Exam 2 Lecture 5 Flashcards
The problem with people
- Participants can lie, deceive, make mistakes, misinterpret, and plainly not realize who they really are
- Researchers can also be the source of problems, we are human, but we have devised strategies to minimize the ways we introduce bias in our study protocols
- These strategies increase the chances that we will get meaningful and accurate results, but not everyone uses these strategies
- Data literacy requires that you know when such strategies should be used and call would when they are missing
Blinding- Overview
Participants can influence the outcome of a study if they know what is expected of them
- You got an antidepressant, so your depression symptoms should go down.
Researchers can influence the outcome of a study by accidentally or purposefully treating groups differently- things that ‘hint’ at what should happen
- The researcher says to participants in the ‘active’ exercise condition: “How are you feeling today? You ready to give it your all today! Let’s see how much we can do. I’m here for you.”
- The researcher says to participants in the ‘comparison’ exercise condition: “OK, let’s see what we can do. Take it slow and steady.”
This sets up the study to get the results the researchers want (although may be unintentional)
Single blind (in blinding)
Participant doesn’t know which group they are in. This gets rid of ‘expectancies’ (like a self-fulfilling prophecy)
Double blind (in blinding)
Neither participant nor researcher knows the group assignment. This gets rid of researcher’s treating one group different than the other.
- How can you be double blind? Third party!
- Someone in the lab that does not interact with the participant knows and prepares experiment
Ex:
- Study includes an active medication and a placebo group.
- Only ‘3rd party’ knows whether the pill a participant receives is real or placebo. The researcher handing the participant the pill and performing the experiment does not know.
Is double-blind possible in an… alcohol experiment?
exercise experiment?
Could be. Depends on what you are measuring/how you are measuring.
Alcohol experiment- could have different alcohol contents in same drink to test effect of alcohol vs. placebo
Leading the witness (What should the researcher be doing to follow protocol?)
Checks on wording of questions and response options to avoid incidentally pointing the participant to the ‘right’ answer
- You don’t like this class, do you? You like this class, don’t you?
- Do you exercise? When was the last time you exercised?
- Even being enthusiastic about some things but not others (I love protein shakes too! Yuck, red meat!!)
Checking experimental protocols to make sure you aren’t increasing the likelihood that you’ll see what you want (or giving them ideas about what you want to see)
- Measuring endurance in 1 group before lunch and 1 group after lunch
- Telling participants in 1 group but not the other how long the experiment will last
- Telling participants in 1 group “this should make you feel stronger”
Intentionally leading the witness
Experiment- Participants five saliva samples on some nights (researcher tell them when). Participant gets paid $20 for every sample. The study flyers say the study is about enzymes in saliva after eating meat.
Actually looking at alcohol metabolizing enzymes and only asking for saliva on nights they drink.
Masking: Each day, participants get a survey-
- Did you eat meat today?
- Did you exercise today?
- Did you drink today?
Is deception ethical?
If I want to study the effects of cannabis on study habits, should I tell participants?
Pros?
Cons?
If I want to study math skills under stress, could I make participants take a math test and tell them all their answers were wrong even if they were right?
Pros?
Cons?
Just saying.
Social desirability- ‘demand’ characteristics
- Just knowing that you’re being studied can make a person change their behavior (nutrition study-> I won’t eat candy this week!)
- Sometimes you just know the right answer (no, I don’t ever cheat!)
- Sometimes you want to help the researcher (I DO feel better)
Protocols need to think about who is asking, where, and when
- Is information private, data topic illegal?
- Could the data have emotional value (weight, BMI)?
- Could the data create competition/insecurity (1-RM)?
- Will person report same behavior to a peer versus professor?
What are the types of biases?
- Recall bias
- Observer bias
- Experimental order bias
- Selection bias
Recall bias
It’s been years since I used drugs!
Observer bias
They don’t look like a drug user
Experimental order bias
- Treadmill test before vs after 3 scoops of ice cream
- Asking about anxiety symptoms before vs after a cognitive test
Selection bias
- Cherry picking ‘best’ candidates (sampling bias)- antidepressant studies
- Marketing to specific pool (Self-selection bias)- ad with pot leaves
- Experiment is harder for some people so they drop out (attrition bias)- most research happens during work hours
