Day 13
Study Design: Causal Claims



EPSY 5261 : Introductory Statistical Methods

Learning Goals

At the end of this lesson, you should be able to …

  • Describe the benefits of randomly assigning participants to treatment groups.

Drawing Causal Conclusions from a Study

When can we make a causal claim?

  • In a lot of research we are interested in making a causal claim.
    • Example: This curriculum improved learning!
  • To be able to make this claim we need to be confident there aren’t any other factors that actually influenced the outcome!

When can we make a causal claim?

To mitigate uncertainty about external factors we want to compare equivalent groups. {.v-center-container}

Ideally…

We have two identical groups (think identical twins). One group is given the treatment and the other isn’t (the control group—often given a placebo). That is the only difference between them.

If there is a difference between the groups in the end, we can be confident it was from the treatment!

Identical Groups?

  • We can’t create identical groups … necessarily.
  • Instead, we create probabilistically equivalent groups.
    • a.k.a. Mathematically equivalent groups
  • These groups are, on average, the SAME on all characteristics (observable and not observable).
  • To do this we randomly assign people to be in the treatment group and the control/placebo group.
  • You will explore this in an activity!

Social Media Activity

Random Assignment

  • The process of random assignment creates, on average, groups that are “identical”.
  • It doesn’t do this perfectly every time we create groups, but on average it does—and that is good enough!
  • Creating groups through random assignment is the gold standard for making causal claims.
  • There are more advanced methods, but we won’t get to those in this course.

Can you make a causal claim?

Summary

  • Randomly assigning participants to groups allows us to create groups that are probabilistically equivalent.
    • The same on average
  • This allows us to infer that our treatment was the only cause of a difference between the groups (if a difference is detected).
    • Then we can make a causal claim