11: Statistical Methods Flashcards
Define p-values.
Probability of obtaining data this extreme or more extreme, assuming the null hypothesis is true.
Industry standard to call something “statistically significant” is what?
p < .05.
r = .30, p = 0.03. What does this mean?
Assuming null hypothesis is true (r = .00), 3% chance you would find a correlation of .30 or larger in the long run, over repeated samples.
What are four criticisms of statistical significance?
p < .05 is an arbitrary cutoff.
With large enough sample size, virtually any finding could be statistically significant (as N increases, so does power).
Only make statements in relation to null hypothesis; lacks intuitive ease of understanding.
Can’t endorse the null, can only fail to reject it.
What is the definition of a 95% confidence interval?
Conduct 1000s of similar experiments and calculate a confidence interval for each one, 95% of those CIs would contain the population parameter.
What are the three types of studies in equivalence testing?
Traditional comparative; equivalence; noninferiority.
What is the equivalence margin?
Range of effect sizes which might be non-zero, but are small enough to be considered practically and clinically insignificant.
One straightforward way to assess clinical significance is to set the null hypothesis at something higher than 0, and use what?
Confidence intervals.
The Jacobson & Truax Approach to clinical significance offers three possibilities. Describe each.
Post-intervention score 2 SD away from mean of the dysfunctional population.
Post-intervention score 2 SD within the mean of the functional population.
Post-intervention score closer to mean of the functional population than mean of the dysfunctional population.
List the four categories of clinically significant change.
Recovered: “clinically significant change.”
Improved: “statistically significant change” but not enough to be considered full recovery.
Unchanged: no statistically significant change.
Deteriorated: statistically significant worsening of symptoms over time.
What is the Number Needed to Treat (NNT)?
Number of patients that need to be treated for one to benefit compared with a control in a clinical trial.
A slope of 2 means what?
Every 1-unit change in X yields a 2-unit change in Y.
In mediation, the c-path (the total effect) should get _____ with the addition of a mediator.
Smaller.
With X, Y, Z, in mediation, what do you do with the a-path (X to Y) and the b-path (Y to Z)?
a-path: get slope of relationship.
b-path: get slope while controlling for X.
Mathematicians have shown that (a-path * b-path) = c-path – c’ path. What does this mean?
If a * b (“the indirect effect”) is statistically significant, mediation has occurred.
