16. Correlation (INCOMPLETE) Flashcards
r
“correlation coefficient”
describes relationship btwn 2 NUMERICAL variables
from -1 to 1, larger absolute value means stronger correlation
+ = as one increases, other increses
- = as one increases, other decreases
r^2
coefficient of determination
describes PROPORTION of variation in one variable that can be predicted from the other variable
mathematical def of r
Covariance(X,Y)/ sqrt(var(x)var(y))
Covariance
how much X/Y vary together
+ relationship contribute positive numbers to the sum
- relationship contribute - numbers of a sum
sum of cross products
numerator of r equation
sum of (Xi - Xbar) * (Y1 - Y bar)
sum of squares for x
sum of (Xi - Xbar)^2
Stating conclusion of hypothesis test for correlation
t = __ is greater than t0.05(#), df = ___ so we can reject the null hypothesis and say that _______ (P < ______>
***** when stating conclusion, use 0.05, but state P value at end as accurately as possible
assumptions of correlation
LOTS
normal dist for x and y, all for each same variance
Alt to correlation that does not make so many assumptions
Spearman’s rank correlation
Based on ranking x and y lowest to highers, smallest x should correlate with smallest y OR low with high
assumptions of spearman’s rank correlation
random sample
what does r estimate
population parameter p (rho)
rho = 0
no relationship between variables
coefficient of variance vs coefficient of determination
co var = r = describes relationship between 2 numerical
do det = r^2 = proportion of variation in ONE population PREDICTED from the other variable (always positive)
is r^2 very important
YESYESYESYES
covariance
how much variances change together
