6.3 Subjects and Groups:  The Importance of Randomness

---

Suppose that a beverage company wishes to test out an experimental energy drink called “AwesomeSauce.”  The company’s scientists claim that this drink improves mental alertness, physical stamina, and explosive strength – all by delivering a completely natural mix of vitamins and minerals, thereby preventing the sugar-and-caffeine crash that consumers might expect after quaffing other energy drinks.

To test AwesomeSauce’s performance, the beverage company wants to use college students – the company believes that a college campus would be ideal for recruiting new consumers of the product (and for finding a willing pool of research volunteers!) 

The company plans to use 50 students in an experiment.  25 of the students will receive a 16 ounce can of AwesomeSauce (treatment group), while the other 25 will receive a placebo (control group).  The placebo drink will look and taste like AwesomeSauce, but will contain none of AwesomeSauce’s special ingredients.  Each group will be asked to consume the beverage given to them, and then to perform a series of cognitive and physical tasks.  

After gathering the students who agree to participate, AwesomeSauce needs to be sure that students are randomly assigned to each group. 

If, instead, AwesomeSauce allowed self-selection by students into either the control group or treatment group, what sort of biases might creep into the results?   Perhaps the most adventurous students, who would be most excited about trying some kind of unproven but promising energy drink, would volunteer to try the new drink, rather than receive a placebo.  Furthermore, what if that group of students is both brave and suggestible?  What if they believe that such a drink will really work, and that belief gives them extra confidence in the physical and cognitive tasks? 

Random assignment of students into the two distinct groups will help to correct for some of the problems with bias that would occur otherwise.