# Module08 - Statistical Modelling for Business QBUS2810...

• LouiseWei
• 46

This preview shows pages 1–15. Sign up to view the full content.

Statistical Modelling for Business QBUS2810 Module 8: Statistical Inference for MLR II Marcel Scharth Discipline of Business Analytics, The University of Sydney Business School 1 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

MLR modules Multiple linear regression. Statistical inference for MLR I. Statistical inference for MLR II . 2 / 46
Outline Module 8: Statistical Inference for MLR II Standardised coefficients. Inference for one coefficient. F-tests for multiple coefficients. Appendix Reference: Fox Section 6.2. 3 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

Standardised coefficients Standard errors The standard error for b β j is SE ( b β j ) = SER q ( 1 - R 2 j ) ( n i =1 ( x ji - x j ) 2 ) , where the standard error of the regression SER = v u u t 1 n - p - 1 n X i =1 e 2 i estimates σ . 4 / 46
Standardised coefficients Robust standard errors You should use robust standard errors if the assumption of constant error variance is not satisfied for the data. We do not discuss the technical details of robust standard errors for the MLR case. 5 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

Standardised coefficients Sampling distribution in Gaussian MLR model If the errors are Gaussian, we can show that b β j - β j SE ( b β j ) t n - p - 1 6 / 46
Standardised coefficients Sampling distribution (general case) If the distribution of the errors is unspecified but the sample size is sufficiently large, we use the CLT approximation: b β j - β j SE ( b β j ) N (0 , 1) or b β j - β j SE ( b β j ) t n - p - 1 7 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

Topics Standardised coefficients. Inference for one coefficient . F-tests for multiple coefficients. Appendix 8 / 46
Inference for one coefficient Hypothesis testing Relationship test: Two-sided : H 0 : β j = 0 H 1 : β j 6 = 0 One-sided : H 0 : β j = 0 H 1 : β j > 0 9 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

Inference for one coefficient Hypothesis testing H 0 : β j = 0 H 1 : β j 6 = 0 Interpretation: If the null hypothesis is correct, there is no relationship between predictor j and the response conditional on the other predictors. Alternatively, we can say that variable j does not predict the response, after taking the other predictors into account. 10 / 46
Inference for one coefficient Hypothesis testing General test: Two-sided : H 0 : β j = b H 1 : β j 6 = b One-sided : H 0 : β j = b H 1 : β j > b 11 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

Inference for one coefficient Test statistic t b β j = b β j - b j SE ( b β j ) t n - p - 1 We can carry out hypothesis testing on β j using the standard t-statistic with degrees of freedom n - p - 1 . The test is a large sample approximation when the errors are non-Gaussian. 12 / 46
Hypothesis testing Summary One-tailed (left) Two-tailed One-tailed (right) H 0 : β j = b H 0 : β j = b H 0 : β j = b H 1 : β j < b H 1 : β j 6 = b H 1 : β j > b Test statistic t b β j = b β j - b SE ( b β j ) t n - p - 1 (under H 0 and assumptions) Rejection region t b β j < - t n - p - 1 t b β j < - t n - p - 1 ,α/ 2 t b β j > t n - p - 1 and t b β 1 > t n - p - 1 ,α/ 2 p-value ( t b β j t n - p - 1 ) P ( t n - p - 1 < t b β j ) 2 × P ( t n - p - 1 > | t b β j | ) P ( t n - p - 1 > t b β j ) 13 / 46

This preview has intentionally blurred sections. Sign up to view the full version.

Inference for one coefficient Why the p-value does not measure coefficient importance t b β j = b β j SER h ( 1 - R 2 j ) ( n - 1) s 2 x j i 1 / 2 Interpretation, all else equal: A higher coefficient magnitude leads to a larger test statistic.
This is the end of the preview. Sign up to access the rest of the document.
• One '14
• Statistical hypothesis testing, fstat, multiple coefficients

{[ snackBarMessage ]}

### What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern