Of course, multiple and contradictory discourses
circulate in a given society. The argument that underlies
H3 should therefore be understood in a relative sense –
not in an absolute one.
12 The first author, who is fluent in both French and
English, translated all of the questions (originally formulated
in English) into French. The French questionnaire
was subsequently checked for clarity by four individuals:
one director of the Ordre des comptables agre´e´s du
Que´bec; one retired CGA and his wife; and one academic
in sociology. Only minor modifications resulted
from their comments.
13 In Canada, all professions are regulated under provincial
jurisdiction. Accordingly, Canadian Chartered
Accountants are governed in every province by a regulatory
body (commonly named ‘‘provincial institute’’) that
has been delegated authority by the provincial government.
14 A copy of the survey instrument is available from
the first author.
15 A temporary cookie was placed on the respondent’s
computer once s/he initially answered the survey to ensure
that people would only complete the survey once.
16 The total number of respondents from Alberta is
288. At the time of data collection, the total population
of active CAs in Alberta was 7,579. The initial request
was sent to a sample of 911 subjects through e-mail or
fax on October 21, 2002. Three subsequent reminders
were sent to the total population of members (active and
inactive); the reminders were therefore sent to a much
bigger group than the initial request. The first reminder
(October 28) consisted of an advertisement put on the
Alberta Institute’s website, specifically on the ‘‘News
Briefs’’ page. The second reminder (November 7)
appeared on WebLink, which is the monthly guide of
the Institute to updates on its website. The third reminder
(December 13) was part of the Institute’s
monthly package sent either electronically or through
conventional mail to every member.
17 The lower response rate in Que´bec may be due to
various reasons – including doubt as to whether
researchers from English-speaking universities were able
to conceive a meaningful questionnaire in French. One
French-speaking respondent indeed sent an e-mail to us
after having completed the survey, highlighting that he
almost decided not to answer the survey for this reason,
and how surprised he was about the high quality of the
French questionnaire.
18 In these three institutes members voluntarily provide
the information on their hierarchical position. In
two institutes, data on hierarchical position was available
for about 30% of the membership. In the third institute
the corresponding proportion is 82%.
19 Since the surveys went out at different times for the
four provincial institutes, the first 25% and last 25% of
the respondents for each institute were identified. The
early respondent group was formed by combining the
four sets of first 25% responses while the late respondent
group was formed by combining the four sets of
last 25% responses.
20 Exceptions were that the late respondents were a
little older, held more senior level positions, and fewer
of them came from the medium size accounting firms.
When using another method comparing the means for
participants who responded within 20 days (therefore
around the time the second reminder was sent) of the
survey start date in the province, and for those who
responded afterwards, two of the significant differences
identified above disappeared, leaving only the medium
size accounting firm variable as significant. This seems
to suggest that there is unlikely to be serious systematic
bias due to non-response, especially if the non-response
group is compared with the larger sample rather than
just those who responded promptly.
21 For all the dichotomous variables in this study,
unless specified differently, a value of ‘‘1’’ is always
assigned to responses that fit the variable name description,
and ‘‘0’’ otherwise. For example, respondents in
public accounting practices will have a code of ‘‘1’’ for
public practice organization.
22 For example, recent data suggests that the capacity
for moral judgment in public accounting organizations
does not differ across hierarchical rank, thereby casting
doubt on a widely taken-for-granted assumption of
empirical literature in accounting ethics (Scofield et al.,
2004).
22 We also re-ran the regressions in Tables IV and V
by replacing the dummy variable province of work: Que´bec
with a dummy variable that takes into account the
language in which respondents answered the questionnaire
(French language = 1). The results are very similar
to those reported in Table IV. For Table V, the coefficients
are still similar although the p-value for the
language variable was slightly higher, at 0.06 (twotailed)
or 0.03 (one-tailed).