Pre-experiments
Due to the weaknesses of pre-experimental designs, researchers must heavily rely on what two elements to support their conclusions?
Theories and literature reviews
Non-experimental design relies on what kind of method to make inferences?
Statistical control (ex: multiple regressions)
What actions should we take if there's a history threat to internal validity?
Add a non-equivalent comparison group or remove the treatment to help determine if external factors are causing change
What is name of Dr. Elliot's program focused on building friendships between people living with dementia and volunteers as they engage in art-making activities?
Opening Minds Through Art
What pre-experimental design is this?
X O
One-shot case study
When a control variable is statistically significant, what relationship does it have with the dependent variable?
Correlation (not causation)
What actions should we take if there's maturation or regression to the mean threats to internal validity?
Add measurements over the time period to help determine what effects are from the treatment and what are natural changes
What pre-experimental design is this?
O X O
Single group pretest-posttest
Why would a non-experimental design not be the most appropriate choice here? Participants are randomly assigned to either receive a new drug or a placebo. Researchers then compare outcomes between the two groups.
There's opportunities to use more rigorous designs (comparison groups/random assignment/just provide a reason tbh)
What actions should we take if there's instrumentation threats to internal validity?
Remove measurements over the time period to prevent opportunities for these threats to occur
What pre-experimental design is this?
X O
O
Static group comparision
A school wants to evaluate whether a new teaching method improves students' math scores. Students take a pre-test at the beginning of the semester and a post-test at the end.
There's opportunities to use more rigorous designs (really just provide any reason pleaseeeeee)
What actions should we take if there's a selection bias threat to internal validity?
Remove the non-equivalent comparison group
All other things being equal, as researchers would we prefer to use a single group pretest-posttest or a one-shot case study?
The single group pretest-posttest case study is slightly better than the one-shot case study, as it has two time points as compared to one, allowing an evaluator to know the changes before and after the program intervention
When would a non-experimental design be the best design choice?
Situations where it would not unethical or impractical to manipulate variables (ex: assigning people to experience trauma to study its effects)
What actions should we take if there is no pretest available?
Use proxies in place of the variable you are actually trying to measure