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?
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?
Which do you want, a random sample or matched pairs?
http://stattrek.com/AP-Statistics-2/Experimental-Design.aspx?Tutorial=AP
To randomly divide the 10 students into Group A and 10 students into Group B in a cafeteria full of students, the least biased way to do it would be to follow these steps:
1. Identify the total number of students in the cafeteria. Let's assume that there are more than 20 students present, so you have enough to randomly divide between the two groups.
2. Assign a number to each student, starting with 1 and ending with the total number of students. For example, if there are 30 students, assign numbers 1 to 30 to each student.
3. Use a random number generator, either electronically or manually, to generate a random number between 1 and the total number of students. This will be used to select the first student for Group A.
4. Once the first student is selected for Group A, remove that student's number from the pool of available numbers.
5. Repeat steps 3 and 4 until 10 students have been selected for Group A. Each time a student is selected, remove their number from the pool of available numbers.
6. Once Group A is complete, all remaining students will be assigned to Group B.
By following these steps, you will ensure a fair and unbiased random division of the 10 students into Group A and 10 students into Group B, without any personal bias or influence on the selection process.