QM

Question 1

P-valor?

Answer 1

P-valor menor que o nível de significância, rejeita Ho.

Question 2

Fórmula e o que é o R^2?

Answer 2

R2 = RSS/SST
Quanto da variável dependente é explicado pelas independentes

Question 3

5 assumptions de multiple regression

Answer 3

1. Linear relationship between x and y
2. Residuals are normally distributed
3. Variance of the error term is constant
4. Residual of one of obs is not correlated with a another
5. Independent variables are not random

Question 4

Pq usar o R2 ajustado?

Answer 4

Pq ele ajusta o R2 pelo número de variáveis que você tem. Evita overfutting the model.

Question 5

Outros dois métodos de avaliação de modelos de regressão?

Answer 5

AIC: utilizado para um modelo de previsão (futuro)

BIC: Avaliar o modelo atual

MENOR VALOR MELHOR!

Question 6

What’s the Join- F test?

Answer 6

Used to jointly test a subset of variables in a multiple regression. Restricted model (less variables) and Unrestricted model (broader).

• checar se as variáveis adicionais fazem sentido para o modelo (diferente de zero)

Question 7

O que é héterecedasticidade e está relacionada com qual assumption?

Answer 7

• Corr (x,e) dif 0
• Assumption: error term has constant variance. It violated when subsamples are more spread out then others.

Question 8

What are the two types of heterecedasticidade?

Answer 8

Type 1: Unconditional
Not related to the independent variable (no problem)

Type 2: Conditional
Related to the independent variable (is a problem
- Afeta o desvio padrão (unreliable)
- Ocorre type I error

Question 9

What are the two types of heterecedasticidade?

Answer 9

Type 1: Unconditional
Not related to the independent variable (no problem)

Type 2: Conditional
Related to the independent variable (is a problem
- Afeta o desvio padrão (unreliable)
- Ocorre type I error

Question 10

How to detect heterocedasticidade?

Answer 10

1. Scatter diagramas (via gráfico)
2. Breush- Pagan (BP): teste chi quadrado nos residuals com a independent variable.

Question 11

How to correct Heterecedasticidade?

Answer 11

Calculate the robust standard error and then use to recalculate the t-statistic

Question 12

What is serial correlation and is related with witch assumption?

Answer 12

1.Corr (e, e) dif 0
2. Tô small standard error ( t=b/s)
3. Ocorre erro tipo I
Assumption: Error terms are no correlated

Question 13

Type I error?

Answer 13

Reject something that is true

Question 14

Tô detect serial correlation?

Answer 14

1. Scatter plots
2. Derbin-Watson (1 lag)
3. Breush-Godfrey (BG): Use the residuals from the original regression as the dependent variable

Question 15

Tô correct serial correlation?

Answer 15

Use robust standard errors (usar novos desvios padrões)

Question 16

O que e multicolinearidade?

Answer 16

Cor (x, x) dif 0
Quando tem correlação entre a variáveis independentes.

Essa condição infla o standard error e torna variáveis falsamente importantes (Type II error)

Question 17

Detecting multicolinearidade

Answer 17

VIF (Variance inflation factor)

VIF = 1 / (1-R2)

VIF=1 (ok)
VIF > 5 (ruim)
VIF > 10 (péssimo) - alta multicolinearidade

Question 18

Correção de multicolinearidade

Answer 18

1.Tira uma variável e vê o que acontece
2. Usar outra proxy
3. Aumentar a amostra

Question 19

Dummy variable trap?

Answer 19

Use (n-1) dummy variables to avoid multicolinearidade

Ex: 3 dummies para avaliar 4 anos

Question 20

Métodos de regressa para variáveis dependentes qualitativas?

Answer 20

1. Logit models - uses log odds.

Ln(p/1-p)