Feasible generalized least squares correction of autocorrelation in interrupted time series models
DOI:
https://doi.org/10.64497/jssci.265Keywords:
interrupted time series, autocorrelation, Cochrane–Orcutt procedure, COVID-19 vaccination, feasible generalized least squaresAbstract
Interrupted Time Series (ITS) models are widely used to evaluate public health interventions but may produce biased estimates and misleading inferences when autocorrelation is ignored. This study applied a Feasible Generalized Least Squares (FGLS) approach with the Cochrane–Orcutt transformation to correct first-order autocorrelation in assessing the impact of the COVID-19 vaccination programme, which commenced in Nigeria on 5 March 2021, on weekly reported COVID-19 cases over 127 weeks. Diagnostic tests revealed substantial autocorrelation in the Ordinary Least Squares (OLS) residuals, which was effectively eliminated, improving the Durbin–Watson statistic from 0.64 to 1.94. Model comparison further showed that the Akaike Information Criterion (AIC) for the OLS model (2244.0) was higher than that of the Cochrane–Orcutt model (2158.4), indicating a superior fit for the corrected model. The adjusted model indicated a significant immediate reduction and sustained decline in COVID-19 cases following the introduction of vaccination, while counterfactual analysis estimated that approximately 82,631 infections were averted. Although the analysis did not explicitly account for seasonal effects, nonlinear trends, lockdown measures, viral variants, or higher-order autocorrelation structures, the findings highlight the importance of addressing serial correlation in ITS studies. The results demonstrate that the Cochrane–Orcutt method provides robust, reliable, and policy-relevant inference for evaluating epidemiological interventions and strengthening the validity of public health impact assessments.
Downloads
References
[1] C. Hategeka, H. Ruton, M. Karamouzian, L. D. Lynd, and M. R. Law, "Use of interrupted time series methods in the evaluation of health system quality improvement interventions: A methodological systematic review," BMJ Global Health, vol. 5, no. 10, Art. no. e003567, Oct. 2020, doi: 10.1136/bmjgh-2020-003567. DOI: https://doi.org/10.1136/bmjgh-2020-003567
[2] N. Marisetty, "Mitigating assumption violations in regression analysis: Insights from 20 years of currency pair data," SSRN Electronic Journal, Preprint, 2024, Art. no. 5068940. doi: 10.2139/ssrn.5068940. DOI: https://doi.org/10.2139/ssrn.5068940
[3] Z. Zhang, Z. Li, and Y. Song, "On ignoring the heterogeneity in spatial autocorrelation: Consequences and solutions," International Journal of Geographical Information Science, vol. 38, no. 12, pp. 2545-2571, 2024. DOI: https://doi.org/10.1080/13658816.2024.2391981
[4] B. H. Baltagi, "Autocorrelation in regression," in International Encyclopedia of Statistical Science, 2nd ed., M. Lovric, Ed. Berlin, Germany: Springer, 2025, pp. 131-134. DOI: https://doi.org/10.1007/978-3-662-69359-9_43
[5] B. Modu and A. M. Inuwa, "Time series regression modeling with AR(1) errors," European Journal of Statistics, vol. 3, pp. 13-13, 2023. DOI: https://doi.org/10.28924/ada/stat.3.13
[6] B. Modu, B. Maijamaa, and A. M. Mabu, "Interrupted time series analysis of COVID-19 cases and clinical interventions administered in Nigeria: A public health perspective," in Proc. 8th Int. Conf. Quantitative Sciences and Its Applications (ICOQSIA), Sintok, Kedah, Malaysia, Sept. 17-19, 2024.
[7] N. S. Altman, "An iterated Cochrane-Orcutt procedure for nonparametric regression," Journal of Statistical Computation and Simulation, vol. 40, nos. 1-2, pp. 93-108, 1992. DOI: https://doi.org/10.1080/00949659208811368
[8] M. Y. Al-Saqry, R. A. A. Eltayib, A. H. A. Al-Busaidi, and M. F. Chan, "Comparing sociodemographic, health status and resources, macroeconomic status, and environmental factors on infant mortality rates in Bahrain, Kuwait, and Oman: Longitudinal time-series study," JMIR Pediatrics and Parenting, vol. 8, no. 1, Art. no. e73203, 2025. DOI: https://doi.org/10.2196/73203
[9] S. Adrianto, I. H. N. Balqis, C. Z. N. Soetanto, and M. Ohyver, "Cochrane-Orcutt method to overcome autocorrelation in modeling factors affecting the number of hotel visitors in Indonesia," Procedia Computer Science, vol. 216, pp. 630-638, 2023. DOI: https://doi.org/10.1016/j.procs.2022.12.178
[10] L. Pizzimbone, "Harmonic regression analysis of time-series measurements," Authorea Preprints, 2023. DOI: https://doi.org/10.36227/techrxiv.24155097.v1
[11] K. Takahashi, "Tourism boom on island regions in Japan during the 70s and 80s: Analyzing with Cochrane-Orcutt estimation," Journal of Global Tourism Research, vol. 9, no. 2, pp. 145-148, 2024. DOI: https://doi.org/10.37020/jgtr.9.2_145
[12] N. K. Kumar, "Autocorrelation and heteroscedasticity in regression analysis," Journal of Business and Social Sciences, vol. 5, no. 1, pp. 9-20, 2023. DOI: https://doi.org/10.3126/jbss.v5i1.72442
[13] P. Wang and P. Wang, "Assessment on estimations of currency basket weights—With coefficient correction for common factor dominance," International Journal of Finance & Economics, vol. 27, no. 1, pp. 1401-1418, 2022. DOI: https://doi.org/10.1002/ijfe.2222
[14] M. Nguyen, "Linear regression," in Regression Techniques for Data Analysis. Cham, Switzerland: Springer Nature Switzerland, 2026, pp. 9-201. DOI: https://doi.org/10.1007/978-3-032-01835-9_2
[15] R. K. Hamza and W. J. Husain, "Robust methods for detecting and correcting autocorrelation in financial time series modeling," International Journal on Advanced Science, Engineering & Information Technology, vol. 15, no. 5, 2025. DOI: https://doi.org/10.18517/ijaseit.15.5.12901
[16] R. K. Hamza, "Detecting and correcting autocorrelation in time-series models of unemployment rates," Industrial Engineering & Management Systems, vol. 24, no. 4, pp. 509-517, 2025. DOI: https://doi.org/10.7232/iems.2025.24.4.509
[17] A. A. Chaudhary, N. Nazir, A. Riaz, N. Sadiq, and N. Riaz, "Autocorrelation: What happens if the error or disturbance terms are correlated in time-series data," Competitive Education Research Journal, vol. 3, no. 2, pp. 154-163, 2022.
[18] S. Sathyanarayana and T. Mohanasundaram, "Correlation methodologies revisited: From classical statistics to contemporary residual diagnostics," Asian Journal of Economics, Business and Accounting, vol. 25, no. 11, pp. 86-103, 2025. DOI: https://doi.org/10.9734/ajeba/2025/v25i112039
[19] G. Baiocchi and W. Distaso, "GRETL: Econometric software for the GNU generation," Journal of Applied Econometrics, vol. 18, no. 1, pp. 105-110, 2003. DOI: https://doi.org/10.1002/jae.704
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Bilkisu Maijamaa, Babagana Modu
This work is licensed under a Creative Commons Attribution 4.0 International License.
- Abstract 574
- PDF 182
