Lecture 6 - correlation and regression

Question 1

what is correlation used to test?

Answer 1

the relationship between two variables

Question 2

what is linear regression used for?

Answer 2

used to predict one variable from another

more powerful than correlation

Question 3

can linear regression infer causation?

Answer 3

no

Question 4

what is pearson correlation coefficient used for?

Answer 4

measuring the extent of the relationship between 2 variables

Question 5

what are the two conditions for the use of peasons?

Answer 5

both variables must be continuous and follow a normal distribution

Question 6

what if the two conditions of pearsons do not apply?

Answer 6

spearman rank is used instead

Question 7

how does traditional regression work?

Answer 7

regress the dependent variable ( outcome ) on the independent variable

Question 8

what do the variables in the regression line mean?

Answer 8

y = ax + b
y= dependent variable
x=independent variable
b= y- intercept
a=gradient of the line

Question 9

what is the least squares approach

Answer 9

estimates should be made to minimise the sum of squared deviations –> residuals

Question 10

what are the assumptions behind the linear regression model

Answer 10

1. errors must be normally distributed
2. linear association between x and y
3. Variance of y should be equal across all values of x