An egocentric model of the relations among the opportunity

Description
A model of the relations among taxpayers’ opportunity, social norms, ethical beliefs, and tax compliance is proposed
and tested using structural equation modeling. High opportunity taxpayers, who may personally benefit from evasion,
judged evasion as less unethical than low opportunity taxpayers. High and low opportunity taxpayers judged social
norms similarly. Further, ethical beliefs partially (fully) mediate the relation between opportunity (social norms) and
underreporting.

An egocentric model of the relations among the opportunity
to underreport, social norms, ethical beliefs,
and underreporting behavior
Cindy Blanthorne
a,
*
, Steven Kaplan
b
a
University of Rhode Island, Department of Accounting, 7 Lippett Road, Kingston, RI 02881, United States
b
Arizona State University, School of Accountancy, W.P. Carey School of Business, P.O. Box 873606, Tempe,
AZ 85287-3606, United States
Abstract
A model of the relations among taxpayers’ opportunity, social norms, ethical beliefs, and tax compliance is proposed
and tested using structural equation modeling. High opportunity taxpayers, who may personally bene?t from evasion,
judged evasion as less unethical than low opportunity taxpayers. High and low opportunity taxpayers judged social
norms similarly. Further, ethical beliefs partially (fully) mediate the relation between opportunity (social norms) and
underreporting. Implications from our study to tax compliance researchers and policy makers are discussed.
Ó 2008 Elsevier Ltd. All rights reserved.
Introduction
The US income tax system has been character-
ized as a voluntary reporting system and the costs
of noncompliance with the federal income tax laws
in the United States are recognized as sizable. Esti-
mates of the tax gap exceed $345 billion per year
(IRS, 2006). Within this voluntary reporting sys-
tem, the income for many individuals is highly vis-
ible to the IRS (e.g., third party reporting to the
IRS) such that there is little opportunity to e?ec-
tively evade income taxes. Kagan (1989) observed
that compliance is extremely high on those items
that are reported to both the taxpayer and to the
IRS. Not surprisingly, previous research generally
has found that noncompliance occurs to a greater
extent when the opportunity to evade is high
(Antonides & Robben, 1995; Carnes & Englebr-
echt, 1995; Madeo, Schepanski, & Uecker, 1987;
Smith, 1990; Young, 1994). The importance of
opportunity is grounded in economic models (All-
ingham & Sandmo, 1972), which are premised on
taxpayer self-interest. Generally, under an economic
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.02.001
*
Corresponding author. Tel.: +1 401 874 4363; fax: +1 401
873 4312.
E-mail addresses: [email protected] (C. Blanthorne),
[email protected] (S. Kaplan).
Available online at www.sciencedirect.com
Accounting, Organizations and Society 33 (2008) 684–703
www.elsevier.com/locate/aos
model of tax evasion, opportunity to evade is a
proxy for the probability of detection. Consistent
with an economic model, evasion is positively
associated with opportunity (Smith, 1990).
Researchers have generally concluded that eco-
nomic models focusing on opportunity and penal-
ties provide only a partial understanding of tax
compliance choices made by taxpayers and that
social norms and ethical beliefs also play a role
(Alm, McClelland, & Schultze, 1992; Alm, McC-
lelland, & Schultze, 1999; Jackson & Milliron,
1986; Wenzel, 2005b). However, research explor-
ing relations among taxpayers’ economic variables
such as opportunity to evade, their social norms
and ethical beliefs about evasion, and their tax
compliance intentions and behaviors has been lim-
ited. Generally, previous research has character-
ized the relations between ethics and deterrence
variables such as opportunity in one of two ways
(Wenzel, 2004). Under one view, economic and
ethical beliefs/social norms are considered to rep-
resent independent in?uences on taxpayers. That
is, the in?uence of each is grounded in an indepen-
dent process (Wenzel, 2004), and consequently,
implicitly assumes no relation between opportu-
nity and ethical beliefs/social norms. This view is
re?ected in tax compliance research that only
examines either economically based or social/ethi-
cal beliefs on tax compliance or considers both but
treats each as an independent in?uence (e.g.,
manipulates economically based variables and
measures and treats social norms and ethical
beliefs as covariates). Within this view, research
has commonly found that the inhibiting e?ects
from economic variables are smaller than from
ethical beliefs (Wenzel, 2004).
Alternatively, ethical beliefs (and presumably
social norms as well) have been characterized as rep-
resenting a screening mechanismin which economic
variables such as opportunity are only considered
when one does not consider evasion to be highly
unethical (or social norms encourage evasion). This
view, based on Etzioni (1988, 1993), in e?ect, ‘‘dis-
cards” evasionas a behavioral option when the indi-
vidual considers evasionas a highly unethical act. In
this regard, Alm (1991, p. 584), states, ‘‘some indi-
viduals pay taxes because they believe that cheating
is wrong”. Under this view, economic consider-
ations only matter for taxpayers who do not con-
sider evasion to be highly unethical. Consistent
with this view, Reckers, Sanders, and Roark
(1994) report that withholding position (e.g., over-
withhold or tax due) ‘‘hardly in?uenced” (Reckers
et al., 1994, p. 832) taxpayers who strongly agreed
that tax evasionwas morally wrong but had a signif-
icant impact among taxpayers holding opposing
ethical beliefs. However, under this view, taxpayers’
ethical beliefs are implicitly assumed to arise inde-
pendently from the economic constraints facing
taxpayers. Thus, under both views, ethical beliefs
and social norms are implicitly assumed to arise
independently from how taxpayers are economi-
cally situated.
Our study contributes to the taxpayer compli-
ance literature by proposing a new model of the
links among opportunity to evade, social norms
about evasion, ethical beliefs about evasion, and
tax compliance behavior/intentions. Central to
our model is the proposition that taxpayers’ ethical
beliefs do not arise independently from economic
considerations, but instead, are in?uenced by tax-
payers’ opportunity to evade. Our model builds
upon prior research that has found in other con-
texts that self-interest in?uences one’s ethical beliefs
(Babcock, Loewenstein, Issacharo?, & Camerer,
1995; Thompson &Loewenstein, 1992). Thompson
and Loewenstein (1992) refer to this tendency as an
egocentric interpretation or bias and it occurs when
individuals’ ethically related judgments ‘‘will be
biased in a manner that favors themselves”
(Thompson & Loewenstein, 1992, p. 177).
Speci?cally, our model proposes that high
opportunity taxpayers, who are able to personally
bene?t from tax evasion behavior, are expected to
judge tax evasion as less unethical than low oppor-
tunity taxpayers, who are less able to personally
bene?t from evasion behavior. Thus, under our
model, taxpayers with a high opportunity to evade
di?er from taxpayers with a low opportunity to
evade in several important respects. First, from an
economic standpoint, the expected values of eva-
sion behavior are much higher for high opportunity
taxpayers than low opportunity taxpayers. Second,
high opportunity taxpayers are expected to viewtax
evasion less unethically than low opportunity
taxpayers. Thus, ethical beliefs that have the capac-
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 685
ity to inhibit taxpayers from engaging in evasion
(e.g., the discarding e?ect) are less likely to do so
among high opportunity taxpayers. Third, we
examine whether egocentric interpretations extend
into judgments about social norms. Cialdini and
Trost (1998, p. 152), de?ne social norms as ‘‘rules
and standards that are understood by members of
a group, and that guide and/or constrain social
behavior without the force of law”. Previous
research has not considered whether social norms
are interpreted egocentrically. Our model proposes
that taxpayers also construct social norms in a self-
serving fashion. Social norms, in turn, along with
opportunity, are expected to in?uence taxpayers’
ethical beliefs. Lastly, ethical beliefs are expected
to directly in?uence tax compliance intentions and
behavior. Under our model, the e?ect of opportu-
nity on tax compliance intentions and behavior
may be indirect through ethical beliefs or there
may also be a direct relationship between opportu-
nity and tax compliance intentions and behavior.
Our model is tested in the context of underre-
porting taxable income
1
and as such we examine
taxpayers’ opportunity to underreport, taxpayers’
social norms and ethical beliefs about underreport-
ing, and intentions and behavior related to und-
erreporting. The model is tested using responses
from taxpayers to a survey. Taxpayers are assigned
into a high or low opportunity to evade group
based on their survey responses. Survey responses
also included information regarding taxpayers’
social norms about underreporting, ethical beliefs
about underreporting, underreporting behavior,
intentions to underreport given hypothetical sce-
narios, and demographic information. Structural
equation modeling (SEM) is used to analyze
responses and test the proposed tax compliance
model. SEM o?ers two key advantages to regres-
sion based forms of analysis. First, several of the
underlying constructs in tax compliance models,
such as ethical beliefs and social norms, are inher-
ently latent (unobservable) variables. SEM pro-
vides a theoretically error-free measure of latent
variables (Hatcher, 1994). Second, SEM simulta-
neously examines the relations among a set of vari-
ables, which is particularly important in models
including mediating variables. Mediating variables,
such as ethical beliefs, enhance our understanding
of ‘‘how” or ‘‘why” a predictor variable (e.g., tax-
payer opportunity) is related to a dependent vari-
able (e.g., taxpayer underreporting behavior/
intentions) (Baron & Kenny, 1986).
The following section provides background for
the study. This section also presents the proposed
model and related hypotheses. Subsequent sections
present the research methodology, the results, and
a discussion of the results.
Proposed tax compliance model and hypothesis
development
Fig. 1 presents our proposed tax compliance
model. Our study extends economic models (All-
ingham & Sandmo, 1972) and previous research
(Antonides & Robben, 1995; Carnes & Englebr-
echt, 1995; Henderson & Kaplan, 2005; Madeo
et al., 1987; Young, 1994) by demonstrating a rela-
tionship between noncompliance opportunity and
compliance behavior. Furthermore, while tax com-
pliance researchers recognize that social norms and
ethical beliefs are likely to play a role in taxpayers’
reporting decisions (Wenzel, 2005a), our model
develops speci?c hypotheses about the relations
among opportunity, social norms, ethical beliefs,
and underreporting intentions and behavior.
Fundamentally, our model applies the notion of
egocentric interpretations of ethics to the context
of tax compliance. Research in psychology (Epley
& Caruso, 2004; Messick & Sentis, 1979), organi-
zational behavior (Neale & Bazerman, 1983;
Thompson & Loewenstein, 1992), and accounting
(Kachelmeier & Towry, 2002; Kaplan, 2001; Luft
& Libby, 1997), demonstrates that individuals tend
to make egocentric interpretations of fairness and
ethics. In situations such as tax compliance where
consensus on acceptable behavior may be lacking,
multiple interpretations of ethical actions are likely
to arise. In this regard, Thompson and Loewen-
stein (1992) contend that interpretations will be
self-serving. That is, individuals who bene?t from
1
To simplify the discussion, unless indicated di?erently, in
the remainder of the manuscript, we use the phrase ‘‘underre-
porting” or ‘‘underreporting income” to mean ‘‘underreporting
taxable income”.
686 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
an unethical or questionable act will not assess the
act as negatively as other individuals who do not
bene?t from the act.
In an early example, Messick and Sentis (1979)
examine self-serving interpretations of fairness in a
work setting. The study presented a scenario in
which a worker had spent 7 hours on a job and
another worker had spent 10 hours on a job. Some
of the participants were assigned the role of the
7 hours worker and the remaining participants were
assignedthe role of the 10 hours worker. Inall cases,
the 7 hours worker was paid $25 andin all cases par-
ticipants were asked to indicate a fair payment for
the 10 hours worker. The mean payment from par-
ticipants assigned to the role of the 7 hours worker
was $30.29 whereas the mean payment from partic-
ipants assigned to the role of the 10 hours worker
was $35.24. That is, in setting a fair wage, partici-
pants set the wage higher when the wage was being
set for oneself than for the other worker. This evi-
dence is consistent with self-serving fairness beliefs.
More recently, Luft and Libby (1997) found that
participants set transfer prices egocentrically. Spe-
ci?cally, participants in the role of a selling division
set higher transfer prices compared to the transfer
prices set by participants in the role of buying divi-
sion. Further, the tendency to set transfer prices in
an egocentric fashion was greater when the use of
market prices did not result in an equal pro?t split
between the two division managers compared to
when the use of market prices did result in an equal
pro?t split. Kachelmeier and Towry (2002) extend
Luft and Libby (1997) by considering both judg-
ments about expected transfer prices as well as the
results from actual rounds of negotiation. Kac-
helmeier and Towry (2002) replicate Luft and Lib-
by’s ?ndings with respect to expected transfer prices
and ?nd that the negotiated transfer price was sig-
ni?cantly di?erent from one based exclusively on
market prices when use of the market price did
not result in an equal pro?t split between the two.
Neither of these studies, however, explicitly exam-
ined ethical or fairness beliefs. Further, because
tax compliance is a voluntary system, tax reporting
does not involve negotiations between parties.
Now consider two taxpayers, one with high
opportunity to underreport income and one with
low opportunity to underreport income. As dis-
cussed above, and consistent with economic theory,
the expected value for underreporting is greater for
Measured Variable
Latent Construct
Self
Underreporting
Behavior
H 3 & 5
Opportunity
Hypothetical
Underreporting
Intention
Underreporting
Ethics
Social
Norms
H 6
H 1
H 5 H 2 & 4
H 4
H 7
H 7 & 8
Fig. 1. Hypothesized model of tax compliance.
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 687
the high opportunity taxpayer than for the low
opportunity taxpayer. Given the ethical ambiguity
of tax evasion, taxpayers are likely to form ethical
beliefs about underreporting egocentrically. The
low opportunity taxpayer has little chance of
directly bene?ting from underreporting. Conse-
quently, such a taxpayer has no (economic) self-
serving interest in the formation of their ethical
beliefs about tax compliance.
2
Alternatively, consider an egocentric ethical
interpretation by a taxpayer with a high opportu-
nity to underreport income. Because the probabil-
ity of detection is very low, this taxpayer will
personally bene?t from underreporting. However,
Bazerman, Curhan, Moore, and Valley (2000, p.
292), state, ‘‘People are motivated to think of
themselves as ethical”. Egocentric interpretations
provide a mechanism in which the taxpayer can
take advantage of the high opportunity to under-
report, without sacri?cing one’s self-image. That
is, by judging tax evasion as less unethical, ethical
beliefs become less of a barrier to underreporting.
Thus, the economic bene?t from tax evasion is
expected to induce the taxpayer to form an ethical
belief about underreporting that is favorable
towards maintaining one’s self-esteem. This dis-
cussion leads to the following hypothesis:
Hypothesis one: Taxpayers with high opportu-
nity to underreport income will judge underre-
porting to be less unethical than taxpayers
with low opportunity to underreport income.
Previous research (Bobek & Hat?eld, 2003;
Erard & Feinstein, 1994; Ghosh & Crain, 1996;
Henderson & Kaplan, 2005; Kaplan, Newberry,
& Reckers, 1997; Kaplan, Reckers, & Roark,
1988; Wenzel, 2005a) has consistently found a neg-
ative association between ethical beliefs and eva-
sion behavior. That is, as suggested above,
stronger beliefs that tax evasion is unethical func-
tion as an inhibitor to underreporting behavior. As
discussed below, underreporting behavior is based
on a self-report of past behavior, and conse-
quently, we label this as self-underreporting
behavior. This discussion leads to the following
hypothesis, which may be viewed as an attempt
to replicate a common ?nding:
Hypothesis two: Beliefs that underreporting
income is unethical will be negatively associated
with self-underreporting behavior.
As proposed above in hypotheses one and two,
taxpayer opportunity is expected to directly in?u-
ence ethical beliefs over underreporting, which in
turn, are expected to directly in?uence underreport-
ing decisions. Together these two hypotheses imply
that ethical beliefs mediate the relationship between
taxpayer opportunity and underreporting behav-
ior.
3
Baron and Kenny (1986, p. 1176) explain a
mediator variable as follows: ‘‘In general, a given
variable may be said to function as a mediator to
the extent that it accounts for the relation between
the predictor and the criterion.” Mediation is used
to establish ‘‘how” or ‘‘why” the predictor variable
is relatedtoanoutcome or criterionvariable (Frazier
et al., 2004) and mediator variables ‘‘explain how
external physical events take on internal psycholog-
ical signi?cance” (Baron & Kenny, 1986, p. 1176).
In our context, taxpayer opportunity represents
a predictor variable, underreporting behavior rep-
resents the criterion and taxpayer ethical beliefs
about underreporting are proposed as a mediator
to test why opportunity is related to underreport-
ing behavior. Baron and Kenny (1986) also note
that a mediator variable may fully or partially
mediate the relation between a predictor and the
criterion. Generally, a mediator variable is a full
mediator when the direct path from the predictor
to the criterion is insigni?cant when the mediator
variable is included in the model. Alternatively, a
mediator variable is a partial mediator when the
direct path from the predictor to the criterion is
reduced, but remains signi?cant, when the
2
It is also possible that for low opportunity taxpayers, who
are generally unable to underreport income, to in?ate or
exaggerate the extent to which underreporting income is judged
to be unethical. We thank an anonymous reviewer for bringing
this point to our attention.
3
The terms mediated e?ect (primarily used in psychology
literature) and indirect e?ect (primarily used in sociology
literature) are typically used interchangeably (Frazier, Tix, &
Barron, 2004).
688 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
mediator is included in the model. We expect that
ethical beliefs will partially mediate the relation-
ship because the expected value of underreporting
remains greater for high opportunity taxpayers
compared to low opportunity taxpayers. This dis-
cussion leads to the following hypothesis:
Hypothesis three: Controlling for ethical beliefs,
opportunity will have a signi?cant direct rela-
tionship on taxpayers’ self-underreporting
behavior.
Fig. 1 also includes taxpayers’ underreporting
intentions in response to hypothetical underreport-
ing scenarios. These scenarios are constructed such
that opportunity to underreport is high for the
hypothetical income described. By design, the
hypothetical scenarios place all taxpayers in a high
opportunity condition for the income in the sce-
nario. Comparing underreporting intentions in
response to hypothetical scenarios, relative to tax-
payers’ self-underreporting behavior, is intended
to further inform the proposed model. As pro-
posed in hypothesis four, we expect that ethical
beliefs about underreporting will have a similar
relation with underreporting intentions based
upon a hypothetical scenario and with self-und-
erreporting behavior. As proposed in hypothesis
?ve, we expect the relationship between opportu-
nity (in fact) and underreporting intentions based
upon a hypothetical scenario to be signi?cantly
weaker than the relationship between opportunity
(in fact) and self-underreporting behavior. This
discussion leads to the following hypotheses:
Hypothesis four: Beliefs that underreporting
income is unethical will be negatively associated
with underreporting intentions, and equivalent
to the association between beliefs that underre-
porting income is unethical and self-underre-
porting behavior.
Hypothesis ?ve: Controlling for ethical beliefs,
the direct relationship between opportunity
and reporting intentions based upon a hypo-
thetical scenario will be signi?cantly weaker
than the direct relationship between opportu-
nity and self-underreporting behavior.
Social norms are generally considered to repre-
sent a potentially important factor in tax compli-
ance decisions (Alm et al., 1999; Bobek, Roberts,
& Sweeney, 2007; Wenzel, 2004). Social norms
are de?ned as ‘‘rules and standards that are under-
stood by members of a group, and that guide and/
or constrain social behavior without the force of
law” (Cialdini & Trost, 1998, p. 152). In this study,
we examine whether social norms are formed in an
egocentric fashion. If so, social norms discourag-
ing underreporting income may function as an
inhibitor to underreporting; this could be costly
for a high opportunity taxpayer. However, in con-
trast to ethical beliefs, which are internal to the
individual, social norms are externally grounded,
and as such, may be less pliable or subjectively
formed than ethical beliefs. This discussion leads
to the following hypothesis:
Hypothesis six: Taxpayers with high opportu-
nity to underreport income will judge social
norms for underreporting to be less unethical
than taxpayers with low opportunity to under-
report income.
Research has generally found an association
between social norms andtax compliance intentions
and behavior (Alm, 1991; Alm & Sanchez, 1995;
Bobek & Hat?eld, 2003; Davis, Hecht, & Perkins,
2003; Hanno &Violette, 1996; Wenzel, 2004). How-
ever, previous research has generally considered
tests of association without controlling for ethical
beliefs (Wenzel, 2005b). Such ?ndings are consistent
with the importance of social norms emerging from
either or both an ethical and/or economic perspec-
tive. First, froman ethical perspective, social norms
relate toone of the components of Jones (1991) issue
contingent model of ethical decision making, social
consensus. Under Jones’ (1991) model, increases in
the social consensus about the ethical aspects of a
behavior reduce the moral ambiguity related to
the behavior.
Consequently, as proposed by hypothesis seven,
individuals are less likely to engage in behavior
when the social consensus is strong regarding the
unethical nature of that behavior. Implicitly, under
this view, social norms shape and inform one’s
ethical beliefs. That is, to some extent, social norms
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 689
are internalized, which will be re?ected in one’s eth-
ical beliefs (Kelman, 1958; Wenzel, 2004), which in
turn, will in?uence compliance behavior.
Secondly, social norms for underreporting also
may be viewed from an economic perspective
(Alm & Sanchez, 1995; Pommerehne, Hart, &
Frey, 1994; Scholz & Pinney, 1995; Wenzel,
2004). Imagine engaging in a behavior for which
the social norm (e.g., the social consensus) consid-
ers the behavior to be inappropriate or wrong.
Under an economic perspective, the community
members impose economic costs on the individual
who engages in the inappropriate behavior. For
example, the individual’s reputation may be dam-
aged, and consequently, the individual’s economic
future jeopardized. Under an economic perspec-
tive, social norms, by imposing costs, directly
in?uence taxpayers’ compliance decisions and are
not linked to the individual’s ethical beliefs.
Implicitly, under this economic view, individuals
are judging social norms in a reasonably objective
fashion and believe that community members
become aware of a taxpayer’s noncompliance. Pre-
sumably, taxpayers may be able to avoid any neg-
ative consequences triggered by noncompliance by
hiding their behavior from community members.
This discussion leads to the following two compet-
ing hypotheses regarding the relationship between
social norms and self-underreporting. The ?rst is
based on an ethical perspective, whereas the sec-
ond is based on an economic perspective.
Hypothesis seven: Social norms will have a sig-
ni?cant direct relationship with beliefs that und-
erreporting income is unethical. Controlling for
ethical beliefs, the direct relationship between
social norms and self-underreporting behavior
will not be signi?cant.
Hypothesis eight: Social norms will have a sig-
ni?cant direct relationship with self-underre-
porting behavior.
Research methodology
A survey was distributed to taxpayers. We
describe the participants, including how they were
classi?ed into the high or low opportunity condi-
tion, and the administration of the survey below.
While the survey contained a broad range of mea-
sures, including demographic measures, we pri-
marily focus on measures of ethics and social
norms about underreporting and measures of
actual and hypothetical underreporting.
Participants, opportunity, and administration
The researchers approached nine di?erent
groups, described below, and requested individuals
to complete the survey. As shown in Table 1, a
total of 456 survey instruments were distributed
by one of the researchers. Approximately 85%
returned the survey to the researcher. Generally,
the survey was completed it in approximately
20 min. Of those returned, another approximately
7% were incomplete. Thus, the analysis is based on
the 355 usable surveys and represents a response
rate of 78% of the distributed surveys.
Participants were classi?ed into either the high
opportunity or low opportunity condition based
on their responses to two of the items on the sur-
vey. One item asked about business ownership
and a second item asked about receiving cash
income that was not reported to the IRS. Partic-
ipants were placed into the high opportunity con-
dition if they indicated business ownership (with
at least some cash sourced income) or if they indi-
cated receiving any cash income that was not
reported to the IRS. All remaining participants
were placed in the low opportunity category.
Based on the responses to these two items, 160
participants were categorized as high opportunity
and 195 participants were categorized as low
opportunity.
4
Nine di?erent groups were approached to par-
ticipate in the study. Certain groups, by design,
were selected because they were expected to be
composed of mainly high opportunity taxpayers.
For example, participants from the swap meets
4
Of the 195 respondents with opportunity: 157 were small
business owners (with cash sourced income) and 38 additional
people who received cash sourced income that were not small
business owners.
690 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
were vendors.
5
and were expected to be mainly
high opportunity taxpayers. These subjects are
classi?ed by the IRS as informal suppliers. Tax-
payers classi?ed as informal suppliers have been
found to underreport a substantial amount of their
income (IRS, 1996; IRS, 2005a). As expected, and
as shown in Table 1, the majority of swap meet
vendors were categorized as high opportunity tax-
payers. Certain other groups, such as the driving
school, were selected because the majority of the
participants were expected to be low opportunity
taxpayers. This was also found to be the case.
For each of the nine groups, Table 1 presents
information about the surveys distributed, surveys
completed, overall usable surveys, and usable sur-
veys by opportunity category.
Table 2 presents information about partici-
pants’ socioeconomic backgrounds for each
opportunity condition. As shown, participants
with high opportunity tended to be older, less edu-
cated, and employed a paid tax preparer less often.
The two opportunity groups were similar in terms
of gender and income. Across di?erent studies,
each of these measures has been associated with
various tax compliance measures (Fischer, War-
tick, & Mark, 1992; Jackson & Milliron, 1986)
and thus, will be evaluated as control variables.
Social norms
To measure social norms, subjects indicated the
extent to which each of ?ve di?erent social referents
represented a source of relative encouragement or
discouragement with respect to underreporting
income. The following item was used to elicit social
norms: ‘‘My ________ has the following type of
e?ect on whether or not I would underreport
income on my income tax return”. Subjects com-
pleted the item for the following ?ve referent
groups: spouse or signi?cant other, family, tax
return preparer, friends, and business contact/
peers.
6
Subjects responded using a seven-point
Table 1
Surveys by source and opportunity condition
a
Source Surveys
distributed
Surveys
collected
Surveys
usable
Without
opportunity
With
opportunity
Swap meet
b
#1 120 88 74 9 65
Swap meet
b
#2 63 52 44 9 35
MBA class entrepreneurship 68 65 65 53 12
State Driving School 107 98 93 58 35
City Kiwanis Club 33 29 24 14 10
Software training class (commercial) 20 18 17 8 9
Restaurant 20 16 16 4 12
XXX, Attorney at Law 15 13 13 1 12
Business network chapter meeting 10 9 9 4 5
Totals 456 388 355 160 195
Response rate (%) 100 85 78
Distribution (%) 100 45 55
a
Researcher attended meetings or went to places of business to administer surveys to respondents.
b
A swap meet, also known as a ?ea market, is a place where vendors come to sell their goods. Most ?ea markets sell second-hand
goods. However, newer items are sometimes found in larger, urban-area ?ea markets (Wikipedia, 2007). Almost all of the vendors
surveyed sold exclusively new products.
5
A swap meet, also known as a ?ea market, is a place where
vendors come to sell their goods. Most ?ea markets sell second-
hand goods. However, newer items are sometimes found in
larger, urban-area ?ea markets (Wikipedia, 2007). Almost all of
the vendors surveyed sold exclusively new products.
6
Family, friends and spouse were common referents from
prior tax compliance literature (see Hanno & Violette, 1996).
Tax return preparer was included because it is estimated that
over 60% of US income tax returns are professionally prepared
(IRS, 2005b). Business contacts was included because many of
the intended subjects were informal vendors and perhaps likely
to share tax reporting strategies with other informal business
owners.
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 691
Likert scale ranging from ‘‘1” (discourages und-
erreporting) to ‘‘7” (encourages underreporting).
The mid-point of the scale, 4, was labeled ‘‘No
e?ect”.
7
To facilitate interpretation, responses were
reverse coded.
8
By reversing the scores, lower val-
ues are indicative of social norms encouraging und-
erreporting, which, as discussed below, is consistent
with the values of underreporting ethics (e.g., lower
values re?ect lower underreporting ethics).
Underreporting ethics
To measure ethical beliefs about underreporting
income, subjects were asked to indicate their level
of agreement with ?ve statements about underre-
porting. The ?ve statements relate to (1) not feel-
ing guilty if underreport; (2) underreporting goes
against principles; (3) morally wrong to underre-
port; (3) not ethically wrong to underreport; and
(5) dishonest to underreport.
9
For each of the ?ve
statements a seven-point Likert scale was used
ranging from ‘‘1” (de?nitely disagree) to ‘‘7” (def-
initely agree). Two of the items, not feeling guilty
and a judgment that it is not ethically wrong to
underreport, were reverse coded. Thus, higher val-
ues indicate that underreporting is judged to be
more unethical, and lower values indicate that
underreporting is judged to be less unethical. We
selected a direct measure of underreporting ethics,
rather than a general measure of ethical orienta-
tion, based on the tendency of ethical behavior
to be situation-speci?c (Arrington & Reckers,
1985; Haan, 1975).
Self-underreporting behavior
Two measures were used to capture participants’
underreporting behavior. The ?rst itemfocuses spe-
ci?cally on the prior year’s tax return and asked,
‘‘how much of the income that you believed was tax-
able did you include on your tax return?” A seven-
point Likert scale was used ranging from ‘‘1”
(included all) to ‘‘7” (included none). The second
measure focuses on past years, and asked, ‘‘approx-
imately how often do you think you underreported
your income on your tax return?” Again, a seven-
point Likert scale was used ranging from ‘‘1”
(never) to ‘‘7” (always). By design, neither question
sought speci?c amounts or frequencies. While it is
possible that such measures may be somewhat
noisy due to individual di?erences in interpretation,
we believed that given the sensitive nature of the
Table 2
Participant demographics by opportunity condition
Demographic indicator High opportunity
e
Mean
(Std. Dev.) n = 195
Low opportunity
e
Mean
(Std. Dev.) n = 160
Variable mean di?erence
M
w
À M
w/o
Age (in years) – range from 18 to 80
d
43.28 (13.54) 35.19 (12.07) 8.09
***
Income (categorical, in $)
a,d
3.46 (1.14) 3.61 (1.15) À.15
Education
b,d
2.61 (.85) 2.98 (.77) À.37
***
Paid tax preparer (%)
c
50 64 À14
**
Gender
Male (#) 116 94
Male (%)
c
59.5 58.8 .7
Female (#) 79 66
Female (%) 40.5 41.2
a
Income had ?ve categories: (1) 10k; (2) 10–25k; (3) 25–50k; (4) 50–75k; (5) 75k.
b
Education had four categories, completion of: (1) less than high school; (2) high school; (3) bachelor’s degree; (4) advanced degree.
c ***,**,*
Di?erences statistically signi?cant at p < .01, p < .05, p < .10, one-way ANOVA.
d ***,**,*
Di?erences statistically signi?cant at p < .01, p < .05, p < .10, two-tailed t-test.
e
Participants were classi?ed as ‘‘high opportunity” if they indicated business ownership (with at least some cash sourced income) or
if they received cash income that was not reported to the IRS. All other participants were classi?ed as ‘‘low opportunity”.
7
The questionnaire indicated that if a certain person (or
group) is not applicable, then ‘‘no e?ect” should be selected.
8
By reverse coding, a response of 1 (discourages underre-
porting) was recoded as a 7, and a response of 2 was recoded as
a 6, etc.
9
The ?rst three items are based on Beck and Ajzen (1991).
The remaining two are synonyms for morality.
692 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
items (e.g., reporting taxable income) participants
would feel more comfortable and likely to self-
report truthfully using subjective scales.
Hypothetical underreporting intentions
Two hypothetical scenarios were included in the
survey. Both scenarios described situations involv-
ing an individual receiving ‘‘high opportunity”
revenue. That is, the individual in the scenario
had received a certain amount of cash revenue that
was not being separately reported to the IRS, and
thus, the opportunity to evade was high. The use
of these scenarios placed all participants in identi-
cal ‘‘high opportunity” settings. One scenario
involved the receipt of $5000 of business revenue
and the other scenario involved the receipt of
$2500 of rental income. Following each scenario,
participants were asked that if they were the indi-
vidual in the scenario how much of the $5000 cash
fees ($2500 rental amount) they would include as
income on their tax return. Thus, participants were
asked to provide a speci?c dollar amount. Given
the hypothetical nature of the two scenarios we felt
that participants would be willing to truthfully
provide behavioral intentions of the amount of
income they would underreport.
Analysis – structural equation modeling
Structural equation modeling (SEM) is particu-
larly suited to testing the proposed model. First,
the underlying constructs are inherently latent
(or unobserved) variables, such as taxpayer’s ethi-
cal beliefs toward underreporting. Structural equa-
tion modeling provides a theoretically error-free
measure of latent variables (Hatcher, 1994, p.
258). Second, SEM simultaneously examines the
relations among constructs, which is particularly
important for the current proposed model which
includes mediating variables.
SEM includes the estimation of a measurement
model and a structural model. The purpose of the
measurement model is to evaluate howwell the indi-
vidual measures serve as a measurement instrument
for each underlying latent constructs. As discussed
above, the questionnaire included multiple individ-
ual measures related to underreporting ethics, social
norms, self-underreporting behavior and hypothet-
ical underreporting intentions. The structural
model speci?es the relationships among the latent
constructs (i.e. the hypothesized causal relation-
ships). While the measurement and structural mod-
els are estimated simultaneously, each model is
separately evaluated and interpreted below.
Results
SEM: measurement model ?t
SEM was conducted using the maximum likeli-
hood method of parameter estimation in LISREL
8.54 (Joreskog & Sorbom, 1996, 2003). The factor
loadings and squared multiple correlation (R
2
) are
used to evaluate the validity and reliability, respec-
tively, of the multiple measures of each latent con-
struct.
10
Validity is evidenced by a signi?cant
factor loading that is P.70 (Fornell & Larcker,
1981). As a general rule of thumb, individual mea-
sures exhibiting squared multiple correlations
(R
2
) P.50 are considered reliable (Anderson &
Gerbing, 1988).
The questionnaire includes ?ve indicator state-
ments related to underreporting ethics. Factor
loadings for each indicator statement are pre-
sented in Table 3 Panel A. As shown, the factor
loadings for each measure is signi?cant and above
the recommended threshold of .70 indicating valid-
ity. Furthermore, the reliability of all measures is
acceptable (R
2
P.50). Consequently, the latent
construct of underreporting ethics is based on all
?ve measures.
The questionnaire includes ?ve measures
intended to capture aspects related to social norms.
The factor loadings of all ?ve measures are signi?-
cant. As shown in Table 3 Panel B, three of the ?ve
factor loadings are greater than .7 and have squared
multiple correlations (R
2
) greater than .5. However,
the factor loadings for friends was .618 and for busi-
10
The only modi?cation made to the original measurement
model speci?cation was that errors were allowed to correlate:
(1) between individual measures of underreporting ethics and
social norms – within each latent construct only; and (2)
between the two dependent latent constructs (self-underreport-
ing behavior and hypothetical underreporting intentions).
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 693
ness contacts/peers was .510, indicating weak inter-
nal consistency. Furthermore, the reliability (R
2
) of
friends and business contacts/peers was also low, at
.38 and .26, respectively. Therefore, friends and
business contacts/peers were eliminated as indica-
tors of social norms.
11
Consequently, the latent con-
struct of social norms is based on the remaining
three measures.
The questionnaire includes two measures
related to self-underreporting behavior and two
measures related to hypothetical underreporting
intentions.
12
Measurement model results for the
two underreporting latent constructs are presented
in Table 4. The validity of each measure is accept-
able as evidenced by a signi?cant factor loading
Table 3
Descriptive statistics for underreporting ethics and social norms for underreporting income by opportunity condition
Indicators Standardized factor
loading
c
(R
2
)
d
n = 355
High opportunity
e
Mean (Std. Dev.)
N=195
Low opportunity
e
Mean (Std. Dev.)
n = 160
Variable mean
di?erence
f
M
w
À M
w/o
Panel A: Descriptive statistics for underreporting ethics
Measures of underreporting ethics
a
I would not feel guilty if I excluded some of my
income from my income tax return
b
.715 (.51) 4.21 (2.33) 4.88 (2.06) À.67
***
When ?ling my income tax return,
underreporting any amount of income goes
against my principles
.921 (.85) 4.57 (2.22) 5.49 (1.55) À.92
***
It would be morally wrong for me to
underreport my income when ?ling my income
taxes
.851 (.72) 4.51 (2.17) 5.33 (1.79) À.82
***
I do not think it is ethically wrong for me to
exclude small amounts of income when ?ling
my income taxes
b
.750 (.56) 4.39 (2.22) 4.89 (2.01) À.50
**
It would be dishonest for me to exclude any
amount of income when calculating my income
tax
.705 (.50) 4.70 (2.26) 5.27 (1.81) À .57
***
Standardized factor
loading
h
(R
2
)
d
n = 355
High opportunity
e
Mean (Std. Dev.)
N = 195
Low opportunity
e
Mean (Std. Dev.)
n = 160
Variable mean
di?erence
f
M
w
À M
w/o
Panel B: Descriptive statistics for social norms for underreporting income
Measures of social norms
g
Spouse/signi?cant other .831 (.69) 4.56 (1.46) 4.73 (1.33) À.17
Family .800 (.64) 4.66 (1.30) 4.34 (1.45) .32
**
Tax return preparer .799 (.64) 4.70 (1.32) 4.66 (1.55) .04
a
Measured on 7-point Likert scale ranging from de?nitely (1) disagree to (7) agree.
b
Items were reverse coded.
c
Factor loading measures indicator validity. All factor loadings are statistically signi?cant at p < .0001.
d
R
2
measures indicator reliability.
e
Participants were classi?ed as ‘‘high opportunity” if they indicated business ownership (with at least some cash sourced income) or
if they received cash income that was not reported to the IRS. All other participants were classi?ed as ‘‘low opportunity”.
f ***,**,*
Di?erences statistically signi?cant at p < .01, p < .05, p < .10 two-tailed t-test.
g
Measured from (1) encourages underreporting to (7) discourages underreporting.
h
Factor loading measures indicator validity. All factor loadings are statistically signi?cant at p < .001.
11
Dropping measures related to friends and business contacts/
peers is not intended to indicate that these groups are not
sources of social norms. They very well may be sources of social
norms. While the correlations between friends and business
contacts/peers and the other three measures range between .40
and .52, the two measures do not appropriately load on a single
factor.
12
To simplify the discussion of results, the terms ‘‘behavior”
and ‘‘intention” are used to mean ‘‘self-underreporting behav-
ior” and ‘‘hypothetical underreporting intention”, respectively.
694 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
exceeding .70. Furthermore, the reliability of each
measure is acceptable (R
2
P.50). Consequently,
the latent constructs of behavior and intentions
are each based on two measures. Overall, the mea-
surement model provides an acceptable ?t to the
data.
SEM: structural model ?t
SEM ?t indices measure the extent to which the
covariance matrix derived from the hypothesized
model is di?erent from the covariance matrix
derived from the sample. Numerous ?t indices
associated with the SEM are reported in Table 5.
Although the v
2
-test of ?t is statistically signi?-
cant.
13
The model provides a good ?t. Speci?cally,
the v
2
/df is less than the recommended acceptable
cut-o? of 2.0 (Hatcher, 1994, p. 393), the root
mean square error of approximation (RMSEA)
is .044, which is below the recommended cut-o?
of .06, and the remaining ?t indices are all greater
than .90, indicting an acceptable ?t. Overall, the
hypothesized model provides an acceptable ?t to
the data.
Hypothesis testing
The SEM path results are shown in Fig. 2.
Hypothesis one proposes that high opportunity
taxpayers will judge underreporting as less unethi-
cal compared to low opportunity taxpayers. In
testing the model, high opportunity taxpayers are
coded 1, and low opportunity taxpayers are coded
0. Thus, hypothesis one predicts a negative path
between opportunity and underreporting ethics.
As shown in Fig. 2, the coe?cient for the path
from opportunity to underreporting ethics is nega-
tive and signi?cant (z-stat À4.55, p = .00), which
supports hypothesis one.
The descriptive statistics and univariate statisti-
cal results presented in Table 3 Panel A provide
additional detail regarding the individual measures
of underreporting ethics and how they di?er
between the two taxpayer groups. Table 3 Panel
A shows that the mean responses among high
opportunity taxpayers (in comparison to low
opportunity taxpayers), indicate that: (1) they
would feel less guilty if underreporting income;
(2) underreporting went against their principles
to a lesser extent; (3) underreporting was less mor-
ally wrong; (4) underreporting was less unethical;
and (5) underreporting was less dishonest. This
pattern is consistent with hypothesis one.
Table 4
Descriptive statistics for underreporting income measures by opportunity condition
Indicators Standardized factor
loading
e
(R
2
)
f
n = 355
High opportunity
g
Mean (Std. Dev.)
N = 195
Low opportunity
g
Mean (Std. Dev.)
n = 160
Variable mean
di?erence
h
M
w
À M
w/o
Self-underreporting behavior – amount
a
.799 (.64) 1.97 (1.39) 1.38 (.93) .59
***
Self-underreporting behavior – frequency
b
.782 (.61) 2.48 (1.85) 1.68 (1.17) .80
***
Hypothetical underreporting intentions –
business (in $)
c
.837 (.70) 1941 (2155) 1084 (1917) 857
***
Hypothetical underreporting intentions –
rental (in $)
d
.736 (.54) 1033 (1156) 657 (1062) 376
***
a
Measured on 7-point Likert scale ranging from (1) included all income to (7) included none.
b
Measured on 7-point Likert scale ranging from (1) never underreported to (7) always underreported.
c
Maximum amount of underreporting possible was $5000.
d
Maximum amount of underreporting possible was $2500.
e
Factor loading measures indicator validity. All factor loadings are statistically signi?cant at p < .001.
f
R
2
measures indicator reliability.
g
Participants were classi?ed as ‘‘high opportunity” if they indicated business ownership (with at least some cash sourced income) or
if they received cash income that was not reported to the IRS. All other participants were classi?ed as ‘‘low opportunity”.
h ***,**,*
Di?erences statistically signi?cant at p < .01, p < .05, p < .10 two-tailed t-test.
13
In applied settings, the v
2
index is often not valid and should
be supplemented with other goodness of ?t indices (Hatcher,
1994, p. 189).
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 695
Hypothesis two predicts a negative relationship
between taxpayer underreporting ethics and behav-
ior. As shown in Fig. 2, the path from underreport-
ing ethics to behavior is negative and signi?cant
(z-stat À9.61, p = .00). This result is consistent with
prior research and supports hypothesis two.
Hypothesis three predicts that, controlling for
ethical beliefs, the direct association between
opportunity and behavior will be signi?cant. Appli-
cable descriptive statistics are shown in Table 4. For
each measure of behavior, the mean value for high
opportunity taxpayers is signi?cantly greater (e.g.,
less compliant) than the mean value for low oppor-
tunity taxpayers. However, hypothesis three holds
that a signi?cant direct path between opportunity
and behavior will persist after controlling for und-
erreporting ethics. As shown in Fig. 2, the direct
path from opportunity to behavior is positive and
signi?cant (z-stat 4.49, p = .00) which supports
hypothesis three.
Taken together, the results of hypotheses one,
two and three indicate that underreporting ethics
partially mediates the relation between taxpayers’
opportunity to evade and behavior. While the indi-
rect paths are signi?cant (H1 and H2), indicating
mediation, the signi?cant direct path between
opportunity and behavior suggests that underre-
porting ethics does not fully account for (e.g.,
mediate) the relation between opportunity and
behavior.
Hypothesis four includes two related parts. The
?rst part predicts a negative relationship between
taxpayer underreporting ethics and intentions.
Fig. 2 shows that the path between these two vari-
ables is negative and signi?cant (z-stat À9.42,
p = .00). The second part predicts that the relation
between taxpayer underreporting ethics and inten-
tions will be similar to (e.g., not signi?cantly di?er-
ent from) the relation between underreporting
ethics and behavior. That is, the coe?cients of
the two paths are predicted to be equivalent. As
shown in Fig. 2, the standardized coe?cients of
these two paths appear similar, at À.65 and
À.66, respectively.
To test the second part of the hypothesis, the
paths from underreporting ethics to intentions
and from underreporting ethics and to behavior
are constrained to be equal. This constrained
model is compared to the original model.
14
A v
2
di?erence test is used to assess whether constrain-
ing the paths to be equal a?ects the v
2
-test statistic
in a signi?cant manner. If the di?erence is insignif-
icant, the paths can be interpreted as not signi?-
cantly di?erent from each other. The v
2
di?erence test between the constrained and origi-
nal SEMs indicates that the two models are not
signi?cantly di?erent (v
2
d
¼ :40, df = 1, p = .527).
This result is consistent with the second part of
hypothesis four. Overall, these results support
both parts of hypothesis four.
Hypothesis ?ve predicts that, controlling for
ethical beliefs, the direct relationship between
opportunity and intentions will be signi?cantly
weaker than the direct relationship between
opportunity and behavior. Speci?cally, this
Table 5
SEM ?t indices
Selected goodness of ?t indices Original SEM results
Indices not adjusted for parsimony
v
2
p-value
a
.0028
v
2
/degrees of freedom
b
1.63
Root mean square error of
approximation (RMSEA
c
)
.044
Goodness of ?t Index (GFI)
d
.96
Normed ?t index (NFI)
d
.98
Comparative ?t index (CFI)
d
.99
Relative ?t index (RFI)
d
.97
Incremental ?t index (IFI)
d
.99
Notes: Fit indices measure if the covariance matrix derived from
the hypothesized model is di?erent from the covariance matrix
derived from the sample. Thus, a nonsigni?cant di?erence
supports the model ?t to the data.
a
Tests closeness of ?t. p-Value associated with test of
covariance models.
b
Tests closeness of ?t. v
2
/df = 1.0 for exact ?t. Values < 2.0
indicate acceptable ?t.
c
Root mean square error of approximation. Values < .06
indicate acceptable ?t.
d
Hypothesized model compared to standard model (e.g.,
independence or null model). Indices range from to 1.0. Val-
ues > .9 indicate acceptable ?t.
14
Additional discussion about the use of constrained models
in hypothesis testing is provided by Blanthorne, Jones-Farmer,
and Almer (2006). Wouters, Anderson, and Wynstra (2005)
provide a recent example of using of this approach as well.
696 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
hypothesis predicts that the coe?cients of the two
paths will not be equivalent. To test this hypothe-
sis, the paths from opportunity to intentions and
from opportunity and to behavior are constrained
to be equal. This constrained model is compared
to the original model. As before, a v
2
di?erence
test is used to assess any di?erence.
Fig. 2 shows that the direct paths from opportu-
nity to intentions (z-stat 2.04, p = .04) and to
behavior (z-stat 4.49, p = .00) are each signi?cant
and the coe?cients for the two paths are .11 and
.23, respectively. The v
2
di?erence test between
the constrained and original SEMs indicates that
the two models are signi?cantly di?erent
(v
2
d
¼ 4:03, df = 1, p = .045). This result supports
hypothesis ?ve and indicates that the standardized
coe?cient from opportunity to behavior is signi?-
cantly stronger than the path from opportunity to
intentions.
The descriptive statistics and univariate statisti-
cal results are presented in Table 4. As shown, the
mean underreporting amounts in the two hypothet-
ical scenarios for high opportunity taxpayers signif-
icantly exceed the mean underreported for low
opportunity taxpayers by $857 and $376, for the
business and rental scenarios, respectively. Further-
more, the SEM results provide two additional
insights. The indirect paths from opportunity to
underreporting ethics and from underreporting
ethics to intentions are both signi?cant. Further-
more, the path between opportunity and intentions
remains signi?cant after controlling for ethical
beliefs which indicates ethical beliefs partially medi-
ate the relationship between opportunity and
intentions.
Hypothesis six predicts that taxpayers with
high opportunity to underreport income will
judge social norms for underreporting to be less
unethical than taxpayers with low opportunity
to underreport income. As shown in Fig. 2, the
path coe?cient from opportunity to social norms
is insigni?cant. Table 3 Panel B reports the
Path Significant
Path Not Significant
se = Standard Error
Self
Underreporting
Behavior
.23 z-stat 4.49
se .05, p=.00
.11 z-stat 2.04
se .05, p=.04
-.66 z-stat -9.61
se .07, p=.00
-.26 z-stat -4.55
se .06, p=.00
-.06 z-stat -.99
se .06, p=.32
.22 z-stat 3.36
se .07, p=.00
-.65 z-stat -9.42
se .07, p=.00 -.07 z-stat -1.36
se .05, p=.17
Opportunity
Hypothetical
Underreporting
Intention
Underreporting
Ethics
Social
Norms
-.07 z-stat -1.30
se .06, p=.19
Fig. 2. SEM results.
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 697
descriptive statistics and univariate statistical
results regarding social norms about underreport-
ing. As shown, the means for high and low
opportunity groups di?er signi?cantly only in
regards to family. The higher mean value for high
opportunity taxpayers suggests that family mem-
bers discourage underreporting which is inconsis-
tent with social norms being interpreted in an
egocentric fashion. Overall, the results do not
support hypothesis six.
Hypotheses seven and eight o?er competing
predictions regarding the relations among social
norms, ethical beliefs, and behavior. Hypothesis
seven predicts the path from social norms to und-
erreporting ethics will be signi?cant but that the
path from social norms to behavior will not be sig-
ni?cant (i.e. full mediation). Alternatively, hypoth-
esis eight predicts that the path from social norms
to behavior will remain signi?cant (i.e. partial
mediation). As shown in Fig. 2, the path coe?cient
from social norms to underreporting ethics is posi-
tive and signi?cant (z-stat 3.36, p = .00) indicating
that social norms discouraging underreporting are
associated with one’s beliefs that underreporting is
unethical. Fig. 2 also shows that the path from
social norms to behavior is insigni?cant.
15
Thus,
after controlling for ethical beliefs, social norms
do not have a direct e?ect on taxpayer behavior.
The univariate correlation between social norms
and behavior was also determined and found to
be signi?cant (r = .23, p < .01). This correlation
indicates that when ethical beliefs are excluded
from the analysis, social norms and behavior are
signi?cantly related. Overall, these results indicate
that ethical beliefs mediate the relationship
between social norms and behavior, and provide
support for hypothesis seven but not for hypothe-
sis eight.
Participant demographic measures (see Table 2
for descriptions and measurement) were added to
the SEM to control for individual di?erences that
might relate to underreporting behavior and/or
intentions. The demographic measures were tested
in two ways. First, because the demographic vari-
ables conceivably a?ect taxpayer opportunity
choice, the demographic variables were tested as
inputs (i.e. precursors) to opportunity. Signi?cant
e?ects (p < .05) were found for age and level of
taxpayer education. Consistent with univariate
results shown in Table 2, high opportunity taxpay-
ers were older and had less formal education.
However, these in?uences do not alter the reported
SEM results. Second, demographic measures were
added in a more traditional manner, as inputs to
the latent behavior construct and to the latent
intention construct. While some of the control
variables were signi?cant to underreporting, their
inclusion in the SEM did not alter the reported
results.
16
Discussion
The purpose of the current study was to test a
model of the proposed relations among opportu-
nity to underreport income, ethical beliefs and
social norms about underreporting income, and
income underreporting. The model is grounded
in research that has found that individuals have
a tendency to form ethical judgments in an
egocentric manner (Epley & Caruso, 2004; Neale
& Bazerman, 1983; Thompson & Loewenstein,
1992). Existing tax compliance research implicitly
assumes that social norms and ethical beliefs are
independent from taxpayers’ opportunity to
evade. The current study extends academic tax
compliance research by proposing and testing a
model in which the value of noneconomic vari-
ables are contingent upon one’s economic condi-
tion or situation.
Our results provide evidence that the e?ects of
opportunity to evade on evasion behavior are
direct and indirect. Because opportunity changes
the expected values, increases in opportunity were
found to directly in?uence intentions and decisions
to evade. Our study also shows that the opportu-
15
While not hypothesized, the path from social norms to
hypothetical underreporting intentions was also tested. The
path is insigni?cant.
16
Age was signi?cant (p < .05) and use of a paid tax preparer
was marginally signi?cant (p < .10) to behavior and intentions.
In addition, gender was marginally signi?cant to intentions.
698 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
nity to evade, which increases the expected values,
also in?uences the formation is one’s ethical beliefs
about evasion, which, in turn, in?uences one’s
intentions and decisions to evade. Results from
our study can be used to reconsider Alm’s (1991)
comment presented earlier. While it is no doubt
true that ‘‘some individuals pay taxes because they
believe that cheating is wrong” (Alm, 1991, p.
584), our ?ndings indicate that one’s ethical beliefs
about evasion are shaped, in part, by economic cir-
cumstances. Speci?cally, our results suggest that
ethical beliefs about tax evasion are least likely
to inhibit those taxpayers that are most likely,
based on economic self-interest, to engage in eva-
sion. Our results also indicate that social norms
also in?uenced ethical beliefs, suggesting that indi-
viduals internalize social norms when forming
their own ethical beliefs. Thus, our evidence indi-
cates that social norms, through ethical beliefs,
have an indirect e?ect on evasion intentions and
behavior.
From a tax policy perspective, our model sug-
gests that policy approaches that restrict taxpay-
ers’ opportunity to evade are likely to yield dual
bene?ts (e.g., economic and ethical). Conse-
quently, the bene?ts of policies that further
restrict taxpayers’ opportunity are likely to be
greater than one would expect assuming ethical
beliefs are independently formed. Alternatively,
the bene?ts of policies aimed at instilling beliefs
that tax evasion is unethical may be less success-
ful than one would expect assuming ethical beliefs
are formed independently from economic
considerations.
Our results have additional implications for tax
compliance researchers and policy makers. First,
consider our ?nding that social norms indirectly
in?uence self-underreporting behavior (and inten-
tions) through ethical beliefs. Wenzel (2005b)
recently found that taxpayers generally overesti-
mate other taxpayers’ acceptance of tax evasion.
To the extent that social norms directly or indi-
rectly in?uence compliance behavior, overestimat-
ing social norms will likely lead to more tax
noncompliance. As discussed above, Wenzel
(2005b) found that an intervention in the form
of a communication about a general mispercep-
tion of tax compliance social norms (e.g., others
are far more compliant than perceived) signi?-
cantly improved compliance behavior in both an
experimental and a ?eld study. This ?nding,
placed in the context with our tax compliance
model, suggests that policy makers should explore
ways to better align taxpayers’ social norms with
respect to actual compliance behavior (e.g., tax-
payers generally overestimate the level of noncom-
pliance). Tax compliance researchers can play an
important role by providing evidence on the rela-
tive e?ectiveness of alternative methods that
might be used to alter taxpayers’ misperceptions
of social norms of tax compliance. For example,
would letters or emails sent to individual taxpay-
ers be more or less e?ective than a broader adver-
tising campaign aimed at addressing taxpayers’
misperceptions?
Regarding ethical beliefs about underreport-
ing, we believe our model suggests that using
various forms of communication intended to
strengthen ethical beliefs will be challenging. Pre-
vious research has explored the e?ectiveness of
certain kinds of communications to strengthen
tax compliance through ethical beliefs (Jackson
& Jaouen, 1989; Kaplan et al., 1997; Violette,
1989) and reported mixed results. Our results
suggest that high opportunity taxpayers have
economic reasons to discount ethics-related com-
munications by dismissing the source as biased
or neutralizing the relevance of the message.
Berso? (1999) has explored ways to address situ-
ations where individuals have incentives to
engage in unethical behavior. Berso? contends
that individuals will be less able to engage in
denial of harm or victim neutralization when
the ‘‘victim” is personalized. In the context of
tax compliance, instead of describing losses and
harm to a faceless federal government, messages
indicating how speci?c individuals in their com-
munity might be harmed (e.g., inability to fund
certain types of health care for senior citizens,
or Head Start education programs for young
children) may be more e?ective. We encourage
further research to explore the use of messages
that personalize the harm from noncompliance
to in?uence ethical beliefs and/or tax compliance
intentions and behavior, particularly among high
opportunity taxpayers.
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 699
In considering our results it is important to take
into account several limitations. First, it is not pos-
sible to completely rule out self-selection as an
explanation for our results. That is, it is possible
that taxpayers who tend to believe that tax evasion
is less unethical self-select into high opportunity
income occupations. Recent results from a study
by Parker (2003) bear on this issue. He provides
evidence that opportunities for tax evasion do
not in?uence occupational choice. Using several
British micro-data sets, his ?ndings indicate that
one’s occupational choice between self-employ-
ment and paid employment is unrelated to oppor-
tunities to evade taxes.
Our ?ndings show that taxpayers’ opportunity
to evade is unrelated to their social norms. This
result, which was not expected, provides some
comfort against a strong self-selection bias. That
is, in terms of tax compliance, both high and low
opportunity taxpayers appear to have been sur-
rounded by similar social networks. This ?nding
would appear to rule out the possibility of high
opportunity taxpayers seeking out a social net-
work that holds similar beliefs about tax compli-
ance. Furthermore, the SEM results remain
stable when participant demographic measures
are added to the model. Thus, while we can not
rule out self-selection bias, our results are consis-
tent with our theoretical model and additional evi-
dence suggests that self-selection is not in?uencing
our results.
Evidence to test our model was obtained from
a survey of taxpayers. While a survey allowed us
to collect evidence across a range of constructs,
often involving multiple measures from taxpay-
ers, results from surveys generally are unable to
demonstrate causality between constructs. For
example, while our model indicates underreport-
ing ethics in?uences compliance decisions, our
analysis using SEM cannot de?nitively determine
the direction of causality. A recent study by
Wenzel (2005a) provides evidence of the issue
of causality between ethics and evasions. Using
a unique data set that follows taxpayers for
two years, Wenzel reports that ethics ‘‘motivates”
tax compliance (e.g., ethics at time 1 is associated
with compliance at time 2, controlling for other
variables) as well as evidence that compliance
fosters ‘‘rationalization” (e.g., tax compliance at
time 1 is associated with ethics at time 2, control-
ling for other variables). Thus, while this evi-
dence does not rule out rationalization, it does
indicate that ethical beliefs motivate compliance
behavior as proposed in our model. Perhaps,
experimental methods may be used in future
research to provide additional, possibly stronger,
evidence on the direction of causality between
variables.
Finally, a limitation concerns our proxies for
underreporting income. Our actual measures of
underreporting are based on self-reports. It is
possible that such measures are not completely
reliable. Our measures for hypothetical underre-
porting represent behavioral intentions. Prior
research has found intentions to be a strong pre-
dictor of behavior (Ajzen, 1988; Ajzen & Fishbein,
1980), but in the tax compliance setting, some
studies have shown such measures are undepend-
able (Hessing, El?ers, & Weigel, 1988). A recent
paper by Webley, Cole, and Eidjar (2001) follows
on Hessing et al. (1988), examining the di?erences
in hypothetical tax evasion measures and self-
reported tax evasion behavior. Findings from
Webley et al. (2001) ease concerns about relying
on hypothetical scenarios and on self-reported
behavior by demonstrating that the di?erences
are traceable to constraints on actual behavior.
Finally, our model was examined exclusively in
the context of underreporting income. Thus, it is
unclear whether our model will generalize to over-
reporting expenses or other forms of tax evasion.
While we believe the model should generalize to
broader settings, this remains an issue for further
research.
Acknowledgement
This paper has bene?ted from workshop partic-
ipants at the Accounting Behavioral Organization
Conference, the Ethics Symposium of the Ameri-
can Accounting Association, Arizona State Uni-
versity, the University of Rhode Island, the
University of North Carolina at Charlotte, the
University of Delaware, and the University of
New Mexico.
700 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
Appendix A. Correlation matrix of SEM measures
opp eth1 eth2 eth3 eth4 eth5
opp 1.000
eth1 À0.195 1.000
eth2 À0.295 0.628 1.000
eth3 À0.253 0.545 0.801 1.000
eth4 À0.148 0.632 0.613 0.643 1.000
eth5 À0.175 0.505 0.668 0.640 0.515 1.000
snSpouse À0.079 0.201 0.147 0.192 0.195 0.163
snFamily À0.146 0.188 0.136 0.147 0.233 0.154
snPrep À0.019 0.153 0.081 0.086 0.243 0.095
snFriend À0.107 0.173 0.138 0.159 0.213 0.154
snPeer À0.046 0.141 0.189 0.174 0.218 0.183
Self1Amt 0.343 À0.454 À0.505 À0.454 À0.449 À0.419
Self2Fre 0.299 À0.471 À0.551 À0.493 À0.521 À0.426
Hyp1Bus 0.269 À0.444 À0.522 À0.481 À0.514 À0.420
Hyp2Rent 0.204 À0.417 À0.445 À0.377 À0.437 À0.341
snSpouse snFamily snPrep snFriend snPeer Self1Amt
snSpouse 1.000
snFamily 0.665 1.000
snPrep 0.497 0.633 1.000
snFriend 0.496 0.502 0.522 1.000
snPeer 0.457 0.410 0.411 0.703 1.000
Self1Amt À0.179 À0.210 À0.199 À0.204 À0.215 1.000
Self2Fre À0.165 À0.146 À0.111 À0.206 À0.316 0.641
Hyp1Bus À0.132 À0.145 À0.176 À0.166 À0.166 0.491
Hyp2Rent À0.175 À0.212 À0.203 À0.216 À0.161 0.438
Self2Fre Hyp1Bus Hyp2Rent
Self2Fre 1.000
Hyp1Bus 0.435 1.000
Hyp2Rent 0.396 0.620 1.000
References
Ajzen, I. (1988). From intentions to actions. Attitudes, person-
ality and behavior. London, England: Open University
Press, pp. 112–145.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and
predicting social behavior. Englewood Cli?s, NJ: Prentice-
Hall.
Allingham, M. G., & Sandmo, A. (1972). Income tax evasion: A
theoretical analysis. Journal of Public Economics, 323–338.
Alm, J. (1991). A perspective on the experimental analysis of
taxpayer reporting. Accounting Review, 66(3), 577–593.
Alm, J., McClelland, G. H., & Schultze, W. D. (1992). Why
do people pay taxes? Journal of Public Economics, 48(1),
21–38.
Alm, J., McClelland, G. H., & Schultze, W. D. (1999).
Changing social norms of tax compliance by voting.
KRYLOS, 52(2), 141–171.
Alm, J., & Sanchez, I. (1995). Economic and noneconomic
factors in tax compliance. KRYLOS, 48(1), 3–19.
Anderson, J. C., & Gerbing, D. W. (1988). Structural
equation modeling in practice: A review and recom-
mended two-step approach. Psychological Bulletin, 103(3),
411–423.
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 701
Antonides, G., & Robben, H. S. J. (1995). True positives and
false alarms in the detection of tax evasion. Journal of
Economic Psychology, 16(4), 617–640.
Arrington, C. E., & Reckers, P. M. J. (1985). A social
psychological investigation into perceptions of tax evasion.
Accounting and Business Research, 15(Summer), 163–176.
Babcock, L., Loewenstein, G., Issacharo?, S., & Camerer, C.
(1995). Biased judgments of fairness in bargaining. Amer-
ican Economic Review, 85(5), 1337–1343.
Baron, R., & Kenny, D. (1986). The moderator–mediator
variable distinction in social psychological research: Con-
ceptual, strategic and statistical considerations. Journal of
Personality and Social Psychology, 51(6), 1173–1182.
Bazerman, M. H., Curhan, J., Moore, D., & Valley, K. (2000).
Negotiations. Annual Review of Psychology, 51, 279–314.
Beck, L., & Ajzen, I. (1991). Predicting dishonest actions using
the theory of planned behavior. Journal of Research in
Personality, 25(3), 285–301.
Berso?, D. M. (1999). Why good people sometimes do bad
things: Motivated reasoning and unethical behavior. Per-
sonality and Social Psychology Bulletin, 25, 28–39.
Blanthorne, C., Jones-Farmer, A., & Almer, E. D. (2006). Why
you should consider using SEM: A guide to getting started.
Advances in Accounting Behavioral Research, 9, 179–207.
Bobek, D., Roberts, R., & Sweeney, J. (2007). The social norms
of tax compliance: Evidence from Australia, Singapore, and
the United States. Journal of Business Ethics, 74(1), 49–64.
Bobek, D., & Hat?eld, R. (2003). An investigation of the theory
of planned behavior and the role of moral obligation in tax
compliance. Behavioral Research in Accounting, 15, 13–38.
Carnes, G. A., & Englebrecht, T. D. (1995). An investigation of
the e?ect of detection risk perceptions, penalty sanctions
and income visibility on tax compliance. Journal of the
American Taxation Association, 17(1), 26–41.
Cialdini, R., & Trost, M. (1998). Social in?uence: Social norms,
conformity, and compliance. In D. Gilbert, S. Fiske, & G.
Lindzey (Eds.), The handbook of social psychology (4th ed.).
New York: Oxford University Press.
Davis, J. S., Hecht, G., & Perkins, J. D. (2003). Social
behaviors, enforcement, and tax compliance dynamics.
Accounting Review, 78(1), 39.
Epley, N., & Caruso, E. M. (2004). Egocentric ethics. Social
Justice Research, 17(2), 171–187.
Erard, B., & Feinstein, J. S. (1994). The role of moral
sentiments and audit perceptions in tax compliance. Public
Finance, 49(Suppl.), 70–89.
Etzioni, A. (1988). Normative-a?ective factors: Toward a new
decision making model. Journal of Economic Psychology, 9,
125–150.
Etzioni, A. (1993). Normative–a?ective choices. Human Rela-
tions, 46(9), 1053–1069.
Fischer, C. M., Wartick, M., & Mark, M. M. (1992). Detection
probability and taxpayer compliance: A review of the
literature. Journal of Accounting Literature, 11, 1–46.
Fornell, C., & Larcker, D. G. (1981). Evaluating structural
equation models with unobservable variables and measure-
ment error. Journal of Marketing Research, 18, 39–50.
Frazier, P. A., Tix, A. P., & Barron, K. E. (2004). Testing
moderator and mediator e?ects in counseling psychology
research. Journal of Counseling Psychology, 51(1), 115–134.
Ghosh, D., & Crain, T. L. (1996). Experimental investigation of
ethical standards and perceived probability of audit on
intentional noncompliance. Behavioral Research in Account-
ing, 8(supplement), 219–244.
Haan, N. (1975). Hypothetical and actual moral reasoning in a
situation of civil disobedience. Journal of Personality and
Social Psychology, 32(August), 255–270.
Hanno, D., & Violette, G. (1996). An analysis of moral
obligation and social in?uences on taxpayer behavior.
Behavioral Research in Accounting, 8, 57–75.
Hatcher, L. (1994). A step-by step approach to using SAS system
for factor analysis and structural equation modeling. Cary,
NC: SAS Institute Inc..
Henderson, B. C., & Kaplan, S. E. (2005). An examination of
the role of ethics in tax compliance decisions. Journal of the
American Taxation Association, 27(1), 39–72.
Hessing, D. J., El?ers, H., & Weigel, R. H. (1988). Exploring
the limits of self-reports and reasoned action: An investiga-
tion of the psychology of tax evasion. Journal of Personality
and Social Psychology, 54(3), 405–413.
Internal Revenue Service. (1996). Federal tax compliance
research: Individual income tax gap estimates for 1985,
1988 and 1992. Publication 1415 (Rev. Apr.) Washington,
DC: US Government Printing O?ce.
Internal Revenue Service. (2005a). The IRS research bulletin:
Recent research. Publication 1500. Washington, DC.
.
Internal Revenue Service. (2005b). Tax year 2004. Taxpayer
usage study. Weekly Report Number 15. .
Internal Revenue Service. (2006). IRS updates tax gap esti-
mates. IR-2006-28, February. .
Jackson, B. R., & Jaouen, P. R. (1989). In?uencing taxpayer
compliance through sanction threat or appeals to con-
science. Advances in Taxation, 2, 131–147.
Jackson, B. R., & Milliron, V. C. (1986). Tax compliance
research: Findings, problems and prospects. Journal of
Accounting Literature, 5, 125–165.
Jones, T. M. (1991). Ethical decision making by individuals in
organizations: An issue-contingent model. Academy of
Management Review, 16(2), 366–395.
Joreskog, K., & Sorbom, D. (1996). LISREL 8: User’s reference
guide. Chicago: Scienti?c Software International.
Joreskog, K., & Sorbom, D. (2003). LISREL 8.54. Chicago:
Scienti?c Software International.
Kachelmeier, S. J., & Towry, K. L. (2002). Negotiated transfer
pricing: Is fairness easier said than done? Accounting
Review, 77(3), 571–593.
Kagan, R. A. (1989). On the visibility of income tax law
violations. In J. A. Roth & J. T. Scholz (Eds.), Taxpayer
compliance, Volume 2: Social science perspective (pp. 76–125).
Philadelphia, PA: University of Pennsylvania Press.
702 C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703
Kaplan, S. E. (2001). Ethically related judgments by observers
of earnings management. Journal of Business Ethics, 32(4),
285–298.
Kaplan, S. E., Newberry, K. J., & Reckers, P. M. J. (1997). The
e?ect of moral reasoning and educational communications
on tax evasion intentions. Journal of the American Taxation
Association, 19(2), 38–54.
Kaplan, S. E., Reckers, P. M. J., & Roark, S. J. (1988). An
attribution theory analysis of tax evasion related judgments.
Accounting, Organization and Society, 13, 371–379.
Kelman, H. (1958). Compliance, identi?cation, and internali-
zation: Three processes of attitude change. Journal of
Con?ict Resolution, 2, 51–60.
Luft, J. L., & Libby, R. (1997). Pro?t comparisons, market
prices and managers’ judgments about negotiated transfer
prices. Accounting Review, 72(2), 217–229.
Madeo, S. A., Schepanski, A., & Uecker, W. C. (1987).
Modeling judgments of taxpayer compliance. Accounting
Review, 62(2), 323–342.
Messick, D. M., & Sentis, K. P. (1979). Fairness and preference.
Journal of Experimental Social Psychology, 15, 418–434.
Neale, M. A., & Bazerman, M. H. (1983). The role of
perspective-taking ability in negotiating under di?erent
forms of arbitration. Industrial and Labor Relations Review,
36(3), 378–389.
Parker, S. C. (2003). Does tax evasion a?ect occupational
choice? Oxford Bulletin of Economics and Statistics, 65(3),
379–394.
Pommerehne, W., Hart, A., & Frey, B. (1994). Tax morale: Tax
evasion, and the choice of tax policy instruments in di?erent
political systems. Public Finance, 49(Suppl.), 52–69.
Reckers, P. M. J., Sanders, D. L., & Roark, S. J. (1994). The
in?uence of ethical attitudes on taxpayer compliance.
National Tax Journal, 47(4), 825–836.
Scholz, J., & Pinney, N. (1995). Duty, fear, and tax compliance:
The heuristic basis of citizenship behavior. American Jour-
nal of Political Science, 39(2), 490–512.
Smith, K. W. (1990). Integrating three perspectives on
noncompliance. Criminal Justice and Behavior, 17(3),
350–369.
Thompson, L., & Loewenstein, G. (1992). Egocentric inter-
pretations of fairness and interpersonal con?ict. Organi-
zational Behavior and Human Decision Processes, 51,
176–197.
Violette, G. R. (1989). E?ects of communicating sanctions on
taxpayer compliance. Journal of the American Taxation
Association, 11(1), 92–104.
Webley, P., Cole, M., & Eidjar, O. (2001). The prediction of
self-reported and hypothetical tax evasion: Evidence from
England, France and Norway. Journal of Economic Psy-
chology, 22(2), 141.
Wenzel, M. (2004). An analysis of norm processes in tax
compliance. Journal of Economic Psychology, 25, 213.
Wenzel, M. (2005a). Motivation or rationalisation? Causal
relations between ethics, norms, and tax compliance.
Journal of Economic Psychology, 26, 491–508.
Wenzel, M. (2005b). Misperceptions of social norms about tax
compliance: From theory to intervention. Journal of Eco-
nomic Psychology, 26, 862–883.
Wikipedia. (2007). Wikipedia, the free encyclopedia. .
Wouters, M., Anderson, J. C., & Wynstra, F. (2005). The
adoption of total cost of ownership for sourcing decisions –
A structural equations analysis. Accounting, Organizations
and Society, 30, 167–191.
Young, J. C. (1994). Factors associated with noncompliance:
Evidence from the Michigan tax amnesty program. Journal
of the American Taxation Association, 16(2), 82–105.
C. Blanthorne, S. Kaplan / Accounting, Organizations and Society 33 (2008) 684–703 703

doc_427076092.pdf