in the social and tax contexts for practitioners, as Revenue practitioners also possess tax knowledge and years of experience working in tax. Equally, moving from a social context to a work-related context is not driving the difference, as tax is also the working domain for Revenue practitioners. The results suggest that the differences observed in the reasoning of tax practitioners in the tax domain arise only in a private practice environment. While the results do not identify the reasons for the differences in moral reasoning in a private tax practice domain, the differences found may be due to a socialization effect in private sector tax practice.

The P scores that emerged from this analysis are interesting in themselves. The mean DIT P scores of all participant groups in this study, as shown in Table 3, range from 28.1 for Revenue practitioners to 32.9 for private sector practitioners. These compare quite poorly with the average scores reported by Rest (1986b, p. ii) on the basis of the norms complied by the CSED in Alabama. These figures are set out in Table 7 below.
Table 7

Mean DIT P scores—Rest (1986b, p. ii)

Group

Mean DIT P score

Moral philosophy and political science doctoral students

65.2

Seminarians in a liberal protestant seminary

59.8

Advanced law students

52.2

Practicing medical physicians

49.5

Average college student

42.3

Average of adults in general

40.0

Average senior high student

31.8

Average junior high student

21.9

Institutionalized delinquent boys, 16 years old

18.9

According to Table 7, the scores from this study are most comparable with those of average senior high students and are well below the level of adults in general and college students. These scores are also much lower than the average P scores of accountants found in other studies which range from 38.1 (Ponemon 1992) to 41.3 (Shaub 1994). The low scores found in this study do suggest that, when studying the impact of work contexts on moral reasoning, account needs to be taken of differences in levels of moral reasoning in relation to social dilemmas across populations. Using scores compiled by the CSED, over hundreds of studies in different jurisdictions and spanning a long period in time, may not be so relevant for this sort of investigation. Many DIT studies date from the 1970s and are based on research undertaken in the United States. In an accounting and tax context, they also pre-date the greater awareness of ethical issues following the Enron and KPMG scandals. It may be unwise to assume that average P scores calculated in the 1970s would be replicated today, as moral standards, especially in financial/business areas, may have changed in the intervening years (see, for example, Cole and Smith 1996; Cordeiro 2003; Glen and Loo 1993).

The results did not reveal any significant differences in the moral reasoning scores of males and females. This finding is consistent with the prior literature in accounting, most of which finds no statistically significant difference in moral reasoning between the sexes (Abdolmohammadi et al. 2003; Ponemon 1990, 1992; Tsui 1996).

Age also does not appear to effect moral reasoning scores in this study. This finding is consistent with Shaub (1994) but not with many other studies that have examined age (e.g., Etherington and Hill 1998; Eynon et al. 1997; Jones and Hiltebeitel 1995; Ponemon 1990).

Conclusions

The aim of this article was to examine the moral reasoning of tax practitioners in both a work and a social domain using a control group of non-specialists. The use of moral reasoning measures has never previously been a focus of attention in the literature on the tax profession, despite the fact that the theory of cognitive moral development and the DIT have been used extensively to examine accountants and, to a more limited extent, lawyers (Earnest and Bebeau 2000; Scofield 1997; Scofield et al. 2004). The design of the study allowed the context in which dilemmas originate to be unpicked from the nature of individuals who enter the tax profession and from the training/socialization of professionals in the area. The inclusion of a control group of non-specialists in the design is another unique contribution of the article, both from the viewpoint of tax research and of context-specific DIT research, allowing this level of unpacking of effects in the analysis.

In that tax practitioners do not use levels of moral reasoning that are statistically significantly different from those used by non-specialists in a social context, the results provide some assurance that individuals who utilize a lower level of moral reasoning than might be considered a societal norm are not self-selecting into the tax profession. However, once the context changes to tax, the moral reasoning of practitioners in private practice declines significantly from that used in a social context and from that used by non-specialists. Interestingly, Revenue practitioners, for whom tax is also a work-related context and who possess technical tax knowledge and experience, do not display the same change in reasoning as private sector practitioners: their tax context reasoning remains consistent with their social reasoning. This suggests that it is the private sector tax practice environment (i.e., a client-oriented one) that drives the difference in tax and social context moral reasoning in tax practitioners. These findings therefore support the notion of a socialization effect in tax practice, a concept widely accepted in the accounting domain but not previously examined separately for tax.

In addition to making a contribution to the academic literature, this research informs several other areas. The knowledge that tax practitioners use much lower levels of moral reasoning in a work context as opposed to a social one should be considered in the design of both academic and professional training/education programs. Rest (1986a) contends that ethical reasoning might be enhanced through certain forms of instruction, and many studies have found that the completion of ethics courses in college is associated with higher level moral reasoning in accountants or accounting students as measured using the DIT (Armstrong 1993; Eynon et al. 1997; Shaub 1994; Sweeney 1995). Our results, however, suggest that ethics training in social contexts may not have a comparable impact on work-related moral reasoning. Rather we suggest that, to be most effective, training programs should incorporate formal ethics training that uses specific tax-related scenarios, as put forward by Bobek and Radtke (2007). The use of the focal context in professional ethics training gives the best potential for challenging socialized responses, perhaps by comparison with similar social scenarios where professionals may find themselves reasoning differently. The tax profession is governed by relevant professional institutes, and such bodies should also take account of the dichotomy in reasoning between contexts when considering how to promote higher levels of moral reasoning in the work environment.

It is common practice for policy makers, professions, and even corporate entities to develop codes of conduct or guidelines as to best practice in particular areas. Codes/guidelines can either be principles based or rules based. Knowing on which basis the individuals concerned approach ethical dilemmas may be useful to the relevant drafting bodies by indicating how best to encourage compliance with codes of conduct or best practice guidelines. Given that tax practitioners appear to reason using a lower level of moral reasoning in a work context, codes or guidelines encompassing prescriptive procedures may be more effective than principles-based guidelines. The more pertinent question that arises, perhaps, is whether it is better for policy makers and/or the profession to support the development of principled thinkers who rely on their own personal reasoning processes to dictate their behavior, or whether the profession would be better served by rule-oriented practitioners who are guided by the profession’s detailed instructions in handling ethical issues. Further research on how principled thinking might be promoted would be needed to support the former approach.

To summarize, this article makes a contribution not only to academic knowledge and debate but also informs policy makers, educators, and the tax profession as represented by the professional institutes.

Footnotes
1

Ireland is a common law jurisdiction, so the results of this study are inherently relevant and applicable.

 
2

The terms “morality” and “ethics” are used interchangeably in the literature on the psychology of moral reasoning (Rest 1994), and we will follow this practice throughout this article. Various authors have proposed distinctions, but there does not seem to be one, generally accepted distinction.

 
3

Intervention studies involve various kinds of ethics courses designed to enhance moral reasoning in participants. Typically participants are tested pre- and post-intervention to assess the effectiveness of the course.

 
4

Proxying entry-level auditors.

 
5

Permission to use the DIT in this study was obtained from the Centre for the Study of Ethical Development in Alabama (formerly Minnesota).

 
6

The remaining two practitioners were academics and were excluded from this analysis.

 
7

The Heinz scenario has been slightly altered from the original Rest (1986b) version in order to update the language slightly for the Irish jurisdiction context used in this study. The original dollar figure mentioned in the scenario has been changed to Euros and the word “druggist” has been replaced by “pharmacist.”

 

Acknowledgments

The authors gratefully acknowledge the helpful comments from anonymous reviewers and delegates on earlier versions of the article at the following conferences: the American Accounting Association Ethics Symposium, 2010, San Francisco; the American Accounting Association Northeast Region Meeting, 2009, Cambridge, Massachusetts; the Society for Judgement & Decision Making Annual Meeting, 2009, Boston, Massachusetts; the IRS Research Conference 2009, Washington DC; and the Tax Research Network Conference, 2009, Cardiff, Wales.

Copyright information

© Springer Science+Business Media B.V. 2012