Median-Unbiased Estimation in Fixed-Effects Dynamic Panels

<p>The linear mixed model is a commonly used model for longitudinal or nested data due to its ability to account for the dependency of nested data. Researchers typically rely on the random effects to adequately account for the dependency due to correlated data, however serial correlation can also be used. If the random effect structure is misspecified (perhaps due to convergence problems), can the addition of serial correlation overcome this misspecification and allow for unbiased estimation and accurate inferences? This study explored this question with a simulation. Simulation results show that the fixed effects are unbiased, however inflation of the empirical type I error rate occurs when a random effect is missing from the model. Implications for applied researchers are discussed.</p>

Download Full-text

Median Unbiased Estimation for Binary Data

The American Statistician ◽

10.1080/00031305.1989.10475597 ◽

1989 ◽

Vol 43 (1) ◽

pp. 7-11 ◽

Cited By ~ 2

Author(s):

Karim F. Hirji ◽

Anastasios A. Tsiatis ◽

Cyrus R. Mehta

Keyword(s):

Binary Data ◽

Unbiased Estimation ◽

Median Unbiased Estimation

Download Full-text

Estimating Nonlinear Selection on Behavioral Reaction Norms

10.32942/osf.io/u26tz ◽

2021 ◽

Author(s):

Jordan Scott Martin

Keyword(s):

Individual Variation ◽

Fixed Effects ◽

Random Effect ◽

Generalized Solution ◽

Reaction Norm ◽

Reaction Norms ◽

Unbiased Estimation ◽

Behavioral Strategies ◽

Non Gaussian ◽

Parameter Values

Individuals' behavioral strategies are often well described by reaction norms, which are functions predicting repeatable patterns of personality, plasticity, and predictability across an environmental gradient. Reaction norms can be readily estimated using mixed-effects models and play a key role in current theories of adaptive individual variation. Unfortunately, however, it remains challenging to assess the effects of reaction norms on fitness-relevant outcomes, due to the high degree of uncertainty in random effect estimates of reaction norm parameters, also known as best linear unbiased predictors (BLUPs). Current approaches to this problem do not provide a generalized solution for modelling reaction norm effects with nonlinear structure, such as stabilizing, disruptive, balancing, and/or correlational selection, which are necessary for testing adaptive theory of individual variation. To address this issue, I present a novel solution for straightforward and unbiased estimation of linear and nonlinear reaction norm effects on fitness, applicable to both Gaussian and non-Gaussian measurements. This solution involves specifying BLUPs as random effects on behavior and fixed effects on fitness within a Bayesian multi-response model. By simultaneously accounting for uncertainty in reaction norm parameters and their causal effects on other measures, the risks accompanying classical approaches to BLUPs can be effectively avoided. I also introduce a new method for visualizing the consequences of multivariate selection on reaction norms. Simulations are then used to validate that the proposed models provide unbiased estimates across realistic parameter values, and an extensive coding tutorial is provided to aid researchers in applying this method to their own datasets in R.

Download Full-text

GMM ESTIMATION FOR DYNAMIC PANELS WITH FIXED EFFECTS AND STRONG INSTRUMENTS AT UNITY

Econometric Theory ◽

10.1017/s026646660909063x ◽

2009 ◽

Vol 26 (1) ◽

pp. 119-151 ◽

Cited By ~ 95

Author(s):

Chirok Han ◽

Peter C. B. Phillips

Keyword(s):

Time Series ◽

Fixed Effects ◽

Estimation Method ◽

Dynamic Panel Data ◽

Small Samples ◽

Moment Condition ◽

Dynamic Panel ◽

Gmm Estimation ◽

Unit Root Testing ◽

Dynamic Panels

This paper develops new estimation and inference procedures for dynamic panel data models with fixed effects and incidental trends. A simple consistent GMM estimation method is proposed that avoids the weak moment condition problem that is known to affect conventional GMM estimation when the autoregressive coefficient (ρ) is near unity. In both panel and time series cases, the estimator has standard Gaussian asymptotics for all values of ρ ∈ (−1, 1] irrespective of how the composite cross-section and time series sample sizes pass to infinity. Simulations reveal that the estimator has little bias even in very small samples. The approach is applied to panel unit root testing.

Download Full-text