Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors

Orawo Luke Akongo

doi:doi:10.11648/j.ijsd.20210704.15

| Peer-Reviewed

Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors

Orawo Luke Akongo

Published in International Journal of Statistical Distributions and Applications (Volume 7, Issue 4)

Received: 21 October 2021 Accepted: 12 November 2021 Published: 24 November 2021

Views: Downloads:

Download PDF

Share This Article

Twitter
Linked In
Facebook

Abstract

The estimators of the slope and the intercept of simple linear regression model with normal errors are normally distributed and their exact confidence intervals are constructed using the t-distribution. However, when the normality assumption is not fulfilled, it is not possible to obtain exact confidence intervals. The Wald method of interval estimation is commonly used to provide approximate confidence intervals in such cases, and since it is derived from the central limit theorem it requires large samples in order to provide reliable approximate confidence intervals. This paper considers an alternative method of constructing approximate confidence intervals for the parameters of a simple linear regression model with Cauchy errors which is based the normal approximation to the Cauchy likelihood. The normal approximation to the Cauchy likelihood is obtained by a Tailor series expansion of the Cauchy log-likelihood function about the maximum likelihood estimate of the parameters and ignoring terms of order greater two. The maximized relative log-likelihood function for each parameter is then derived from the normal Cauchy relative log-likelihood function. The approximate confidence intervals for the parameters are constructed from their respective maximized relative log-likelihood functions. These confidence intervals have closed form confidence limits, are short and have coverage probabilities close to the nominal value 0.95.

Published in	International Journal of Statistical Distributions and Applications (Volume 7, Issue 4)
DOI	10.11648/j.ijsd.20210704.15
Page(s)	108-114
Creative Commons	This is an Open Access article, distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium or format, provided the original work is properly cited.
Copyright	Copyright © The Author(s), 2021. Published by Science Publishing Group

Keywords

Normal Approximations, Relative Likelihood Function, Maximized Relative Likelihood Function, Likelihood Confidence Intervals

References

[1]	Akkaya, A. D. and Tiku, M. L. (1999), Estimating parameters in autoregressive models in non- normal situations: Asymmetric innovations. Communications in Statistics Theory and Methods. 30. 517-536.
[2]	Bartolucci, F. and Scaccia, L. (2005). The Use of Mixtures for Dealing with Non-normal Regression Errors. Computational Statistics and Data Analysis. 48, 821-834.
[3]	Bian, G., Tiku, L. M., 1997. Bayesian inference based on robust priors and MML estimators: Part I, symmetric location-scale distributions. Statistics. 29, 317-345.
[4]	Bianco, A. M., Gercia, B. M. and Victor, J. (2005). Robust Estimation for Linear Regression with Asymmetric Errors. Canadian Journal of Statistics. 33: 511-528.
[5]	Fernandez, C. and Steel, M. F. J. (1998). On Bayesian Modeling of Fat Tails and Skewness. Journal of the American Statistica Association. 93: 359-371.
[6]	Fernandez, C. and Steel, M. F. J. (1999). Multivariate Student-t regression models: Pitfalls and inference. Biometrika. 86: 153-167
[7]	Goria, M. (1978). Fractional Absolute Moments of the Cauchy Distribution, Quaderni di Statistica e Matematica Applicataalle Scienze Economico-Sociali, University of Trento, Trento. 1: 89-96.
[8]	Geweke, J. (1993). Bayesian Treatment of the Independent Student-t Linear Model. Journal of Applied Econometrics. 8: 19-40.
[9]	Howard, M.(2018). Comparison of the Performance of Simple Linear Regression and Quantile Regression with Non-Normal Data: A Simulation Study. (Doctoral dissertation).
[10]	Howlader, H. A., and Weiss, G. (1988). On Bayesian Estimation of the Cauchy Parameters. The Indian Journal of Statistics, Series B, 50: 350-361.
[11]	Islam, M. Q., Tiku, M. L. and Yildirim, F. (2001). Non-normal Regression. I. Skew Distributions, Communications in Statistics - Theory and Methods. 30: 993-1020.
[12]	Johnson, N. L., Kotz, S. and Balakrishnan, N. (2004). Continuous Univariate Distributions. Vol. 1, 2^nd edition, John Wiley & sons Inc, New York.
[13]	Kadiyala, K. R. and Murthy, K. S. R. (1977). Estimation of Regression Equation with Cauchy Disturbances, Canadian Journal of Statistics. 5: 111-120.
[14]	Kalbfleisch, L. G. (1985). Probability and Statistical Inference. 2^nd edition, Springer-Verag, New York.
[15]	Koutrouvelis, I. A. (1981). Estimation of Location and Scale in Cauchy Distributions Using the Empirical Characteristics Function. Biometrika. 69: 205-213.
[16]	Lawless, J. F. and Singhal, K. (1978). Effective Screening of Non-normal regression Models. Biometrics. 34: 318-327.
[17]	Mahdizadeh M. and Zamanzade (2019). Goodness-of-fit testing for the Cauchy Distribution with Application to Financial Modeling, Journal of King Saud University-Science. 31: 1167-1174.
[18]	Stuart, A. and Ord, J. K. (1994). Kendall’s Advanced Theory of Statistics. 6^th edition, Vol. 1, Distribution Theory, Edward Anorld, London.
[19]	Tiku, M. L., Wong W. K., and Bian, G. (1999), Estimating parameters in autoregressive models in non-normal situations. Communications in Statistics - Theory and Methods. 28: 315-341.
[20]	Tiku, M. L., Islam, M. Q., and Selçuk, A. S. (2001). Non-normal Regression. II. Skew Distributions. Communications in Statistics - Theory and Methods. 30: 1021-1045.
[21]	Wong, K. W. and Bian, G. (2004). Robust Estimation of Multiple Regression Model with Non-normal Error: Symmetric Distribution. Monash Economics Working Papers. 09: 1-17.
[22]	Wong, K. W. and Bian, G. (2000). Robust Bayesian Inference in Capital Asset Pricing Model. Journal of Applied Mathematics & Decision Sciences, 4: 65-82.
[23]	Xiang L, Wanling W., Ecosse L. L., Tien and Wong (2012). Are Linear Regression Techniques Appropriate for Analysis When the Dependent Variable is not Normally Distributed? Investigative Ophthalmology & Visual Science Journal, 53: 3082-3083.

Cite This Article

Plain Text BibTeX RIS

APA Style

Orawo Luke Akongo. (2021). Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors. International Journal of Statistical Distributions and Applications, 7(4), 108-114. https://doi.org/10.11648/j.ijsd.20210704.15

Copy | Download

ACS Style

Orawo Luke Akongo. Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors. Int. J. Stat. Distrib. Appl. 2021, 7(4), 108-114. doi: 10.11648/j.ijsd.20210704.15

Copy | Download

AMA Style

Orawo Luke Akongo. Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors. Int J Stat Distrib Appl. 2021;7(4):108-114. doi: 10.11648/j.ijsd.20210704.15

Copy | Download

@article{10.11648/j.ijsd.20210704.15,
  author = {Orawo Luke Akongo},
  title = {Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors},
  journal = {International Journal of Statistical Distributions and Applications},
  volume = {7},
  number = {4},
  pages = {108-114},
  doi = {10.11648/j.ijsd.20210704.15},
  url = {https://doi.org/10.11648/j.ijsd.20210704.15},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.ijsd.20210704.15},
  abstract = {The estimators of the slope and the intercept of simple linear regression model with normal errors are normally distributed and their exact confidence intervals are constructed using the t-distribution. However, when the normality assumption is not fulfilled, it is not possible to obtain exact confidence intervals. The Wald method of interval estimation is commonly used to provide approximate confidence intervals in such cases, and since it is derived from the central limit theorem it requires large samples in order to provide reliable approximate confidence intervals. This paper considers an alternative method of constructing approximate confidence intervals for the parameters of a simple linear regression model with Cauchy errors which is based the normal approximation to the Cauchy likelihood. The normal approximation to the Cauchy likelihood is obtained by a Tailor series expansion of the Cauchy log-likelihood function about the maximum likelihood estimate of the parameters and ignoring terms of order greater two. The maximized relative log-likelihood function for each parameter is then derived from the normal Cauchy relative log-likelihood function. The approximate confidence intervals for the parameters are constructed from their respective maximized relative log-likelihood functions. These confidence intervals have closed form confidence limits, are short and have coverage probabilities close to the nominal value 0.95.},
 year = {2021}
}

Copy | Download

TY  - JOUR
T1  - Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors
AU  - Orawo Luke Akongo
Y1  - 2021/11/24
PY  - 2021
N1  - https://doi.org/10.11648/j.ijsd.20210704.15
DO  - 10.11648/j.ijsd.20210704.15
T2  - International Journal of Statistical Distributions and Applications
JF  - International Journal of Statistical Distributions and Applications
JO  - International Journal of Statistical Distributions and Applications
SP  - 108
EP  - 114
PB  - Science Publishing Group
SN  - 2472-3509
UR  - https://doi.org/10.11648/j.ijsd.20210704.15
AB  - The estimators of the slope and the intercept of simple linear regression model with normal errors are normally distributed and their exact confidence intervals are constructed using the t-distribution. However, when the normality assumption is not fulfilled, it is not possible to obtain exact confidence intervals. The Wald method of interval estimation is commonly used to provide approximate confidence intervals in such cases, and since it is derived from the central limit theorem it requires large samples in order to provide reliable approximate confidence intervals. This paper considers an alternative method of constructing approximate confidence intervals for the parameters of a simple linear regression model with Cauchy errors which is based the normal approximation to the Cauchy likelihood. The normal approximation to the Cauchy likelihood is obtained by a Tailor series expansion of the Cauchy log-likelihood function about the maximum likelihood estimate of the parameters and ignoring terms of order greater two. The maximized relative log-likelihood function for each parameter is then derived from the normal Cauchy relative log-likelihood function. The approximate confidence intervals for the parameters are constructed from their respective maximized relative log-likelihood functions. These confidence intervals have closed form confidence limits, are short and have coverage probabilities close to the nominal value 0.95.
VL  - 7
IS  - 4
ER  -

Copy | Download

Author Information

Orawo Luke Akongo

Mathematics Department, Egerton University, Njoro, Kenya

Download PDF

Sections

Plain Text BibTeX RIS

APA Style

Orawo Luke Akongo. (2021). Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors. International Journal of Statistical Distributions and Applications, 7(4), 108-114. https://doi.org/10.11648/j.ijsd.20210704.15

Copy | Download

ACS Style

Orawo Luke Akongo. Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors. Int. J. Stat. Distrib. Appl. 2021, 7(4), 108-114. doi: 10.11648/j.ijsd.20210704.15

Copy | Download

AMA Style

Orawo Luke Akongo. Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors. Int J Stat Distrib Appl. 2021;7(4):108-114. doi: 10.11648/j.ijsd.20210704.15

Copy | Download

@article{10.11648/j.ijsd.20210704.15,
  author = {Orawo Luke Akongo},
  title = {Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors},
  journal = {International Journal of Statistical Distributions and Applications},
  volume = {7},
  number = {4},
  pages = {108-114},
  doi = {10.11648/j.ijsd.20210704.15},
  url = {https://doi.org/10.11648/j.ijsd.20210704.15},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.ijsd.20210704.15},
  abstract = {The estimators of the slope and the intercept of simple linear regression model with normal errors are normally distributed and their exact confidence intervals are constructed using the t-distribution. However, when the normality assumption is not fulfilled, it is not possible to obtain exact confidence intervals. The Wald method of interval estimation is commonly used to provide approximate confidence intervals in such cases, and since it is derived from the central limit theorem it requires large samples in order to provide reliable approximate confidence intervals. This paper considers an alternative method of constructing approximate confidence intervals for the parameters of a simple linear regression model with Cauchy errors which is based the normal approximation to the Cauchy likelihood. The normal approximation to the Cauchy likelihood is obtained by a Tailor series expansion of the Cauchy log-likelihood function about the maximum likelihood estimate of the parameters and ignoring terms of order greater two. The maximized relative log-likelihood function for each parameter is then derived from the normal Cauchy relative log-likelihood function. The approximate confidence intervals for the parameters are constructed from their respective maximized relative log-likelihood functions. These confidence intervals have closed form confidence limits, are short and have coverage probabilities close to the nominal value 0.95.},
 year = {2021}
}

Copy | Download

TY  - JOUR
T1  - Likelihood-Based Confidence Intervals for the Parameters of a Simple Linear Regression Model with Cauchy Errors
AU  - Orawo Luke Akongo
Y1  - 2021/11/24
PY  - 2021
N1  - https://doi.org/10.11648/j.ijsd.20210704.15
DO  - 10.11648/j.ijsd.20210704.15
T2  - International Journal of Statistical Distributions and Applications
JF  - International Journal of Statistical Distributions and Applications
JO  - International Journal of Statistical Distributions and Applications
SP  - 108
EP  - 114
PB  - Science Publishing Group
SN  - 2472-3509
UR  - https://doi.org/10.11648/j.ijsd.20210704.15
AB  - The estimators of the slope and the intercept of simple linear regression model with normal errors are normally distributed and their exact confidence intervals are constructed using the t-distribution. However, when the normality assumption is not fulfilled, it is not possible to obtain exact confidence intervals. The Wald method of interval estimation is commonly used to provide approximate confidence intervals in such cases, and since it is derived from the central limit theorem it requires large samples in order to provide reliable approximate confidence intervals. This paper considers an alternative method of constructing approximate confidence intervals for the parameters of a simple linear regression model with Cauchy errors which is based the normal approximation to the Cauchy likelihood. The normal approximation to the Cauchy likelihood is obtained by a Tailor series expansion of the Cauchy log-likelihood function about the maximum likelihood estimate of the parameters and ignoring terms of order greater two. The maximized relative log-likelihood function for each parameter is then derived from the normal Cauchy relative log-likelihood function. The approximate confidence intervals for the parameters are constructed from their respective maximized relative log-likelihood functions. These confidence intervals have closed form confidence limits, are short and have coverage probabilities close to the nominal value 0.95.
VL  - 7
IS  - 4
ER  -

Copy | Download