Correlation and Regression

Question 1

Correlation measures (2)

Answer 1

1) direction (positive/negative)
2) linear association between 2 variables

Question 2

range of correlation coefficient

Answer 2

-1 to + 1

Question 3

measures co-movement between 2 variables

Answer 3

function of covariance

Question 4

whats the difference between covariance and correlation

Answer 4

summation of movement between 2 variables within a range of - to + infinity or - to +1

Question 5

What doesn’t a scatter plot show?

Answer 5

The slope of the correlation

Question 6

What is spurious correlation?

Answer 6

correlation that is random chance; correlation is not causation

Question 7

linear relationship between dependent and independent variable

Answer 7

linear regression

Question 8

dependent variable

Answer 8

y

Question 9

independent variable

Answer 9

x

Question 10

assumptions of classic normal linear regression model

Answer 10

1. variables are linear
2. independent variable, x is not random
3. error is 0
4. variance is the same for all observations

Question 11

2 measures of regression quality

Answer 11

1. standard error of estimate
2. coefficient of determination

Question 12

how is standard error of estimate calculated?

Answer 12

root sum of differences between y and x components of the regression

Question 13

why is the coefficient of determination preferred?

Answer 13

SEE shows the change but CoD reveals how much of the variability of Y is because of x

Question 14

3 things to know for hypothesis testing with regression/correlation

Answer 14

1. estimated parameter value
2. hypothesized parameter value
3. CI around the estimated parameter value at a given degree of confidence (probability)

Question 15

What is analysis of variance?

Answer 15

measurement of variances across means

Question 16

how is multi-regression different from regression

Answer 16

effect multiple independent variables (x) have on a a dependent variable

Question 17

assumptions of multi-regression

Answer 17

1. variable relationships are linear
2. independent variables are not random
3. error of independent variable is 0, error terms are not correlated and are part of a normal distribution
4. variance for all observations is the same

Question 18

multi-regression hypothesis differences

Answer 18

1. null - all coefficients are 0
2. alternative hypothesis - at least 1 hypothesis is not 0

Question 19

difference between t-test and f-test

Answer 19

f-test - overall significance of the model
t-test - significance of each coefficient within the model

Question 20

3 violations of multi-regression assumptions:

Answer 20

1. heteroskedasticity
2. serial correlation
3. multico

Question 20

homoskedasticty

Answer 20

variation of errors is constant across observations

Question 21

whats the result of heteroskedasticity

Answer 21

inflated t-statistics and underestimated errors

Question 22

serial correlation

Answer 22

regression error terms are correlated across observations

Question 22

multicollinearity

Answer 22

2 or more independent variables are highly correlated

Question 23

model characteristics:

Answer 23

1. based on cogent economic reasoning
2. variables appropriate for the model with particular roles
3. as simple as possible
4. examined for assumption violations
5. useful out of sample