Coggle requires JavaScript to display documents.
Take a theory.
Deduce its observational predictions.
Test one prediction.
Refutation? --> Reject the theory // No refutation? --> Go back to 3.
// Chance of error // p value
E.g. a survey or sample description
Cannot make inferences (e.g. statistical inferences/analysis)
Can establish basic prevalence
Cannot make predictions
Cannot establish causality
e.g. Cross-sectional or descriptive
Regression or Correlation
Looks at links between phenomena
Can consider confounding variables
Can make some predictions
E.g. Some case control designs, open trial designs
Can look at change over time
Can establish causality in some cases
(e.g. with SEM)
Does not directly isolate the mechanism of change
E.g. a randomised trial RCT
Designed to establish causality
Several levels of integrity and equipoise (e.g. blinding)
Very expensive and complex to conduct