Journals / İstatistik Araştırma Dergisi / 2025 / Cilt: 15 Sayı: 2

An Applied Comparison of Beta Regression Approaches: An Examination of Constant and Variable Dispersion Models

Beta Regresyon Yaklaşımlarının Uygulamalı Karşılaştırması: Sabit ve Değişken Saçılım Modelleri Üzerine Bir İnceleme

Onur Şentürk(GAZİ ÜNİVERSİTESİ, FEN BİLİMLERİ ENSTİTÜSÜ)0000-0002-6752-4963
Pages
32–43
Publication Date
2025-12-31
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The findings indicate that the choice of link function and the modeling of variance structure have significant effects on the performance of beta regression models; in particular, accounting for varying dispersion improves model explanatory power.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:In this study, the applicability of beta regression and varying dispersion beta regression models in modeling proportion-type dependent variables was evaluated using two different datasets.
📊
Key Findings & Data:The first dataset includes macroeconomic indicators of European Union member and candidate countries, and classical beta regression models were applied under the assumption of constant dispersion.
💡
Conclusion & Impact:The findings indicate that the choice of link function and the modeling of variance structure have significant effects on the performance of beta regression models; in particular, accounting for varying dispersion improves model explanatory power.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Beta regresyon · Değişken saçılım · Bağlantı Fonksiyonu

Abstract

In this study, the applicability of beta regression and varying dispersion beta regression models in modeling proportion-type dependent variables was evaluated using two different datasets. The first dataset includes macroeconomic indicators of European Union member and candidate countries, and classical beta regression models were applied under the assumption of constant dispersion. In the analyses comparing different link functions (logit, probit, clog-log, cauchit, and log-log), the clog-log link function was identified as the most suitable model, with the highest log-likelihood and Pseudo R² values and the lowest AIC and BIC values. In the second dataset, an analysis was conducted to explain the proportion of belief in God at the country level. In this dataset, both constant and varying dispersion beta regression models were estimated, where the mean and variance structures were modeled as functions of covariates. As a result of the model comparisons, the varying dispersion logit model emerged as the best-performing model, with the lowest AIC and the highest log-likelihood and Pseudo R² values. The findings indicate that the choice of link function and the modeling of variance structure have significant effects on the performance of beta regression models; in particular, accounting for varying dispersion improves model explanatory power.

Özet

Bu çalışmada, oran türündeki bağımlı değişkenlerin modellenmesinde beta regresyon ve değişken saçılımlı beta regresyon modellerinin uygulanabilirliği iki farklı veri seti üzerinden değerlendirilmiştir. İlk veri seti, Avrupa Birliği üyesi ve aday ülkelerin makroekonomik göstergelerini içermekte olup sabit saçılım varsayımı altında klasik Beta Regresyon modelleri uygulanmıştır. Farklı bağlantı fonksiyonlarının (logit, probit, clog-log, cauchit, log-log) karşılaştırıldığı analizlerde, en yüksek log-olabilirlik ve Pseudo R² değerleri ile en düşük AIC ve BIC değerlerine sahip olan clog-log bağlantı fonksiyonu en uygun model olarak belirlenmiştir. İkinci veri setinde ise, ülke düzeyinde tanımlanmış Tanrı’ya inanç oranını açıklamaya yönelik bir analiz gerçekleştirilmiştir. Bu veri setinde ortalama ve varyans yapısının bağımsız değişkenlerle ilişkilendirildiği sabit ve değişken saçılımlı Beta Regresyon modelleri tahmin edilmiştir. Model karşılaştırmaları sonucunda, değişken saçılımlı logit modeli, en düşük AIC ve en yüksek log-olabilirlik ile Pseudo R² değerleriyle en başarılı model olarak öne çıkmıştır. Sonuçlar, bağlantı fonksiyonu seçiminin ve varyans modellemesinin beta regresyon modellerinin performansı üzerinde önemli etkileri olduğunu göstermekte; özellikle değişken saçılım yapısının dikkate alınmasının modelin açıklayıcılığını artırdığını ortaya koymaktadır.

🔖

Cite This Article

Şentürk, O., Özonur, D., Olmuş, H. (2025). An Applied Comparison of Beta Regression Approaches: An Examination of Constant and Variable Dispersion Models. İstatistik Araştırma Dergisi, Cilt: 15 Sayı: 2, 32–43. https://ajindex.com/articles/1617752

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (28)

  1. Abonazel, M. R., Dawoud, I., Awwad, F. A., & Lukman, A. F. (2022). Dawoud–Kibria estimator for beta regression model: simulation and application. Frontiers in Applied Mathematics and Statistics, 8, 775068.
  2. Abonazel, M. R., Said, H. A., Tag-Eldin, E., Abdel-Rahman, S., & Khattab, I. G. (2023). Using beta regression modeling in medical sciences: a comparative study. Commun. Math. Biol. Neurosci., 2023, Article-ID.
  3. Akram, M. N., Abonazel, M. R., Amin, M., Kibria, B. G., & Afzal, N. (2022). A new Stein estimator for the zero‐inflated negative binomial regression model. Concurrency and Computation: Practice and Experience, 34(19), e7045.
  4. Akrami, H., Zamzam, O., Joshi, A., Aydore, S., & Leahy, R. (2024, April). Beta quantile regression for robust estimation of uncertainty in the presence of outliers. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 7480-7484). IEEE.
  5. Al-Ayashy, H. L. K., & Alshaybawee, T. (2025). New robust beta regression estimation to overcome the effect of high leverage points. Journal of Information Systems Engineering and Management, 10(33s).
  6. Algamal, Z. Y., Abonazel, M. R., Awwad, F. A., & Eldin, E. T. (2023). Modified jackknife ridge estimator for the Conway-Maxwell-Poisson model. Scientific African, 19, e01543.
  7. Altun, E., El-Morshedy, M., & Eliwa, M. S. (2021). A new regression model for bounded response variable: An alternative to the beta and unit-Lindley regression models. Plos one, 16(1), e0245627.
  8. Aranda-Ordaz, F. J. (1981). On two families of transformations to additivity for binary response data. Biometrika, 68(2), 357-363.
  9. Awwad, F. A., Odeniyi, K. A., Dawoud, I., Algamal, Z. Y., Abonazel, M. R., Kibria, B. G., & Eldin, E. T. (2022). New two-parameter estimators for the logistic regression model with multicollinearity. WSEAS Trans. Math, 21, 403-414.
  10. Canterle, D. R., & Bayer, F. M. (2019). Variable dispersion beta regressions with parametric link functions. Statistical Papers, 60(5), 1541-1567.
  11. Cribari-Neto, F., & Souza, T. C. (2013). Religious belief and intelligence: Worldwide evidence. Intelligence, 41(5), 482-489.
  12. Cribari-Neto, F. (2023). A beta regression analysis of COVID-19 mortality in Brazil. Infectious Disease Modelling, 8(2), 309-317
  13. El-Raoof, Z. M. A., M El-Gohary, M., & G Yehia, E. (2023). Robust Estimation for Beta Regression Model in the Presence of Outliers: A Comparative Study. 43,والتمويل التجارة(3), 380-406.
  14. Eurostat. (2025). Eurostat veri tabanı. https://ec.europa.eu/eurostat/web/main/data/database
  15. Ferrari, S., Cribari-Neto, F. 2004. Beta regression for modelling rates and proportions. Journal of Applied Statistics, 31(7), 799-815.
  16. Figueroa-Zúñiga, J. I., Bayes, C. L., Leiva, V., & Liu, S. (2022). Robust beta regression modeling with errors-in-variables: a Bayesian approach and numerical applications. Statistical Papers, 1-24.
  17. Geissinger, E. A., Khoo, C. L., Richmond, I. C., Faulkner, S. J., & Schneider, D. C. (2022). A case for beta regression in the natural sciences. Ecosphere, 13(2), e3940.
  18. Guerrero, V. M., & Johnson, R. A. (1982). Use of the Box-Cox transformation with binary response models. Biometrika, 69(2), 309-314.
  19. Koenker, R., & Yoon, J. (2009). Parametric links for binary choice models: A Fisherian–Bayesian colloquy. Journal of Econometrics, 152(2), 120-130.
  20. Kubinec, R. (2023). Ordered beta regression: a parsimonious, well-fitting model for continuous data with lower and upper bounds. Political analysis, 31(4), 519-536.
  21. Maluf, Y. S., Ferrari, S. L., & Queiroz, F. F. (2024). Robust beta regression through the logit transformation. Metrika, 1-21.
  22. Queiroz, F. F., & Ferrari, S. L. (2023). Power logit regression for modeling bounded data. Statistical Modelling, 1471082X221140157.
  23. Ramalho, E. A., Ramalho, J. J., & Murteira, J. M. (2011). Alternative estimating and testing empirical strategies for fractional regression models. Journal of Economic Surveys, 25(1), 19-68.
  24. Ribeiro, T. K., & Ferrari, S. L. (2023). Robust estimation in beta regression via maximum Lq-likelihood. Statistical Papers, 64(1), 321-353.
  25. Seifollahi, S., Bevrani, H., & Månsson, K. (2025). Bayesian analysis of the beta regression model subject to linear inequality restrictions with application. Hacettepe Journal of Mathematics and Statistics, 1-15.
  26. Simas, A. B., Barreto-Souza, W., & Rocha, A. V. (2010). Improved estimators for a general class of beta regression models. Computational Statistics & Data Analysis, 54(2), 348-366.
  27. Smithson, M., Verkuilen, J. (2006). A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychological methods, 11(1), 54
  28. El-Raoof, Z. M. A., M El-Gohary, M., & G Yehia, E. (2023). Robust Estimation for Beta Regression Model in the Presence of Outliers: A Comparative Study. 43,والتمويل التجارة(3), 380-406.