Gate 1 



Gate 1 provides very little information that can be used 

 to develop the equation. Only general site information 

 (slope and elevation) and early volume estimates are 

 known at this time. It is therefore difficult to predict 

 group membership at this point. 



EASTSroE EQUATIONS 



The gate 1 eastside equations and classification results 

 are found in table 3. In general, the eastside equations 

 indicate that sale size (TOTVOL and TOTSALE) has a 

 positive influence on salability meaning that sales with 

 larger volumes and of larger size increase the likelihood 

 of selling. The standardized discriminant coefficients in- 

 dicate that total volume (TOTVOL) is the most important 

 determinant of salability. 



In terms of significant variables, the logistic regression 

 and the discriminant function are very similar. Their 

 prediction results are noticeably different, however (table 

 3b). The logistic regression correctly classifies 42.9 per- 

 cent of the unsold sales, while the discriminant function 

 correctly classifies 85.7 percent. The logistic regression 

 has the advantage of predicting sold sales, but the differ- 

 ence is not substantial. Given its unsold sales prediction 

 accuracy, the discriminant function correctly classifies a 

 higher percentage of all the sales, 72.5 percent compared 

 to 65.0 percent. Also, the holdout method indicates that 

 the discriminant error rates (percent correctly classified) 

 are quite stable, with the largest percentage change oc- 

 curring within the unsold sales class. 



WESTSroE EQUATIONS 



Table 4 presents the westside equations and classifica- 

 tion results. Both equations contain the same statistically 

 significant variables, total volume (TOTVOL), total sale 

 acres (TOTSALE), and average slope (AVGSLOPE). The 

 westside equations indicate that sale size and slope are 

 significant determinants of sold and unsold sales. On the 

 westside forests, however, the total sale acres have a 

 negative effect on salability. The average slope indicates 

 offerings found on steep slopes are more likely to be un- 

 sold. The standardized discriminant coefficients indicate 

 that total sale acres is the most important determinant of 

 salability. 



Table 4b presents the westside gate 1 classification 

 results. A higher percentage of unsold sales are correctly 

 classified by the discriminant function. But the logistic 

 regression correctly classifies a higher percentage of sold 

 sales. Overall the logisitic regression correctly classifies 

 63.3 percent of the sales, in comparison to 60.2 percent, 

 for the discriminant function. The holdout classification 

 results indicate the discriminant results are quite stable; 

 the percentage correctly classified is identical for both 

 classification measurements. 



Table 3 — Gate 1 eastside equations and dassification results 

 A. Equations 



Logistic regression 



Discriminant analysis 



Variable Coefficient (Std Err) Coefficient (Std Coeff) 



TOTVOL 

 (TOTVOL)"^ 

 TOTSALE 



Constant -.276 



0.00048 (0.00022) 

 (.498) 



0.027 (0.575) 

 .0003 (.487) 

 -1.582 



B. Classification results 



Logistic regression Discriminant analysis 



Actual 



Total 



Correct 



Percent 



Correct 



group 



sales 



predict 



correct 



predict 



Unsold 



14 



6 



42.9 



12 



Sold 



26 



20 



76.9 



17 



All sales 



40 



26 



65.0 



29 



85.7 (78.6)' 



65.4 (61.5)' 



72.5 (67.5)' 



'Indicates percent correctly classified using the holdout method. 



Table 4 — Gate 1 westside equations and classification results 



A. Equations 



Logistic regression 



Discriminant analysis 



Variable 



Coefficient (Std Err) Coefficient (Std Coeff) 



TOTVOL 



0.00008 



(0.00004) 







Ln(TOTVOL) 







0.496 



(0.742) 



(TOTSALE)"^ 



-.032 



(.008) 







Ln(TOTSALE) 







-.734 



(1.135) 



AVGSLOPE 



-.032 



(.010) 



-.053 



(.659) 



Constant 



1.984 



(.380) 



2.562 





B. Classification results 



Logistic regression Discriminant analysis 



Actual 



Total 



Correct 



Percent 



Correct 



Percent 



group 



sales 



predict 



correct 



predict 



correct 



Unsold 



145 



54 



37.2 



92 



63.4 (63.4)' 



Sold 



204 



167 



81.9 



114 



55.9 (55.9)' 



All sales 



349 



221 



63.3 



206 



59.0 (59.0)' 



'Indicates percent correctly classified using the holdout method. 



5 



