1 - regression, correlation + hypothesis testing

when you have an exponential graph/equation, what can u do to make it simpler

take logs and make it linear

write as a linear

.

make it linear

.

how do you know whether its better to use an exponential model or linear model for the data

whichever one the regression line is more perfect for when u enter the data into ur cg50

how does hypothesis testing come into this topic

you test if the r (PMCC) of a sample indicates a linear relationship of a whole population (p)

hypotheses for both 1 and 2 tailed tests for zero correlation

1-TAILED

positive:
H₀: p=0
H₁: p>0

negative:
H₀: p=0
H₁: p<0

2-TAILED

any correlation:
H₀: p=0
H₁: p≠0

u basically always assume no correlation until u prove there is one

how to find the critical value of the test

use critical value table on p37 of formula booklet

match sample size to percentage level given in question to find critical value

TWO TAILED = halve percentage level BEFORE looking at table

how to know whether to reject or accept H₀

given r > critical value —> sufficient evidence to reject H₀ and conclude that p>0 (positive relationship in the population)

given r < - critical value —> sufficient evidence to reject H₀ and conclude that p<0 (negative relationship in the population)

---

but if critical value > r > - critical value then there is insufficient evidence