Key Takeaways

• Often when we want to show that something helps something else, we look to see if they are correlated.  A correlation is when two different things change together (e.g. thing 'a' and thing 'b').  If 'a' and 'b' don't change together (no correlation) then neither causes the other.  If they do change together, 'a' could cause 'b', 'b' could cause 'a' or both could be caused by a third thing 'c'.  

• To know whether thing 'a' causes thing 'b' you must make a comparison.  How you do this comparison is called your experimental design. 

• We covered 3 types of designs:

(1) Compare to a similar group 

(2) Compare people to themselves before and after you do something (often called pre/post)

(3) Or use both strategies together (called a mixed design).

• The best way to know that thing 'a' causes thing 'b' is to randomly assign people to your comparison group.  Random assignment is when you choose who goes in what group a random.  It works well because it makes sure the two groups you compare are the same in everyway except for the thing you are testing.  Random assignment is not always ethical, but if you can do it ethically, you should. 

• If you can't do random assignment you can still increase your confidence that what you are doing works by choosing similar comparison groups,  comparing to more than one group, measuring more than once and measuring in different contexts.