Association
Which of the four major validities (internal, external, construct, statistical) matters LEAST for association studies?
Internal validity - only for experiments
This confound occurs when people get to pick their own experimental condition
Selection Effect
This type of experiment design strategy can help decrease both observer bias and demand effects
Double Blind Study
Between-subjects
Null results almost never get published. This is sometimes called what?
The file drawer problem
Which type of graph is most useful if one variable is categorical and another variable is numeric
Bar Graph
This type of confound happens when participants guess the hypothesis and try to behave in a way which confirms it
Demand Effects
A researcher believes physically manipulating objects will help kids learn about balancing equations & does an experiment investigating how well kids learn after playing domino math game. What major threat is there?
Maturation Effect
This experimental design strategy helps to prevent practice & fatigue effects from skewing the results.
Counterbalancing
Weak manipulation
What thing can researchers do in a correlation study to increase statistical validity?
Get a large sample size!
This confound refers to the fact that treatments are effective if people believe they are effective
Placebo Effect
A researcher studies the effect of a "get outdoors" campaign that launched in January & finds, after three months of the campaign, people report spending more time outdoors! What's the threat?
History Threat
A researcher believes eating oatmeal in the morning increases concentration - they get a group of humans to eat oatmeal every day and measure their concentration afterwards. Is this an experiment? Why or why not?
No - there is no control condition
An experiment studies the effect of viewing cute puppy videos on subsequent generosity among a diverse group of people and gets a null effect. What's a potential reason?
Individual differences / within-group variability (creates noise)
The association between the umber of friends one has and experience of daily stress is: r = .09, 95% CI [–.27, .45]. Are these correlated? Why?
No - the confidence interval contains 0
This type of confound ONLY occurs in within-subject designs can can be prevented by counterbalancing
Order Effects
A first-aid class gives a pre-test and a post-test to evaluate emergency preparedness, but the post-test is harder. What's the threat?
Instrumentation Threat
Every participant in a study watches a lecture video in a swimming pool and then watches a separate lecture video in a traditional classroom. What kind of design is this?
Within-subject
Doing a manipulation check before a study begins can help prevent what sort of issues?
Ceiling/Floor Effects (although weak manipulation could be correct too)
What is one possible thing that can interfere with the measurement of a correlation?
The third variable problem
A group of experimental subjects does exceptionally well on a pre-test measurement for math skills and then, after training, their scores on a similar test go down. What's the threat?
Regression to the Mean
Experimental subjects must choose between a high-risk;high-reward option and a low-risk;low-reward option. What type of design is this?
Concurrent measures
An experiment tried 5 different statistical tests but only reported one of them (the significant one) and it doesn't replicate. The original authors were guilty of what suspicious practice?
p-hacking