| In pharmaceutical research, several tests are performed on a new drug product. One of these tests is a stability study that estimates the shelf life of the drug product, i.e. the amount of time it keeps stable. Normally, stability studies are conducted with a complete factorial design, where the design factors are batch, strength and package type. However, it is also possible to do this with a fractional factorial design, which will save lots of time and money. It is very important, though, to make sure not to lose too much precision in the estimates of interest. In this thesis, several different fractional factorial designs are compared to the complete factorial design in theoretical investigations as well as a case study. These analyses show that some fractional designs can be recommended, whereas others should not be used. |