Fichas
Quantitative Methods - Multiple Regression and Issues in Regression Analysis
Fichas del conjunto (8)
- 1Quantitative Methods - Multiple Regression and Issues in Regression Analysis Assumptions of Multiple Regression
- 2Quantitative Methods - Multiple Regression and Issues in Regression Analysis F-stat formula and H0, Ha, df definitions and decision ruleF = MSR / MSE = (RSS/k)/(SSE/n-k-1)
H0: All 'b's =0
Ha: At least one b <> 0
dfnumerator = k
dfdenominator = n-k-1
Decision rule: reject H0 if F-stat > Fcrit - 3Quantitative Methods - Multiple Regression and Issues in Regression Analysis R2adjR2adj = 1 - { [(n-1)/(n-k-1)] * (1-R2) }
- 4Quantitative Methods - Multiple Regression and Issues in Regression Analysis What to remember about dummy variables?1 - Always use 1 less dummy variable than states of the world
2 - 1 dummy = 1, rest = 0 - 5Quantitative Methods - Multiple Regression and Issues in Regression Analysis Define Heteroskedasticity (Types, implication, How to detect, How to correct)Heteroskedasticity
Definition: variance of residual not cte
Types: unconditional (NO problem) and conditional (PROBLEM)
Implication: low SE => high t-stat (Type I error)
How to detect: Breusch-Pagan chi-squared test = n*Rresid^2
How to correct: White corrected - 6Quantitative Methods - Multiple Regression and Issues in Regression Analysis Define Serial Correlation (Types, implication, How to detect, How to correct)Serial Correlation
Definition: residuals correlated one another
Types: positive and negative
Implication: low SE => high t-stat (Type I error)
How to detect: Durbin-Watson = 2(1-r)
How to correct: Hansen Method - 7Quantitative Methods - Multiple Regression and Issues in Regression Analysis Define Multicolinearity (Types, implication, How to detect, How to correct)Multicolinearity
Definition: independent variables are correlated one another
Types =
Implication: high SE => low t-stat (Type II error)
How to detect: high F + low t's (significant F and insignificant t's)
How to correct: omit one or more X - 8Quantitative Methods - Multiple Regression and Issues in Regression Analysis 6 Model Mispecification1 - Omiting a variable
2 - Variable should be transformed
3 - Incorrectly pooling data
4 - Using lagged dependent vbl. as independent vbl.
5 - Forecasting the past
6 - Measuring independet variables with error
Información
- Descripción
- CFA (Quantitative Methods) CFA Level 2 Fichas sobre Quantitative Methods - Multiple Regression and Issues in Regression Analysis, creado por Victor Koyama el 23/05/2016.
- Creado por
- Victor Koyama
- Visitas
- 1
- Etiquetas
- Valoración