Confounding and Obscuring Variables (Week 11) Flashcards
What are some potential threats to internal validity?
Maturation effect: people are initially nervous in new situations, but they adapt over time
History effect: people may have prior history with situations that will change the way they are effected
What is regression to the mean?
Unusual results measured at time 1 are likely to regress (go back) closer to the average at time 2
ex: you bowl a strike the first time bowling - your first shot was so unlikely, it would be even more unlikely for your second shot to also be a strike
What is Attrition threat? How do you avoid it?
Provide an example
When participants quit a study mid-way
Include a comparison group - compare participants who quit the study on baseline (pre-test) variables
ex: effectiveness of exercise regimen vs. sedentary
- if 25% of people leave halfway, compare that 25% with the remaining 75% on their pre-test weight, BMI, life satisfaction ect..
What is an instrumentation threat? How can this be avoided?
When a measuring instrument changes over time
ex: observers coding co-workers passive-aggressive comments may become more lenient or more strict throughout the study
How to avoid: use a comparison group, develop a solid code book
What are placebo effects? What is a double-blind placebo control?
Participants improve because they think they are getting the active treatment
People are randomly assigned to conditions, experimenter who rates participants does not know who has had which condition
What is a null effect?
What is the file-drawer problem?
When the independent variable makes no significant difference on the DV
Null effects are often not published
Why might a study produce a null effect? (1-3)
Provide examples when relevant
Hint: Real. Internal. Weakness
- The IV really does nothing to the DV
- Internal validity issue, there was not enough between-groups difference
ex: cats are lethargic and sometimes don’t walk faster than turtles so it might not produce much of a difference in participant walking speeds - Weak manipulation, the manipulation was too short or not powerful enough
ex: maybe 15m of turtle watching isn’t enough - maybe it takes several hours
Why might a study produce a null effect? (4-6)
Provide examples when relevant
Hint: Insensitive, Floor to Ceiling!
- Insensitive measures, the dependent variable was not measured sensitively enough
ex: Fitbits may not be sensitive enough to pick up on difference in speeds - Floor effect, almost all participants score low on the dependent variable
ex: puzzles are way too difficult - effect is drowned out by difficulty - Ceiling effect, almost all participants score high on the dependent variable
ex: the puzzles are way too easy - even if there is an effect it is drowned out by how easy it is
Why are statistics so powerful?
More data points = more reliable results
Estimate of the mean becomes more and more precise the larger the samples - estimate of the difference between the groups also becomes more precise
Why run a t-test?
Run test to see whether the data is consistent with the null hypothesis - When n (number of participants) is larger, the larger t can get, which means more chance of rejecting the null
What are the two effects that can produce null effects which you really need to watch 4?
Ceiling and floor effects
