The Effect of Friendly

Persuasion and Gender on

Tax Compliance Behavior

ABSTRACT. Friendly persuasion, in contrast to

deterrent measures like tax audits and penalties on
underreported taxes, is a positive and possibly a cost
effective method of increasing taxpayer compliance.
However, prior studies have failed to show that
friendly persuasion has a significant impact on compliance (Blumenthal et al., 2001; McGraw and Scholz,
1991). In our study, in contrast to prior studies, we
examine the impact of generating and reading reasons
supporting compliance as friendly persuasion on
individuals income reporting behavior as well as
control for gender effects. Specifically, we predict an
interaction effect between friendly persuasion and
gender on compliance behavior. We carried out a 2
(friendly persuasion and control) 2 (men and
women) full factorial experiment, where participants
earned $30 by completing two questionnaires.
Participants in the friendly persuasion group were
required first to generate and second to read a list of
reasons why they should comply fully. Afterwards,
participants in both groups were asked to report the
income they earned and pay tax on the reported
income. The results show a significant main effect
for gender as well as a significant interaction effect
between gender and friendly persuasion on income
reported. Women in the friendly persuasion group

Janne Chung holds a Ph.D. from Edith Cowan University,

Western Australia. She is currently Associate Professor
in Accounting at the Schulich School of Business, York
University, Toronto. Professor Chungs research interests include auditing, ethics, and tax compliance. She
has published articles in Journal of Business Ethics
and Behavioral Research in Accounting.
Viswanath Umashanker Trivedi holds a Ph.D. from
Arizona State University, U.S.A. He is currently
Assistant Professor in Accounting at the Schulich School
of Business, York University, Toronto. Professor Trivedis
research interests include experimental economics, tax
compliance, and ethics among others.

Janne Chung
Viswanath Umashanker Trivedi

reported significantly higher income compared to

men in that group. Other comparisons were not
significant. Policy implications for increasing taxpayers ethics and compliance are highlighted.
KEY WORDS: generating and reading reasons,
income tax reporting, normative appeals, personal
consequences.

I. Introduction
Tax evasion is a form of illegal behavior involving
a moral (ethical) decision where personal benefits
come at the expense of impersonal others or
society as a whole (Kaplan et al., 1997).1
Therefore, increasing the level of income tax
compliance, i.e., tax ethics, is an important issue
for the tax authority of countries where a large
proportion of government spending is funded by
taxes collected. However, many tax authorities
have a limited budget to spend on the various
programs designed to increase tax compliance
such as aggressive enforcement measures and
friendly persuasive techniques. Therefore, discovering which approach or combination of
approaches is more effective is therefore important for the efficient use of the limited budget
in increasing individuals tax ethics.
Aggressive enforcement measures include
auditing selected taxpayers and prosecuting
alleged tax evaders. Friendly persuasion refers to
appeals or influences and may include normative appeals and appeals that promote compliance
out of self-interest by providing information
about the personal consequences to the individual consequent to noncompliance, whether
social, ethical, or legal. While aggressive enforce-

Journal of Business Ethics 47: 133145, 2003.

2003 Kluwer Academic Publishers. Printed in the Netherlands.

134

Janne Chung and Viswanath Umashanker Trivedi

ment measures have traditionally been used to

increase compliance, the use of friendly persuasion on increasing taxpayers ethics has generated
some research interest recently. Notwithstanding
that these studies (McGraw and Scholtz, 1991;
Blumenthal et al., 2001) report that friendly
persuasion does not increase compliance, but
given its relative economy and the limited compliance budgets of many tax authorities, friendly
persuasion warrants further investigation. The
first purpose of the present study is to examine,
using a laboratory experiment, the effect of
friendly persuasion on individuals tax ethics
using another set of techniques by requiring
individuals to first write reasons explaining why
they should report their income fully and second
read a list of reasons on why they should do so.
Generation of reasons involves individuals writing
reasons that either speak for or provide evidence
for the chosen behavior, or speak against or point
against the alternative behavior(s) rejected (Koriat
et al., 1980) and has been shown to be effective
in changing behavior in the cognitive psychology
and accounting literatures (e.g., Hoch, 1984;
Koonce, 1992). This study attempts to extend the
use of generating and reading reasons to a tax
ethics situation to discover whether these actions
will affect individuals tax compliance behavior.
The second purpose of this study is to attempt
to explain why the use of friendly persuasion did
not significantly change behavior in Blumenthal
et al.s (2001) and McGraw and Scholzs (1991)
studies. Due to the nature of their data, neither
of these studies was able to control for gender
even though the tax literature reports a gender
effect on tax compliance behavior and taxpaying
attitude (e.g., Scholz and Pinney, 1995; Beron
et al., 1992; Kinsey, 1992; Smith, 1992;
Steenbergen et al., 1992; Spicer and Hero; 1985).
Because of womens growing importance in
many economies and as taxpayers, we examine
whether the effect of friendly persuasion on taxpayers income reporting decisions is gender
specific.
This study contributes to the literature by
examining the use of another friendly persuasive
technique generating and reading reasons
supporting compliance on tax compliance. The
results from this study also provide a better

understanding of the condition under which the

use of friendly persuasion is effective by examining whether it has gender effects and may
explain why prior studies report friendly persuasion as ineffective on compliance behavior.

II. Theory development

Friendly persuasion
In countries where a large proportion of government spending is financed by taxes collected
from its people, increasing the level of tax
compliance among individuals is an important
concern of their tax authority. This issue is
especially important for the Canada Customs and
Revenue Agency (CCRA) given the size of the
amount spent by the Government of Canada on
welfare measures and social assistance. However,
underreporting of income and taxes remains a
significant problem in Canada particularly among
low-income individuals including students and
the young (Hill and Kabir, 1996).
In an attempt to understand tax compliance
behavior, prior studies show that individuals tax
ethics are influenced by various factors (e.g.,
Alm, 1991), and the extant tax compliance literature reports that individuals pay taxes because
they fear detection and punishment. Based on
this finding, traditionally, tax authorities (e.g., the
Internal Revenue Service of the U.S.) have used
aggressive enforcement measures such as audits
and prosecution of tax evaders as a way of
encouraging tax compliance. Recent research
studies and the tax authorities own data show
that increasing audits and penalties do little to
increase compliance if taxpayers are ignorant of
these changes (e.g., Scholz and Pinney, 1995).
Furthermore, taxpayers as well as policymakers
are increasingly averse to overtly intrusive and
coercive methods of increasing compliance like
increasing the audit rate, as evidenced by the
steady decline in the audit rate over the past
decade and a half as well as resistance to the
administration of later installments of the intense
and random audits conducted under the Taxpayer
Compliance Measurement Program in the
United States. This being the case, tax authori-

The Effects of Friendly Persuasion and Gender on Tax Compliance Behavior

ties would benefit from understanding what the
other determinants of compliance behavior are,
especially given their limited enforcement
budgets.
One approach examined in the tax compliance
literature is the use of friendly persuasion
(Blumenthal et al., 2001; McGraw and Scholz,
1991). This literature reports that individuals are
motivated by two competing approaches to tax
compliance a normative approach that emphasizes moral reasoning about the appropriateness
of norms and principles related to a perceived
legal obligation and asks the question What
should I do? or What am I obligated to do?
(McGraw and Scholz, 1991, pp. 471472). In
contrast, the personal consequences approach
emphasizes the trade-off between the cost of legal
sanctions and the gain from underreporting and
asks the question What will make me better
off?
Based on theories proposed by Rest (1984) for
ethical decision-making, McGraw and Scholz
(1991) propose a four-stage choice process for
tax-paying decisions. First, the individual
identifies whether the decision context involve
a moral or ethical principle; second, s/he formulates a course of action that applies the
moral/ethical principle; third, the competing
values (goals) and costs are weighed and a course
of action that is consistent with these competing
values and costs is contemplated; and fourth, the
chosen action is implemented. It is during the
critical third stage that moral/ethical principles
and/or personal consequences are considered
before a decision on a course of action is reached
(McGraw and Scholz, 1991). This principle gave
rise to a line of research that examines whether
messages emphasizing norms and personal consequences will affect income tax-reporting
behavior. This area of research is also based on
the principle of cognitive consistency that states
that an individual usually has attitudes that are
consistent with related attitudes and there is also
a consistency between the individuals attitudes
and his/her behavior (Cialdini, 1989).
Consequently, a change in attitude(s) would
result in a change in behavior. Cialdini (1989)
proposes that taxpayers attitudes can be changed

135

through friendly persuasion in an attempt to

affect their compliance behavior.
In response to this, McGraw and Scholz
(1991) with the cooperation of the Internal
Revenue Service (IRS) of the U.S. solicited the
participation of taxpayers prior to them filing
their 1987 tax return. The taxpayers were
randomly assigned to one of three groups. The
first group watched a video that emphasized the
normative principles for compliance. The second
group watched a video that emphasized the
personal consequences of compliance. A third
group acted as control. The IRS provided the
researchers with data for both the 1985 and 1987
tax returns as well as the differences between
these two years. The results show that the
messages contained in the videotapes were
successful in changing attitudes but these attitude
changes did not result in increased compliance.
In explaining their results, McGraw and Scholz
(1991) conclude that impacts on compliance are
difficult to achieve.
Blumenthal et al. (2001) extend McGraw and
Scholzs (1991) study with the collaboration of
the Minnesota Department of Revenue. Instead
of requiring taxpayers to watch a video, they sent
letters to three large groups of taxpayers at the
beginning of the 1994 tax season. One group
received a letter that described the services that
were paid for by taxes while a second group
received a letter that refuted the claim that
cheating on taxes is widespread. A third group
acted as control. The Minnesota Department of
Revenue provided tax return data for the tax
years 1993 and 1994. The results show little or
no evidence that either type of friendly persuasion affected behavior. In explaining their results,
the authors suggest that either the appeals did not
have any effect on taxpayers attitudes, or the
resultant attitude changes did not translate to
behavioral changes, or the attitudes that were
changed were not related to tax compliance
behavior. While these two studies do not find
friendly persuasion to have an effect on taxpayer
behavior, the relative economy of its use warrants
further investigation of its effectiveness.
As beliefs are the subparts of an attitude (Sirsi
et al., 1996),2 a technique that has often been
used in the cognitive psychology and accounting

136

Janne Chung and Viswanath Umashanker Trivedi

literatures to change behavior through changing

beliefs is to require individuals to generate
reasons that either speak for, or provide evidence
for, the chosen alternative, or to generate reasons
that speak against, or point against, the alternative(s) rejected (Koriat et al., 1980). A review of
these literatures suggests that the generation of
such reasons may affect subsequent judgment.
Having generated reasons to support a target
action, an individual is likely to judge the target
action as more likely to occur relative to an alternative action. For example, Koonce (1992)
divided auditor participants into two groups
control and treatment. She gave both groups of
participants case materials that comprised financial and non-financial information about the
client and the clients industry and detailed
discussions with the client identifying the change
in sales mix as the cause of fluctuations detected
by analytical procedures. After reading the case
materials, the treatment participants generated
reasons explaining why the detected fluctuations
could be caused by the change in sales mix. The
control participants did not generating supporting
reasons. Subsequently, all participants assessed the
likelihood that the fluctuations were caused by
the change in sales mix. The results show that
generating reasons supporting a target event leads
to higher likelihood judgments of the target
event.
In addition, the requirement to generate
reasons supporting a target action may interfere
with an individuals ability to consider alternative action(s) that will, in turn, affect his/her
likelihood judgments of the target action (Hoch,
1984). Hoch (1984) required a group of participants to generate reasons why a future event
might and might not happen. A second group
generated reasons in the reverse order. The results
show that the first set of reasons generated by
both groups was generated in greater quantity
and less time compared to the second set of
reasons. It appears that generating the first set of
reasons interferes with an individuals ability to
generate the second set of reasons. When used
as a friendly persuasive technique to affect tax
ethics, generating reasons supporting compliance
should increase the salience of compliance beliefs,
reduce non-compliance beliefs and attitudes, and

increase compliance behavior. However, many

individuals may be able to produce only a partial
list of the reasons that justify compliance.
Therefore, the compliance increasing impact of
generating supporting reasons on individuals
compliance behavior could be enhanced by
requiring these individuals to read a potentially
more comprehensive list of reasons justifying
compliance. Therefore, we propose the following
hypothesis in the alternate form:
H1 Participants who generate and read
reasons supporting compliance will report
higher income compared to participants
who did not generate and read reasons
supporting compliance.

Gender
Many studies report that women, compared to
men, are more ethical in terms of their tax
compliance behavior and taxpaying attitudes
(e.g., Scholz and Pinney, 1995; Beron et al.,
1992; Kinsey, 1992; Smith, 1992; Steenbergen et
al., 1992; Spicer and Hero; 1985). This difference is observed across studies that made use of
experimental, archival, and survey data. For
example, an experimental economics study with
ten income reporting rounds and the possibility
of an audit at the end of each round shows that
men tend to evade more taxes than women do
(Spicer and Hero, 1985). Another experimental
study reports that compared to men, women
perceive the probability that the IRS would
discover underreported income to be higher
(Steenbergen et al., 1992). In response to a
question on the severity of the penalty the IRS
would impose if an underreported income of
$500 were detected, women rated the severity
of the penalty higher than the men did (Scholz
and Pinney, 1995).
Data provided by the IRS show that, except
for low-income wage and salary workers, geographical areas where there was a relatively large
number of female-headed families reported
higher adjusted gross income compared to areas
where the proportion of male-headed families
were higher. This improved compliance rate was

The Effects of Friendly Persuasion and Gender on Tax Compliance Behavior

a result of these female-headed households
reporting higher income on their tax returns
(Beron et al., 1992).
Survey data show that compared to men,
women (i) are more likely to consider the tax
system to be fairer (Smith, 1992; Kinsey, 1992),
(ii) think that people who cheat on their taxes
(whether the amounts are large or small) are
more likely to get caught (Smith, 1992; Kinsey,
1992), (iii) are less likely to have underreported
their income, and when they do underreport,
such underreported amounts are smaller (Smith,
1992), and (iv) think that the penalty for noncompliance they would receive would be more
severe (Kinsey, 1992). Consistent with the literature, we expect a gender difference in income
reporting behavior between men and women.
Therefore, we propose the following hypothesis
in the alternate form:
H2 Women will report higher income
relative to men.
Interaction between friendly persuasion and gender
Friendly persuasion affects attitudes in order to
affect behavior. Friendly persuasion affects
behavior by affecting an individuals beliefs and
interfering with his/her ability to consider alternative(s). Used in an income tax-reporting
context, it should strengthen beliefs in the norm
and an individuals ability to consider the positive
(negative) personal consequences of complying
(evading) while interfering with the individuals
ability to consider the positive (negative) personal
consequences of evading (complying). In the
following discussion, we offer three theories that
explain why women and men may be affected
differently by friendly persuasion.
First, gender socialization theory (Davenport
and Yurich, 1991; Chodorow, 1978) suggests that
men and women may be affected differently by
the friendly persuasion treatment. The normative appeals contained in this treatment are more
consistent with womens communal orientation
and act by focusing their attention on the benefits
to society of compliance. They appeal to
womens ethical reasoning that is an inherent
aspect of their communal concerns. Under-

137

reporting of their income for tax purposes

contradicts these goals. In addition, the personal
consequences of compliance (non-compliance)
are consistent (inconsistent) with womens image
of themselves as nurturing individuals and would
make the importance of being compliant more
salient. The friendly persuasion treatment is not
expected to have the same effect on mens
income tax-reporting behavior. In the context of
tax paying, emphasizing normative appeals and
drawing their attention to the benefits to society
of full compliance may be inconsistent with
mens agentic goals. This is because the benefits
are defused and are often indirect. Additionally,
the appeals related to the personal consequences
of compliance or non-compliance would not
affect men to the same degree as they affect
women.
Second, if taxpaying is viewed as a special form
of gambling with the underreported tax being
the gain and the penalty imposed by the tax
authority being the loss, then risk aversion is a
relevant attitude in compliance behavior. The
cognitive psychology literature consistently
reports a gender difference in risk aversion, that
is, women are more risk averse relative to men
(e.g., Shavit et al., 2001). Women who are
exposed to the friendly persuasion treatment
would be sensitized to the risk of underreporting
that could bring about legal and social sanctions
while at the same time drawing their attention
to the benefits of full compliance by appealing to
their social conscience as well as to other potential benefits of full compliance. Thus, friendly
persuasion should have a positive impact on
womens compliance. In contrast, this treatment
may not impact mens compliance behavior to
the same degree because men are less risk-averse
and information sensitizing them to the risk of
underreporting should have less of an impact on
them. In fact, in our manipulation (see below),
such sensitization may have helped to highlight
the significant benefit and the insignificant cost
of non-compliance to the men in the friendly
persuasion condition, when they were informed
subsequently that the audit rate was set at a
relatively low four percent. McGraw and Scholz
(1991) also attribute their results to womens risk
aversion.

138

Janne Chung and Viswanath Umashanker Trivedi

A third view of tax evasion is that it is a crime

and the laws of some countries (e.g., Canada and
the U.S.) treat tax evasion as a criminal offence.
The criminal justice literature reports that men
are more likely to engage in criminal activities
than women are (Dezhbakhsh and Rubin, 1998).
In our study, friendly persuasion may make the
risks and the criminal nature of tax evasion salient
to the female participants, positively impacting
their compliance level.
Consequently, an interaction effect between
friendly persuasion and gender on income
reporting is expected. Women who are subject
to the friendly persuasion treatment will report
higher income compared to men who are subject
to the same treatment. Thus, we hypothesize in
the alternate form that:
H3 There is a significant interaction effect
between friendly persuasion and gender
on income reported.

III. Methodology
Procedures
A 2 2 full factorial experiment (friendly persuasion and control; men and women) was
carried out. At the start of the experiment, participants were provided with some introductory
information together with an assurance of confidentiality. Instead of endowing our participants
with income, we required them to earn the
income that was to be reported later for taxpaying purposes. This procedure was chosen
because it more closely reflects real life where
individuals are required to make decisions about
reporting their income to the tax authority. The
participants were required to complete two
questionnaires and were paid $30 for their
efforts.3
Afterwards, participants in the treatment group
were asked to produce reasons why they should
report their income in full and they were
required to write as many reasons as they could
think of. No time limit was placed on them.
When they had completed this task, they were
required to read a list of eight reasons containing

both normative/ethical and self-interested

reasons, on why individuals should report their
income fully. One of the researchers read aloud
from the list while the participants read it to
themselves. To ensure that they read the list
carefully, the participants were also required to
circle those reasons on the list that matched the
reasons generated by them earlier. Participants
in the control treatment did not perform these
tasks.
Next, all participants were briefed on how to
complete their income tax report and the audit
procedure. The tax rate was 40%, the audit rate
was set at 4%, and the penalty for underreporting
if they were audited was 150% of the unpaid tax,
payable over and above the unpaid tax. Again,
to better reflect behavior in the real life, the
participants were asked to report as if they were
in a real tax reporting situation. Specifically, the
participants were asked to report the amount of
income that they would report if they were
placed in a similar tax reporting situation in real
life, i.e. if they were to face a tax rate of 40%,
an audit rate of 4% etc. In addition, they were
asked to assume that the tax authority would not
know what their income was unless they were
later audited. After they completed and submitted
their income tax report on the amount of income
they earned, and tax payable thereon, the audit
procedure was conducted. A bag containing 25
numbered balls was used for this purpose. A ball,
chosen randomly by one of the participants
beforehand, was designated the audit ball
(resulting in an audit rate of 4%). If the audit ball
was drawn, all participants in the room were
audited. If the audit ball was not drawn, no one
in the room was audited. After the completion
of the audit procedure, the participants provided
some demographic data (including their gender)
and were paid ($30 minus tax at 40% on income
reported, and penalty if any) and dismissed.
When all the experiments were completed, the
participants were sent a debriefing letter that
described the purposes and the results of the
study.

The Effects of Friendly Persuasion and Gender on Tax Compliance Behavior

Participants
Hill and Kabir (1996) report that participation in
the underground economy and by extension tax
evasion in Canada is especially high among lowincome individuals (including those who were in
school, unemployed, and receiving social assistance), youths, and unmarried persons. In this
context, Jackson and Milliron (1986) observe that
tax evasion studies are better off focusing on
groups of interest rather than trying to obtain a
diverse set of participants. Consequently, our
study focuses on young low-income individuals.
To recruit participants for this study, we sent
recruiters to a large Canadian university. To
ensure that a broad cross-section of students was
recruited, recruiters were stationed at the universitys main thoroughfare through which most
students must pass to access the teaching rooms,
the main library, public transport, and the food
halls. Participants must have filed at least one tax
return and be an undergraduate of the university to qualify for our study. One hundred and
thirteen participants were recruited. One participant submitted an incomplete questionnaire and
this resulted in 112 useable questionnaires comprising 51 men and 61 women. Their demographic data, including descriptive statistics and
correlations are shown in Table I. Table I shows
that our participants on average filed 2.7 tax
reports suggesting that they had a fair amount
of tax-filing experience, thereby reinforcing their
appropriateness as participants in a tax ethics
study.

IV. Results
ANOVA analysis with treatment and gender as
main effects was carried out to test the
hypotheses in this study. Initially, the correlations
in Table I, panel B were examined to determine
which control variables should be included in the
model. The results in Table I, panel B show that
significant correlations exist between country of
birth4 and annual income5 and the dependent
variable amount of income reported.
Therefore, following Tabachnick and Fidell
(1996), these two variables were included in the

139

model as control variables, specifically as fixed

effects.6 The ANOVA analysis is shown in Table
II, panel A and the means (SDs) are shown in
Table II, panel B.7
While country of birth did not significantly
influence the amount of income reported, the
annual income earned by the participants had a
significant impact on the amount of income
reported. The insignificant result of the country
of birth variable suggests that new immigrants
to Canada may not necessarily be less compliant
than Canadians consequent to home country
norms as some people apprehend. The negative
correlation between annual income and amount
of income reported suggests that participants with
higher annual income were less compliant than
participants with lower annual income. The treatment main effect was not significant. The income
reported by those who generated and read
reasons supporting compliance (mean $13.10
[sd $18.83]) was not significantly higher than the
income reported by those who did not generate
and read reasons supporting compliance (mean
$14.82 [sd $16.41]) (Table II, panel B). Hence,
H1 is not supported.
The gender main effect was however significant with women reporting significantly higher
income (mean $17.58 [sd $20.22]) compared to
the income reported by men (mean $10.34
[sd $15.58]) (Table II, panel B). This result
suggests that women have higher tax ethics than
men do consequently supporting H2.
There was a significant interaction effect
between treatment and gender, which is depicted
in Figure 1. Related multiple comparisons are
provided in Table II, panel C. Women in the
friendly persuasion treatment reported significantly higher income compared to men in the
same treatment (t = 3.21, p < 0.001). In fact,
while friendly persuasion appears to have a compliance increasing effect on women, it appears
to have a compliance decreasing effect on men.
However, both these effects are not statistically
significant by themselves (Table II, panel C).
None of the other comparisons are significant,
consequently H3 is partially supported.
Next, we examine whether the reasons generated by the participants provide further insight
into their income tax-reporting behavior. The

140

Janne Chung and Viswanath Umashanker Trivedi

TABLE I
Demographic data

Panel A: Descriptive statistics

Variable

Level

Average/n (sd)

Age in years

21.66 ( 3.06)

Gender

Female
Male

51
61

Work experience in months

44.98 (32.47)

Number of tax reports filed

2.70 (1.92)

Number of hours of paid work per week

14.34 (11.59)

Annual income

> $5,000
$5,000$9,999
$10,000$14,999
$15,000$19,999
$20,000$25,000
> $25,000
Missing data

57
35
11
2
6
0
1

Marital status

Never married
Common law relation
Married
Divorced

100
6
5
1

First language

English

60

Non-English

52

Canada
Outside Canada

52
60

Country of birth

Panel B: Correlations
Income
reported

Age
0.085
Marital status
0.113
Number of
children
0.039
Work experience 0.031
Tax report filed 0.112
Hours of paid
work
0.071
Annual income 0.215**
Country of
birth
0.231**
First language
0.103

Age

Marital
status

Number Work
Tax
of
experience report
children
filed

Hours
of paid
work

Annual
income

Country
of birth

0.626***
0.597*** 0.516***
0.507*** 0.171* 0.297***
0.306*** 0.011
0.021 0.634***
0.254***
0.169*

0.123
0.010

0.165*
0.086

0.182*
0.198**

0.209**
0.218**

0.154
0.179*

0.226**
0.150
0.330*** 0.505***
0.141
0.089

0.208**
0.155

0.250**
0.129
0.010

0.220**
0.099 0.638***

***, **, and * correlation is significant at the 0.01, 0.05, and 0.10 level respectively (2-tailed).

The Effects of Friendly Persuasion and Gender on Tax Compliance Behavior

141

Figure 1. Graph showing interaction effect between the treatment groups and gender.

authors independently coded the reasons generated into eight categories legal sanction,
ethical reasons, benefits provided, social
conscience, personal financial benefits
provided, social sanction, other benefits,
and other reasons.8 Differences were reconciled
through discussion and reference to the
literature. Correlation analyses were performed
between the various categories of reasons and
gender, and apart from other reasons (p =
0.021) and ethical reasons (p = 0.076), there
were no significant correlations between the
number of these reasons and gender. Men
provided 0.56 (sd 0.71) ethical reasons and
0.20 (sd 0.50) other reasons whereas women
provided 1.00 (sd 1.06) ethical reasons and no
other reasons. These results suggest that
womens higher reported income may be partly
explained by their relatively more ethical nature
as reflected by the greater number of ethical
reasons they generated. However, this should be
interpreted with caution because our friendly
persuasion treatment comprised two parts first
generating and second reading reasons supporting
compliance and the number of reasons generated by the participants form only one part of
the treatment.

V. Discussion and conclusion

The purposes of this study are twofold: to reexamine the effectiveness of friendly persuasion
on tax ethics by requiring participants to generate
and read reasons supporting compliance, and to
examine whether the effectiveness of friendly
persuasion has gender effects. To achieve the
above objectives the study examined the income
reporting decisions of a group of low-income
participants with such decisions having real
economic consequences to them. Therefore, the
following discussion is subject to the caveat that
it may not be reflective of the tax compliance
behavior of higher income individuals. First, we
find a gender main effect women exhibited
higher tax ethics in that they reported higher
income relative to men. This may have important policy implications given the increasing role
of women in the economy and their growing
importance as taxpayers. Second, we do not find
a significant treatment main effect participants
in the friendly persuasion treatment did not
report higher income compared to participants
in the control treatment. The lack of a significant treatment main effect is consistent with the
findings in McGraw and Scholz (1991) and
Blumenthal et al. (2001). These findings suggest

142

Janne Chung and Viswanath Umashanker Trivedi

TABLE II
ANOVA results, means (SDs), and planned comparisons

Panel A: Tests of between-subjects effects)

Dependent variable: Income reporteda
Source
Control variables:
Annual income
Country of birth
Main effects:
Treatment
Gender
Interaction effect:
Gender Treatment
Error
Corrected total

SS

df

Sig.

01,654.60
00,231.15

004
001

3.198
1.787

0.016
0.184

000,76.87
01,189.13

001
001

0.594
9.194

0.442
0.003

00,948.31
13,321.45
18,182.99

001
103
111

7.332

0.008

R Squared = 0.267 (Adjusted R Squared = 0.210).

Panel B: Means (SDs) and cell size
Gender

Treatment
Friendly persuasion

Control

Total

$6.50
($14.25)
(n = 25)
A
$19.71
($17.00)
(n = 35)
C
$13.10
($18.83)
(n = 60)

$14.18
($12.89)
(n = 26)
B
$15.45
($15.42)
(n = 26)
D
$14.82
($16.41)
(n = 52)

$10.34
($15.58)
(n = 51)

Men

Women

Total

$17.58
($20.22)
(n = 61)
$13.96
($22.13)
(n = 112)

Panel C: Planned comparisons

Comparison
Cell
Cell
Cell
Cell
Cell
Cell
a

A Cell C
D Cell A
B Cell A
C Cell B
C Cell D
D Cell B

Mean ($) (I)

Mean ($) ( J)

Diff. ($) ( JI)

SE

06.50
15.45
14.18
19.71
19.71
15.45

19.71
06.50
06.50
14.18
15.45
14.18

13.20
08.95
07.68
05.53
04.26
01.27

4.12
4.24
3.89
3.89
4.24
4.02

Income earned was $30.

*** Significant at = 0.01 controlling for multiple comparisons using the Bonferroni procedure.

t-value
3.21***
2.11
1.97
1.42
1.00
0.32

The Effects of Friendly Persuasion and Gender on Tax Compliance Behavior

that friendly persuasion may not increase the tax
compliance behavior of the general population.
Third, we found a significant interaction effect
between treatment and gender. Specifically, we
find that while friendly persuasion in the form
of generating and reading reasons supporting
compliance has a slightly positive impact on
womens tax ethics it in fact has a slightly
negative impact on mens tax ethics. This is
because we were able to control for gender that
the previous studies could not. This finding
explains the lack of an overall main effect of
friendly persuasion in our study and may also
explain the lack of a main effect of friendly persuasion on the general population in prior
studies. In conclusion, our study adds to the
literature by reporting that while friendly persuasion may positively influence womens tax
ethics it may not have a similar influence on
mens tax ethics.
We attribute the difference in behavior
between the genders in our study to three different theories. There may be a fourth explanation, that is, tax compliance is a moral action,
and moral actions are susceptible to social
desirability bias. As women report higher social
desirability bias than men do (Chung and
Monroe, 2003), it is possible that the experimental and survey data reported in the literature
may be affected by womens higher social desirability bias. However, archival studies using data
provided by the IRS (e.g., Beron et al., 1992)
have also found women to be more compliant
than men. The social desirability bias scale
obtained from one of the filler questionnaires our
participants completed to earn income does not
significantly correlate with the participants compliance decisions. In addition, we paid our participants based on the amount of income they
reported. Thus appearing compliant because it
is socially desirable has economic costs to them.
After considering all the above factors, we
discount social desirability bias inducted in the
laboratory as a reason for our results. The studies
reported in this paper found that women are
more likely than men to both consider the tax
system to be fairer and to estimate the penalties
for noncompliance to be more severe. It is
unclear which of these perceptions is driving

143

their compliance behavior. Future research could

try to distinguish between the effects of these two
perceptions on womens compliance behavior.
As this is the first study to show an interaction effect between friendly persuasion and
gender on income reporting behavior, more
research has to be conducted before firm conclusions can be drawn. Specifically, in the present
study the impact of friendly persuasion on taxpayers ethics was examined by requiring participants to both generate and read reasons
supporting compliance. Thus, a deliberate
confound was created between the impact of
generating and the impact of reading reasons so
as to create a strong manipulation effect for
friendly persuasion and consequently to detect its
effect on taxpayers compliance decisions. This
was done consequent to the fact that prior studies
by Blumenthal et al. (2001) and McGraw and
Scholz (1991) had failed to detect a significant
impact of friendly persuasion on individuals tax
compliance decisions. Given the significant interaction effect detected between friendly persuasion and gender in our study, future research can
now attempt to identify the reason(s) for the
interaction effect generating or reading reasons
supporting compliance. Future research could
also use actual tax compliance data to re-examine
this interaction effect. If future research reports
a similar gender effect on other friendly persuasion techniques, then this could have implications
for tax policy-makers. One question they would
face is how to increase the effectiveness of
friendly persuasion techniques on men. The
answer to this question is especially important
given that in most countries, mens income is
higher than womens. It could be that friendly
persuasive techniques work on men only in a
context where the personal benefits (costs) of
compliance (noncompliance) are high. Thus, the
presence of strong deterrent measures like a relatively high audit rate or penalty rate may be
prerequisites for friendly persuasive techniques to
have a positive impact on mens compliance. This
suggests that both positive reinforcements and
deterrent measures are required and may act as
compliments to each other to maintain and
increase the overall level of compliance in society.
If this conclusion holds true friendly persuasion

144

Janne Chung and Viswanath Umashanker Trivedi

techniques can be used to increase compliance in

the overall population. Conversely, if friendly
persuasion techniques continue to be effective
only on women, tax authorities may be better off
focusing on other techniques to increase compliance, especially those that increase compliance
among men given the fact that, based on tax data,
women are already more compliant than men.
This study made use of an audit rate of four
percent. The general taxpaying population in
Canada is presumably not very informed either
of the overall audit rate, which is slightly above
one percent, or whether the audit rate varies
between different target groups, given CCRAs
practice of not publicizing such information
widely. Therefore, it is unclear what effect, if any,
the use of four percent as the audit rate will have
on the generalizability of these results.

Country of birth was coded 0 for Canadianborn and 1 for non-Canadian-born.

5
There was one missing data point for annual
income and this was filled in with the mode.
6
When too many covariates are used and they are
correlated with each other, a point of diminishing
returns in adjustment of the DV is quickly reached.
Power is reduced because numerous correlated covariates subtract degrees of freedom from the error term
while not removing commensurate sums of squares
for error. (Tabachnick and Fidell, 1996, 350). To
prevent this from occurring, we performed correlation analyses between the DV and different demographic variables and included in our ANOVA model
those variables that have a significant relation with the
DV as control variables.
7
The assumption of homogeneity of variance was
not violated at the p = 0.05 level.
8
The other reasons category was used as a catchall category for reasons that did not fit into any of
the other seven categories.

Acknowledgements
The first author gratefully acknowledges the
partial financial support provided by York
University. The authors thank Bram Cadsby,
Linda Thorne, and participants at the AAA/ABO
Conference 2002, European Accounting
Associations 2003 Annual meeting, and the
Canadian Academic Accounting Associations
2003 Annual meeting for their helpful comments
on an earlier version of the paper.
Notes
1

While fundamental questions may be raised about

the ethicality of taxation and whether the tax system
of a country is just and fair, such examinations are
beyond the scope of this paper. Instead, this paper
argues that tax evasion leading to direct and indirect
personal benefits from government programs at the
expense of others and society as a whole is unethical.
2
The words, beliefs and attitudes, are often used
interchangeably (see Burkhardt, 1994).
3
These were the Tolerance for Ambiguity and Social
Desirability Bias scales and were chosen because they
comprise true or false responses and require little time
to complete. In addition, because of their innocuous
nature, they are unlikely to affect the results of the
study.

References
Alm, J.: 1991, A Perspective on the Experimental
Analysis of Taxpayer Reporting, The Accounting
Review 66(3), 577594.
Beron, K. J., H. V. Tauchen and A. D. Witte: 1992,
The Effect of Audits and Socioeconomic Variables
on Compliance, in Joel Slemrod (ed.), Why People
Pay Taxes: Tax Compliance and Enforcement (The
University of Michigan Press, Ann Arbor), pp.
6789.
Blumenthal, M., C. Christian and J. Slemrod: 2001,
Do Normative Appeals Affect Tax Compliance?
Evidence from a Controlled Experiment in
Minnesota, National Tax Journal 54(1), 125138.
Burkhardt, M. E.: 1994, Social Interaction Effects
Following a Technological Change: A Longitudinal
Investigation, Academy of Management Journal 37(4),
869898.
Chodorow, N.: 1978, The Reproduction of Mothering
(University of California Press, Berkeley, CA).
Chung, J. and G. S. Monroe: 2001, A Research Note
on the Effects of Gender and Task Complexity on
an Audit Judgment, Behavioral Research in
Accounting 13, 111125.
Chung, J. and G. S. Monroe: 2003, Exploring Social
Desirability Bias, Journal of Business Ethics (in
press).
Cialdini, R. B.: 1989, Social Motivations to Comply:
Norms, Values and Principles, in J. A. Roth and

The Effects of Friendly Persuasion and Gender on Tax Compliance Behavior

J. T. Scholz (eds.), Taxpayer Compliance, Vol. 2,
Social Science Perspectives (University of Pennsylvania
Press, Philadelphia), pp. 220227.
Davenport, D. S. and J. M. Yurich: 1991,
Multicultural Gender Issues, Journal of Counseling
& Development 70 (September/October), 6471.
Dezhbakhsh, H. and P. H. Rubin: 1998, Lives Saved
or Lives Lost? The Effects of Concealed-Handguns
Laws on Crime, AEA Papers and Proceedings 88 (2),
468474.
Hill, R. and M. Kabir: 1996, Tax Rates, the Tax
Mix, and the Growth of the Underground
Economy in Canada: What Can We Infer,
Canadian Tax Journal 44(6), 15521583.
Hoch, S.: 1984, Availability and Inference in
Predictive Judgment, Journal of Experimental
Psychology: Learning, Memory and Cognition 10,
649662.
Jackson, B. R. and V. C. Milliron: 1986, Tax
Compliance Research: Findings, Problems, and
Prospects, Journal of Accounting Literature 5,
125161.
Kaplan, S. E., K. J. Newberry and P. M. J. Reckers:
1997, The Effect of Moral Reasoning and
Educational Communication on Tax Evasion
Intentions, The Journal of the American Taxation
Association 19(2), 3854.
Kinsey, K. A.: 1992, Deterrence and Alienation
Effects of IRS Enforcement: An Analysis of Survey
Data, in Joel Slemrod (ed.), Why People Pay Taxes:
Tax Compliance and Enforcement (The University of
Michigan Press, Ann Arbor), pp. 259285.
Koonce, L. 1992: Explanation and Counterexplanation During Audit Analytical Review, The
Accounting Review 67(1), 5976.
Koriat, A., S. Lichtenstein and P. Fischhoff: 1980,
Reasons for Confidence, Journal of Experimental
Psychology: Human Learning and Memory 6,
107118.
McGraw, K. M. and J. T. Scholz: 1991, Appeals to
Civic Virtue versus Attention to Self-Interest:
Effects on Tax Compliance, Law and Society Review
23(2), 209240.
Rest, J. R.: 1984, The Major Components of
Morality, in W. L. Jurtines and J. L. Gewirtz (eds.),
Morality, Moral Behavior, and Moral Development
(Wiley, New York).

145

Roth, J. A., J. T. Scholz and A. D. Witte: 1989,

Taxpayer Compliance. Volume I: An agenda for
Research (University of Pennsylvania Press,
Philadelphia).
Scholz, J. T. and N. Pinney: 1995, Duty, Fear, and
Tax Compliance: The Heuristic Basis of
Citizenship Behavior, American Journal of Political
Science 39(2), 490512.
Schwartz, R. and S. Orleans: 1967, On Legal
Sanctions, University of Chicago Law Review 34,
274300.
Shavit, T., D. Sonsino and U. Benzion: 2001, A
Comparative Study of Lotteries-Evaluation in Class
and on the Web, Journal of Economic Psychology
22(4), 483491.
Sirsi, A. K., J. C. Ward and P. H. Reingen: 1996,
Microcultural Analysis of Variation in Sharing of
Causal Reasoning about Behavior, Journal of
Consumer Research 22, 345372.
Smith, K. W.: 1992, Reciprocity and Fairness:
Positive Incentives for Tax Compliance, in Joel
Slemrod (ed.), Why People Pay Taxes: Tax
Compliance and Enforcement (The University of
Michigan Press, Ann Arbor), pp. 223250.
Spicer, M. W. and R. E. Hero: 1985, Tax Evasion
and Heuristics, Journal of Public Economics 26,
263267.
Steenbergen, M. R., K. M. McGraw and J. T. Scholz:
1992, Taxpayer Adaptation to the 1986 Tax
Reform Act: Do New Tax Laws Affect the Way
Taxpayers Think about Taxes, in Joel Slemrod
(ed.), Why People Pay Taxes: Tax Compliance and
Enforcement (The University of Michigan Press Ann
Arbor), pp. 937.
Tabachnick, B. and L. Fidell: 1996, Using Multivariate
Statistics (Harper Collins College Publishers, New
York, NY).

Schulich School of Business,

York University,
4700 Keele St.,
Toronto ON M3J 1P3,
Canada,
E-mail: jchung@schulich.yorku.ca