Module 2.6 Heteroskedasticity

Question 1

one assumption of multiple regression is variance of residuals is _____

Answer 1

constant across observations

Question 2

one assumption of multiple regression is what kind of relationship exists between independent and dependent variables?

Answer 2

linear

Question 3

assumption of multiple regression is independent variables:

Answer 3

are NOT random and no exact relationship between any two or more independent variables

Question 4

assumption of multiple regression is expected value of error term =

Answer 4

0

Question 5

final two assumptions of multiple regression (related to error term)

Answer 5

error term for one observation is not correlated with that of another and error term is NORMALLY distributed

Question 6

occurs when variance of residuals is NOT the same across all observations

Answer 6

heteroskedasticity

Question 7

occurs when heteroskedasticity is NOT related to level of independent variables, which means it doesn’t systematically increase or decrease with changes in value of independent variables

Answer 7

Unconditional heteroskedasticity

Question 8

usually causes NO MAJOR problems with regression

Answer 8

unconditional HS

Question 9

HS that is related to level of independent variables; exists if variance of residual term increases as value of independent variable increases

Answer 9

conditional HS

Question 10

creates significant problems for statistical inference

Answer 10

conditional HS

Question 11

four effects of HS

Answer 11

1. standard errors are usually unreliable estimates
2. coefficient estimates NOT affected
3. if standard errors are too small, but coefficient estimates not affected, t statistics will be too large and null is rejected too often (opposite if SE is too large)
4. F test is UNRELIABLE

Question 12

two methods to detect HS

Answer 12

1. examine scatter plots

2. chi square test

Question 13

test statistic for Breusch-Pagan (chi square) test

Answer 13

= n x R^2 with k degrees of freedom

Question 14

R^2 used in BP (chi square) test is:

Answer 14

from a second regression

Question 15

chi square test has how many tails

Answer 15

one tail

Question 16

most common remedy to account for HS

Answer 16

robust standard errors

Question 17

robust standard errors are used to:

Answer 17

recalculate t statistics using original regression coefficients

Question 18

second method to correct for HS is use of _____ which attempts to eliminate the HS by modifying original equation

Answer 18

generalized least squares