3 - Correlation & Regression Flashcards
List Pearson’s rule of thumb for effect sizes for r.
Small: .1 < r < .3 or -.1 > r > -.3
Medium: .3 < r .5 or -.3 > r >.5
Large: .5 < r < .7 or -.5 > r > -.7
What is the difference between Variability vs. Covariability
Variability: how much a given variable varies from
observation to observation
Covariability: how much two variables vary together
How is Sum of Squares calculated (variability)? How is Sum of Products calculated (covariability)?
SSx = Sum (Observed score x - Mean scores x) squared
SP = Sum (Observed score x - Mean scores x) (Observed score y - Mean scores y).
How is Pearson’s r calculated?
r = SP / SqrRoot (SSxSSy)
i.e. Covariability of X and Y / Variability of X and Y Separately.
What is regression toward the mean?
Extreme scores usually followed by a value closer to the average.
Extreme scores after due to chance, therefore, following score tends to be followed by less extreme score…
For correlation, what are the degrees of freedom (df)?
DF = n - 2.
How does Spearman’s correlation coefficient use rank order scores?
Deals with outliers & non-linear data
By converting ranks, i.e. 3, 2, 1… All values close to each other.
What is Cronbach’s Alpha?
- A measure of reliability.
- Reqs at least 3 items.
- Ranges 0 to 1… where 1 = 100% reliable.
Influenced by number of items; increasing numbers can produce high reliability.
Rule of Thumb… 0.6 - 0.7 = questionable.
What is the regression formula & what are the error assumptions?
Y = a + bX + e
a is intercept.
b is slope
e is error, residuals; independent, normally distributed, & homoscedastic (equal error variance).
What is the least squares parameter estimates?
Regression prediction that minimises the error.
What is a standardised regression coefficient?
When X and Y scores are transformed into z-scores, aka Beta.
B/w -1 - 1.
