Timothy E. O’Brien,Jack W. Silcox
Abstract:Use of nonlinear statistical methods and models are ubiquitous in scientific research. However, these methods may not be fully understood, and as demonstrated here, commonly-reported parameter p-values and confidence intervals may be inaccurate. The gentle introduction to nonlinear regression modelling and comprehensive illustrations given here provides applied researchers with the needed overview and tools to appreciate the nuances and breadth of these important methods. Since these methods build upon topics covered in first and second courses in applied statistics and predictive modelling, the target audience includes practitioners and students alike. To guide practitioners, we summarize, illustrate, develop, and extend nonlinear modelling methods, and underscore caveats of Wald statistics using basic illustrations and give key reasons for preferring likelihood methods. Parameter profiling in multiparameter models and exact or near-exact versus approximate likelihood methods are discussed and curvature measures are connected with the failure of the Wald approximations regularly used in statistical software. The discussion in the main paper has been kept at an introductory level and it can be covered on a first reading; additional details given in the Appendices can be worked through upon further study. The associated online Supplementary Information also provides the data and R computer code which can be easily adapted to aid researchers to fit nonlinear models to their data.
What problem does this paper attempt to address?
The problems that this paper attempts to solve mainly focus on the understanding and application of nonlinear regression modeling methods. Specifically:
1. **Popularization and Understanding of Non - linear Statistical Methods**: Although non - linear statistical methods are widely used in scientific research, these methods may not be fully understood. The paper points out that the commonly used parameter p - values and confidence intervals may be inaccurate, especially in small - to medium - sized studies.
2. **Problems with Wald Statistics**: The paper emphasizes the limitations of Wald statistics in non - linear regression, especially that the p - values and confidence intervals it provides may be seriously inaccurate. For example, the actual coverage level of a nominally 95% confidence interval may be only 75% or lower.
3. **Advantages of Likelihood Methods**: Through basic diagrams and key reasons, the paper explains why likelihood - based methods are more preferred in non - linear regression. These methods can usually better match the nominal and actual coverage levels.
4. **Parameter Profiles in Multi - parameter Models**: It discusses parameter profiles in multi - parameter models as well as exact or approximate likelihood methods, and explores the relationship between curvature measures and the frequent failure of Wald approximations in statistical software.
5. **Basic Concepts and Methods of Non - linear Regression**: The paper provides a gentle introduction to non - linear regression modeling, including methods of parameter estimation, standard error, point estimation and interval estimation. In addition, it also introduces how to solve non - linear normal equations by numerical methods and computer algorithms.
6. **Application Examples**: Through specific examples (such as the Michaelis - Menten enzyme kinetics model and the analysis of fungal growth data), it shows the application and advantages of non - linear regression models in actual research.
Overall, this paper aims to provide applied researchers with a comprehensive overview of non - linear regression modeling, help them understand and apply these important statistical methods, point out the limitations of Wald statistics, and recommend the use of likelihood - based methods.