Broadway shows, part 6.

We really should have examined the residuals. Here is a scatterplot of the residuals from the regression of Exercise 14.

a) Which assumptions and conditions for regression can you check with this plot? What do you conclude?
Perhaps we should re-express Gross revenue. Here is a new regression and residual plot using the logarithm of Gross:
Response variable is: LogGross R squared = 96.0% R squared (adjusted) = 95.6%
s = 0.0571 with 33 - 4 = 29 degrees of freedom Variable Coefficient SE(Coeff) t-ratio P-value Intercept 1.28979 0.1092 11.8 60.0001 Attendance 0.154219 0.0225 6.85 60.0001 Playing Weeks -3.75769e-4 0.0002 -1.81 0.0807 New Productions 9.70391e-3 0.0026 3.72 0.0009

b) Would you prefer the second (re-expressed) regression model to the first? Explain all the reasons you reach this conclusion.

c) If you wished to predict Gross revenues for a year with larger attendance than has been seen thus far, which model would you choose? Why?

Transcribed Image Text:

Residuals -200 200 0 300 600 900 Predicted