Curtis et al. (1982) proposed to conduct an experiment to study the eect of four factors in the software development process upon the software life cycle cost and the number of residual errors. The factors and levels are shown in the table below:

(a) A 24−1 fractional factorial design was proposed to study these factors since the cost of building and testing a multi-module system can be quite high. The author proposed to use a design with generator C = AB.
Suggest a better design generator and explain why you think it would be better.

(b) If the number of levels for factor A=Design methodologies and B=Programing languages were three rather than two, what is the mini-
mum number of runs required for an orthogonal array design for studying two factors at three levels and two factors at two levels (i.e., 32 × 22)?

(c) Use the FrF2 and AlgDesign packages to create a near orthogonal 12-run fractional factorial design for a 32 × 22.

Transcribed Image Text:

Label Factor A Design methodology B Programming language C Source language debugger D Testing strategy Levels Functionally structured Data structured Ada No Functional CMS-2 Yes Structural