D-3: Identify threats to internal validity (e.g., history, maturation) ©
Want this as a downloadable PDF? Click here!
Want a self-paced video course that covers all the test content and more? Click here!
Target Terms: internal validity, history, maturation
Threats to internal validity
Definition: “Threats” in this sense means something that could get in the way of our understanding. We need to be confident that any change is due to our intervention. Otherwise our “data informed decisions” may not mean what we think they mean! Remember that internal validity basically means the extent to which we can be confident that the change was due to the intervention. Lots of things could create a change in our dependent variable (usually client behavior) other than our intervention. These are called “confounding variables.” (They are also sometimes called “lurking variables” because we may not notice them).
Examples:
- Low treatment integrity: If we aren’t sure how well we implemented the intervention, or if we know we did it “kinda sorta,” then there is really no way to know if the behavioral outcomes we are seeing are because of our intervention! For example, if Rosie’s behavior gets drastically safer after implementing a token economy, but the tokens were only available half the time they were supposed to be, then we don’t know what helped Rosie be safer.
- History: variables that occur outside of the study/treatment that could influence the results – for example, a child in a manding study/treatment receives intensive speech therapy outside of the study/streatment.
- Maturation: This is when a subject matures over the course of the study, and acquires developmental skills that are not the result of the intervention. For example, if you feed a baby applesauce every day for a year, you will notice that they talk a lot more at the end than at the beginning. However, the applesauce did not teach them language!
- Experimenter bias: It’s easy for us humans to be influenced by our perceptions of the context or individuals involved. The person implementing the intervention or taking data may be influenced by their bias in favor of a particular action – such as scoring a permanent product closer to mastery if they know that it was produced by a client who is seen as “more capable.”
Note: The above is a partial list. Confounds could be unlimited. Critical thinking is essential.
Why it matters: Threats to internal validity are important for us to think about and plan for, because otherwise, we may not end up making good decisions. As behavior analysts using within-subject designs in our clinical work, internal validity is the kind of validity that we generally need to be more focused on. It all comes back to understanding the problem, so that we can take action in meaningful ways.