This preview shows page 1. Sign up to view the full content.
Unformatted text preview: agement Integrity on Audit Planning and Evidence 63 information is generally more costly to collect) by going outside the client for data veriﬁcation instead of simply increasing the analysis of the client’s information. Equations (1)–
(4) were also run including two additional control variables YR (the year of the observation)
and ROA (return on assets, included as a proﬁtability measure). The results are robust to
the inclusion of these variables. The variables are not signiﬁcant in the regressions and
therefore are not included in the primary analysis.
Management Integrity and Audit Differences
Table 4 presents logistic regression results testing H3. The coefﬁcient on MI is signif0.001) suggesting that MI is associated
icant and negative in Equation (5) ( 4.332, p
with the likelihood of discovering audit differences. As discussed earlier this ﬁnding may
be the by-product of more diligent testing for low-integrity assessments, so we attempt to
control for the extent of testing to isolate the relationship between misstatement discovery
and integrity assessments. Equation (6) includes PERSUASIVENESS, TIMING, and EXTENT as independent variables to control for the potential effect of increased audit effort
on misstatement discovery. In this equation, MI remains signiﬁcant ( 9.891, p
This evidence provides support for H3. It appears that although the RMM and management
integrity assessments are strongly inﬂuenced by PYERR, the MI assessment outperforms
PYERR as a signiﬁcant predictor of discovering differences in the current year.
We examine the sensitivity of our results to (1) our use of a combined risk measure,
(2) the use of two different metrics for persuasiveness, (3) a scaled measure of the magnitude of audit differences, (4) the inclusion of a fraud risk assessment variable, and (5)
the inclusion of an indicator variable for the presence of prior-year misstatements.
We rerun all of our empirical tests using either IR or CR in place of RMM. The untabled
results of using either IR or CR as the risk measure are consistent with the ﬁndings for H1...
View Full Document
- Spring '10