If you were in a cafeteria full of students and you wanted to randomly divide 10 students into group A and 10 students into group B, what would be the least biased way to do it?

Answer 1

Specifically for a two sample t test on difference in means with a matched-pairs design. What would the step-by-step process be for a simple random sample from the cafeteria population?

Answer 2

Which do you want, a random sample or matched pairs?

http://stattrek.com/AP-Statistics-2/Experimental-Design.aspx?Tutorial=AP