SETTLING DOMESTICS DISPUTES WITH DESIGNED EXPERIMENTS AND JMP~

Twentieth Annual SAS Users Group International Conference

Lucia A. Ward-Alexander ; Blue Cross Blue Shield of Maryland ; Owings Mills ; MD

PID  :  88347

来源: CEUR

【 摘 要 】

One day Mel and Lucia had an argument overthe quickest route to drive between their homeand a nearby shopping center. Mel suggestedthat statistics could be used to settle theargument. However, Lucia felt the issue wastoo complicated to be resolved using statisticsbecause of the numerous factors affectingtravel time. Mel countered that argument bypointing out that a statistically designedexperiment could be used to settle the issueconclusively.This paper will show how SAS software (inparticular JMP) assisted the couple in answeringthe key questions posed by the argument. Animportant lesson learned from this exercisewas: statistical thinking not only works well inscience, business, medicine, and industry, butit can also improve domestic relationships.AUDIENCEThe tutorial assumes some knowledge of basicexperimental design concepts, e.g., factors;levels; blocking; randomization; replication; anda basic familiarity with JMP software, it'smenus and subcommands.THE SITUATION AND OVERVIEWThe situation began when our son got sick andneeded medication from a drug store in anearby shopping center. There were actuallythree routes that could be used to get to theshopping center. The argument ensued overwhich route was quickest. To resolve thesituation we asked two questions: Whose routewas fastest? If one route was faster, whichsets of factors (four main effects andinteractions) affects the variability in traveltimes] JMP software, with its easy userinterface and statistical graphical displays,provided the suitable platform for analyzing thedata to answer the above questions. Followinga discussion of the situation and design295considerations is an outline of the steps used inconstructing the full factorial design data tablein JMP, detailed analysis using JMP, andconclusions.DESIGN CONSIDERATIONSWe identified four factors. The first factor wasthe route with three levels. Rg~re 1 shows thethree routes: Route 1 (Mel's route), Route 2(Lucia's route), and Route 3 (Control route).A third route level was added as a control routebecause it was most commonly traveled byresidents and nonresidents of the communitywhere we lived. This control route allowed usto compare the travel times of both of ourroutes with a standard. The second factor wasdriver set at two levels. Lucia and Mel havedifferent driving styles. The third factor wascar. Lucia drove a subcompact while Me) drovea station wagon. The fourth factor was voltime,the amount of traffic volume during differentperiods of time. During rush hours (weekdays between 7:00 AM9:00 AM and 4:00PM6:00 PM) traffic volume was heavier thanduring nonrush hour times. Therefore wewanted to take this factor into account.