Unit 12: Sampling. When enough is enough.

47 Considerations When Sampling

Considerations When Sampling

Self Selection Bias (in social science)

“When the peeps who were ok with being in your study might be different from the peeps who didn’t – and in important ways.”

Let’s say that you want to learn about how UI students feel about the Cambus routes and timeliness. You put out a survey blast – “Hey y’all! We want to know how YOU feel about Cambus!” Think about the types of students who are gonna click that link and complete the survey. Might they be different from the peeps who DON’T click? Perhaps you surveyed anyone who was walking in/out of the library, or people who were waiting at the Cambus stop? What if you did a simple random sample of all students at UI? How might your results vary? What are other considerations? Does it matter if people live on campus or off campus? If they actually…you know…RIDE the Cambus? These are all important considerations.

Learning Objectives

Warning: Things to be aware of!


Sampling is a multi-faceted process and requires thoughtful execution – particularly in experiments!

When sampling, one of the most important considerations is that your sample pools are equivalent! This means that the groups you use for all intents and purposes must be interchangeable and equal in how they are determined. Often the best way to achieve this is to assign groups using the systematic randomization process, this way all participants are assigned groups randomly and through this, the groups are ensured to be as equal as possible.

This is important because you want all variations that you observe in your groups to be caused by the manipulation of variables, not because they varied to begin with. Without this, it is possible that there are changes in your groups that are not a result of the experimental process.