Ancillary Analyses—B&O Taxes
The results of ANCOVA analyses using the B&O tax base as the dependent variable are given
in Table 3. Of the two experimental treatments, only the Penalty is directly applicable to the B&O
tax. The Affidavit factor is technically only applicable to the use tax. Nonetheless, it is possible
that a "spill-over" effect occurs for the Affidavit manipulation on B&O tax payments. The cell
sizes for these two treatments are as follows: 647 for Penalty, 631 for No Penalty, 635 for
Affidavit, and 643 for No Affidavit. In addition to the two experimental treatments, the same
partitioned Revenue State factor is included in the model. The revenue state factor divided the
sample into 599 firms with declining revenues and 679 with increasing revenues.
The only significant main effect is for Revenue State (p = .000)—reported in Table 3, Panel
A—is not surprising since the B&O tax base is gross income. It is logical that firms whose income
has declined from last period to this period are reporting a lower B&O tax base than firms whose
income has increased as Table 3, Panel B indicates (means: decreasing 6.31 and increasing 7.02).
Therefore, this significance is of no practical importance other than to show that the partition was
effective.
As for the experimental variables. Table 3, Panel A reveals insignificant main effects for
Penalty (p = .547). Affidavit is also not significant (p = .785). Therefore, there is apparently no
spillover accountability effect and the sanction information provided in the educational letter does
not significantly increase the B&O tax base reported on the Combined Excise Tax Report. However,
we find a significant three-way interaction of Affidavit, Penalty, and Revenue State (p
= .045). Since this interaction was not predicted, any discussion of this finding would be mere
speculation.