Table 7.

Model coefficients of the mediation effect of HM on ∆TaxRisk via ∆Certainty

	Dependent variables
Variables	(1) ∆TaxRisk	(2) ∆Certainty	(3) ∆TaxRisk
Direct effects
HM	−1.51**    (0.60)	2.67***    (0.70)	−0.63    (0.65)
∆Certainty	–	–	−0.33**    (0.13)
Years	0.16    (0.10)	0.02    (0.12)	0.17*    (0.10)
Constant	−0.17    (0.92)	−0.56    (1.09)	−0.35    (0.86)
R2	0.32	0.34	0.43
Indirect effects
HM via ∆Certainty	–	–	−0.88**    (0.44)
Total effects
HM	−1.51**    (0.60)	2.67***    (0.70)	−1.51**    (0.60)

Coefficients and standard errors (in parentheses) were estimated using maximum likelihood estimation with robust standard errors. The dummy variable HM represents HM participation, ∆Certainty and ∆TaxRisk reflect perceived changes in tax certainty and tax risk, respectively. Years represents the timeframe for which participants were asked to indicate change. Direct effects correspond to parameter estimates in multiple regressions and represent the effects net of any effect by control variables included in the model. Indirect effects of HM represent the mediation effect and equals the product of the direct effects of HM on ∆Certainty (Column 2) and of ∆Certainty on ∆TaxRisk (Column 3). Total effects of HM represent the sum of direct and indirect effects and corresponds to the coefficient of HM in a regression on ∆TaxRisk without controlling for ∆Certainty (Column 1). ***, **, and * indicate two-tailed statistical significance at the 1, 5, and 10% level, respectively.