A Bayesian approach to retransformation bias in transformed regression.

@article{Stow2006ABA,
  title={A Bayesian approach to retransformation bias in transformed regression.},
  author={Craig A. Stow and Kenneth H. Reckhow and Song S. Qian},
  journal={Ecology},
  year={2006},
  volume={87 6},
  pages={
          1472-7
        }
}
Ecological data analysis often involves fitting linear or nonlinear equations to data after transforming either the response variable, the right side of the equation, or both, so that the standard suite of regression assumptions are more closely met. However, inference is usually done in the natural metric and it is well known that retransforming back to the original metric provides a biased estimator for the mean of the response variable. For the normal linear model, fit under a log… 

Figures from this paper

Nonlinear parsimonious forest modeling assuming normal distribution of residuals
To avoid the transformation of the dependent variable, which introduces bias when back-transformed, complex nonlinear forest models have the parameters estimated with heuristic techniques, which can
A parsimonious approach for modeling uncertainty within complex nonlinear relationships
TLDR
The objective of this research was to provide scientists focused on modeling ecosystem processes with a procedure that supplies parsimonious correct results, and indicates that uncertainty, measured by variance, impacts the analysis in a non-intuitive way when the defining domain of the response variable is restricted.
Explaining the Shortcomings of Log‐Transforming the Dependent Variable in Regression Models and Recommending a Better Alternative: Evidence From Soil CO2 Emission Studies
Log‐transforming the dependent variable of a regression model, though convenient and frequently used, is accompanied by an under‐prediction problem. We found that this underprediction can reach up to
Development of Nonlinear Parsimonious Forest Models Using Efficient Expansion of the Taylor Series: Applications to Site Productivity and Taper
The parameters of nonlinear forest models are commonly estimated with heuristic techniques, which can supply erroneous values. The use of heuristic algorithms is partially rooted in the avoidance of
Random intercept and linear mixed models including heteroscedasticity in a logarithmic scale: Correction terms and prediction in the original scale
TLDR
The correction terms, simulations and real data about the income of elderly are presented to show the importance of using them to obtain more accurate predictions.
Bayesian Hierarchical/Multilevel Models for Inference and Prediction Using Cross-System Lake Data
TLDR
This work uses data from 382 Michigan lakes and reservoirs to develop and evaluate several alternative multilevel models for predicting chlorophyll a concentration from total phosphorus concentration.
Modeling hypoxia in the Chesapeake Bay: Ensemble estimation using a Bayesian hierarchical model
Calculating the variance and prediction intervals for estimates obtained from allometric relationships
Often researchers are interested in obtaining estimates of variables which are quite difficult or expensive to measure. To obtain these estimates, relationships between those variables of interest
Including a Log-Transform of the Data
In this chapter, the Bayesian hierarchical space-time model developed in Chap. 3 has been fitted to log-transformed data of significant wave height. There are two main motivations for this;
...
...

References

SHOWING 1-10 OF 24 REFERENCES
Bias in Hydrologic Prediction Using Log-Transformed Regression Models
: Hydrologic variables are related through a complex set of dynamic processes. Due to this complexity, empirical, usually statistical, models are used for the synthesis of records or extentions of
Estimating constituent loads
Several recent articles have called attention to the problem of retransformation bias, which can arise when log linear regression models are used to estimate sediment or other constituent loads. In
Regression analysis of log‐transformed data: Statistical bias and its correction
Power and exponential relationships are common in most quantitative disciplines. In environmental chemistry and toxicology, predictive applications range from flow-related variation in water quality
Logarithmic transformation bias in allometry
It has been known for some time (DJ Finney, J. Roy. Stat. Soc. Suppl. 7:155–161, 1941) that transformation of an arithmetic data set to logarithms results in biased estimates when predicted values
Estimator bias in a lake phosphorus model with observation error
We investigated observation-error-induced parameter bias using least squares and Stein-corrected least squares estimators in a model for predicting lake phosphorus. The Stein-corrected estimator
Reducing Transformation Bias in Curve Fitting
Figure 1. Scatter diagram and curve resulting from inverse transformation, for sales (Y) vs. number of employees (X). which is graphed in Figure 1 (the lower curve), along with the original data
A Language and Program for Complex Bayesian Modelling
TLDR
This work describes some general purpose software that is currently developing for implementing Gibbs sampling: BUGS (Bayesian inference using Gibbs sampling), written in Modula-2 and runs under both DOS and UNIX.
Estimation in Lognormal Linear Models
Abstract Let Y = ey where the usual assumptions of the general linear models: y = Xa+u, u ∼ N(0, σ2 I) are made. Extending the results of Finney minimum variance unbiased estimators, τ, of parametric
Correcting for Bias in Log‐Transformed Allometric Equations
TLDR
The role of cereal growth stages and crowding in the induction of alatae in Sitobion a'enae and its consequences for population growth are studied.
Empirical models for trophic state in southeastern U.S. lakes and reservoirs
: A cross-sectional data set of 80 lakes and reservoirs in nine southeastern states was examined to specify and parameterize trophic state relationships. The relationships fitted are based on
...
...