Threats to validity can happen anytime there is an independent variable (treatment) that is implemented to a behavior. Meaning anytime you work with a student there is some protocol put in place on how/when/what you should do when that target behavior occurs. Therefore threats could be lurking around every behavior plan waiting to strike.
The best course of action is to know these threats and be prepared to have safe-guards put in place. The nine threats are:
History: when other conditions are changed the same time the independent variable is implemented.
- This means that there is some other skill acquistion going on besides your treatment. An example would be a college student learning how to cook their own meals. If you were investigating weight gain during college this skill could influence that participants weight. Less take-out = decrease in weight.
Disclaimer: No one should be this happy to be making a salad.
Maturation: when the subject has some form natural occurring developmental process and not the implementation of the independent variable.
- By the age of 2 children should be able to walk, this is a milestone that could affect procedures put in place for mobility. We went from crawling, to walking, to running. Say hello to the new Road Runner
Experimental Mortality: when a subject leaves the experiment before it is concluded.
- Simply put: your client/student quits services. Worst break up ever!
Regression to the Mean: when the subjects who have extreme initial scores tend to have future scoring closer to the mean.
- This is common for me when crafting. My first project like a scarf will come out as a hot mess of yarn then the next time I try I will remember all of the mistakes I made it will come out as a decent scarf. I like to call this the Pinterest Fail Formula:
Repeated Testing: participants tend to remember their answers
- If you give the same test participants could remember what answers they got right. Happened in grad school pre and post final exams (remember 50 definitions- challenge accepted!)
Instrumentation: the observers becoming better or worse at observing the target behavior. The way the pre and post tests are delivered.
- This can happen when there is a revision to the operational definition of the behavior. Giving more information on most (if not all) measures of the behavior will increase accurate data collection.
Selection: when conducting a study with multiple groups some groups can become unequal with subject related variables
- When working with a classroom of kids and you have teams of boys vs. girls in a game of Jeopardy. This could be unequal groups based on test scores (the boys team could have one student who has an A in the class while the girls team has 3).
- It rocks when you get the bigger piece of birthday cake but it is not equal
We all who would win this fight! RIP big guy
Reactivity: the change in behavior due to testing
- We all act one way at work and another with friends. That is because swearing behavior is punished at work while with friends it is encouraged.
Experimental Bias: when the researcher already has an idea of what the results should be.
- Self recording: when you write down all of the food you have consumed in a day you hope that it will decrease your over eating. If you do not focus on healthier food options or decreasing portions you will not see a decrease. Put the work in!!
Informative? Yes.
Funny? I can hope.
