The results indicate that 1) traditional corporation have significant agency problems while the agency problems for high tech corporations are insignificant; 2) the pooled regression model for traditional corporations violates the assumption of
homoscedasticity because the residual values of small traditional corporations behave differently from those of large traditional corporations; 3) revenue growth rate is significantly and positively associated with profitability for both large and small high tech corporations; and 4) revenue growth rate is significantly and positively associated with profitability for small traditional corporations, but not for large traditional corporations.
Where assumptions of normality and
homoscedasticity were not met as required for parametric analysis, equivalent non-parametric tests were used.
Not surprisingly, White's test for
homoscedasticity shows that the squared residuals are much larger for countries with higher money growth and deads to rejection of he hypothesis of homoscedastic disturbances Likewise the Jarque-Bera test shows that the residuals are highly skewed so normality must be rejected.
Conditional
homoscedasticity and constant volatility is therefore strongly rejected for the time series.
With the transformed data, the Goldfeld-Quandt test did not reject the null hypothesis of
homoscedasticity (for the pre-TRA86 sample, largest F (43,43) = 1.
Data were transformed as necessary to satisfy the assumptions of normality and
homoscedasticity.
Conditional
homoscedasticity of covariances Assumption A3 is controversial and is not satisfied in many discrete choice models of schooling and labor earnings as we discuss below.
Scatterplots are also used for informal initial data inspection to reveal outlier presence, strong departures from linearity, or violations of
homoscedasticity.
The assumption of
homoscedasticity was tested with Cochran's test (at [Alpha] = 0.
Another assumption is
homoscedasticity of the OLS residuals, i.
For seven of 12 tests for
homoscedasticity (six floral traits examined for
homoscedasticity among weeks and among populations), the ratio of the highest to the lowest sample variance, [s.
Analyses for the Coopersmith indicated no violation of the underlying assumption of
homoscedasticity at the multivariate (Box's M, F = .
