Using these time-since-last-cutting variables, a third set of screening runs was made to 

 combine all data sets for each "vigor" class. Again, blackjack pine had two data sets; one 

 with Taylor Woods data and one without. Table 25 lists the independent variables and how they 

 were grouped. The runs were set up so that the basic independent variables from the previous 

 step were forced into all equations. The screening then picked, at most, one independent 

 variable from each of the remaining groups. These groups were formed by multiplying the three 

 sigmoidal curves (A^ , A2, A3) by all the independent variables except InCS) and In(GRF). 

 For those independent variables in which a time-since-last-cutting variable is also picked, the 

 effect is to provide for a change to the regression coefficient of that independent variable 

 as time-since-last-cutting changes. 



Table 25 . --Definition of independent variables, groups and sets for combined plot, log of basal 



area growth screening runs 



Variable 

 number 



Group 

 number 



Set 

 number 



Variable ^ 



1 





1 



ln{U) 



2 







D 



3 







LBA2 or (LBA2)^ 



4 







MBA2 or (MBA2)^ 



5 







UBA2 or (UBA2)2 



6 







in(S) 



7 







Jn(GRF) 



8 



1 



1 



AiJn(D) 



9 





2 



A2ir!(D) 



10 





3 



AsJnCD) 



11 



2 



1 



AiD 



12 





2 



A2D 



13 





3 



A3D 



14 



3 



1 



A1LBA2 or Ai(LBA2)2 



15 





2 



A2LBA2 or A2(LBA2)2 



16 





3 



A3LBA2 or A3(LBA2]2 



17 



4 



1 



A1MBA2 or Ai(MBA2)2 



18 





2 



A2MBA2 or A2(MBA2)2 



19 





3 



A3MBA2 or A3(MBA2)2 



20 



5 



1 



A1UBA2 or Ai(UBA2)2 



21 





2 



A2UBA2 or A2(UBA2)2 



22 





3 



A3UBA2 or A3(UBA2)2 



23 



6 



1 



Ai 



24 





2 



A2 



25 





3 



A3 



^In those 



cases where two variables 



are given, the first 



is for blackjack pine and second 



is for yellow pine. 



The most promising equations from these runs were selected and the regression coefficients 

 determined. Analysis of these coefficients showed that the coefficients on In(S) were not 

 reasonable (that is, they were negative) for yellow pine and for blackjack pine with the 

 uneven-aged data set alone (plots 61, 62, 71, and 72), and that the coefficient on in(GRF) was 

 not reasonable for blackjack pine with the uneven- and even-aged data sets combined. These 

 runs also indicated that a high degree of multicol linearity existed. 



60 



