continued from previous post
Improving the Model Further
Though the adjusted R-squared value is low (which is the case when using a dummy variable for the dependent variable) improving the regression is still possible.
Other factors that may be relevant include the type of industry that a firm is a member of; by including these variables into the last regression above, the regression model becomes:
Including these explanatory dummy variables improved the adjusted R-squared slightly, but did not affect the probability of a lawsuit, i.e. it remains (.13 probability of getting sued if all industries are relevant.))))) with a 95% confidence interval of (-0.39, 0.51). It did reduce the coefficient of BIG5 so that having a Big 5 auditor only increases the probability of a lawsuit by 1.3%, and it increased the coefficient of LOGSALES such that the probability of a lawsuit is raised from 1.6% to 1.8%.
Probability of lawsuit if in chem. Ind=.06:::::: Prob. Of lawsuit if in comp ind = .08 Prob of lawsuit if in elec=.05
Denis these are the three equations for each individual with average log sales and industry specific.
95% confidence intervals for each probability.
This is probably the best probability ::: 8 % probability of getting sued if you are in the computer industry (Not required financial risk indicators (random company in the computer industry with high sales) and are represented by a big 5 firm.
We should write a mini conclusion here stating that this is a good regression etc and what the coefficient
To understand other interactions that may increase the probability of getting sued by rule out big5 accounting association we further investigate other variables.
We then considered a firm’s undiversified risk relative to the market (its Beta), unrelated to the market (its market returns not explained by Beta) and the variation of the firm’s return during the period observed. These variables were obtained during these three years and are tabulated as BETA, CUMRET, and SIGMARET, respectively. Adding these explanatory variables to the last regression, we obtained: