Details for the R programs used in the article by Tsai and Gilmour (2007)

Three-level designs in 18 runs
Save  level3.zip for the program files.
Save r18dat.zip for the three-level non-isomorphic designs in 18 runs with 3 to 7 quantitative factors that Tsai et al. generated by their columnwise procedure
unzip the two files to your computer.


Under R console open the source code design18.r
> source("design18.r")  # specified noRun=18
noF= #number of factors 3~7

noF=4
Read 13 items
Read 129 records
study design=  # which of the 129 four-factor designs in 18 run to be studied.

study design=3

Output for the design plan and their GWC
-----------------------------------------------------------------
There are  129  main effects designs for 18-run designs with  4 factors.

The design of interest is Design 3
1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3
1 1 2 2 3 3 1 1 2 2 3 3 1 1 2 2 3 3
1 2 1 3 2 3 2 3 1 3 1 2 1 3 2 2 1 3
1 2 3 1 2 3 3 3 2 2 1 1 2 1 1 3 3 2
Generalised word count
(#noF, order , Lin,  Quad :  b_k(i,j))
1     2     0     1     0
1     1     1     0     0
2     4     0     2     0
2     3     1     1     0
2     2     2     0     0
3     6     0     3     0.875
3     5     1     2     0.75
3     4     2     1     1.125
3     3     3     0     0.0833
4     8     0     4     0
4     7     1     3     0
4     6     2     2     0
4     5     3     1     0.6667
4     4     4     0     0
 GMA-aberration
A 1 : 0
A 2 : 0
A 3 : 2.8333
A 4 : 0.6667
 beta-aberration
B 1 : 0
B 2 : 0
B 3 : 0.0833
B 4 : 1.125
B 5 : 1.4167
B 6 : 0.875
B 7 : 0
B 8 : 0
Q second-order :  0.3646
-----------------------------------------------------

WorList for the generalised word count in Tsai and Gilmour (2007)
 
word.A for the word length pattern given by the ANOVA-type generalized minimum aberration criterion defined by Xu and Wu (2002)

word.B for the
word length pattern given by the ß-aberration criterion defined by Cheng and Ye (2004)

Q.2nd is the value of Q for the second-order maximal model given by Tsai et al (2000)


Other useful information can be obtained from these programs

For any three-level design

for a given design plan
Design=
 [1,]    1    1    1    1
 [2,]    1    1    2    2
 [3,]    1    2    1    3
 [4,]    1    2    3    1
 [5,]    1    3    2    2
 [6,]    1    3    3    3
 [7,]    2    1    2    3
 [8,]    2    1    3    3
 [9,]    2    2    1    2
[10,]    2    2    3    2
[11,]    2    3    1    1
[12,]    2    3    2    1
[13,]    3    1    1    2
[14,]    3    1    3    1
[15,]    3    2    2    1
[16,]    3    2    2    3
[17,]    3    3    1    3
[18,]    3    3    3    2


Use the program gwc.r
worList is the generalized word count for the design


delta.mod(a, b, c, noF)
gives the number of eligible models for
the second-order maximal models for
designs with noF factors
in which a factors' linear effects, b of the a factors' quadratic effects and c
linear×linear interactions of the a factors are included.

For example delta.mod(3,0,1,3)=32, 
delta.mod(3,0,2,3)=16 and delta.mod(3,1,1,3)=16 as for the calculation of the Q criterion for three-factor designs in 18 runs.

delta.mod(3,0,1,4)=544,  delta.mod(3,0,2,4)=272 and delta.mod(3,1,1,4)=272 as for the calculation of the Q criterion for four-factor designs in 18 runs.

delta.mod(3,0,15)=17334, 
delta.mod(3,0,2,5)=8626 and delta.mod(3,1,1,5)=8628 as for the calculation of the Q criterion for five-factor designs in 18 runs.

delta.mod(3,0,1,6)=667068,  delta.mod(3,0,2,6)=287228 and delta.mod(3,1,1,6)=287228 as for the calculation of the Q criterion for six-factor designs in 18 runs.

delta.mod(3,0,1,7)= 10758936,  delta.mod(3,0,2,7)= 10758936 and delta.mod(3,1,1,7)= 10758936 as for the calculation of the Q criterion for seven-factor designs in 18 runs.