MODEL SELECTION AND INFERENCE: FACTS AND FICTION

By H. Leeb, B. M. Potscher. Econometric Theory, 2005

712 citations recorded by the source corpus.

Subfields: Control and Systems Engineering, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/307d4f561831d98246c91424b84ff90f81b964eb

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Control and Systems Engineering

In Statistics and Probability

Citations in the corpus

A re‐evaluation of fixed effect(s) meta‐analysis (2018, background): “In a single analysis such procedures may bias effect estimates and seriously underestimate standard errors (Leeb and Pötscher, 2005); they can similarly invalidate inference from any meta-analysis.”
Valid confidence intervals for post-model-selection predictors (2014, background): “…actual coverage probability of ‘naive’ conﬁdence sets can be dramatically smaller than the nominal one), and valid procedures are often hard to ﬁnd; see Leeb and P¨otscher (2005, 2006, 2012), Kabaila and Leeb (2006), P¨otscher (2009) and references therein for an introduction to the issues…”
Pre-Event Trends in the Panel Event-Study Design (2019, neutral): “However, while such tests are valid tests of a null of correct specification, choosing a model based on such tests will create the usual pre-test biases (Leeb and Pötscher 2005; Guggenberger 2010).”
Exact post-selection inference, with application to the lasso (2013, background): “Leeb and Pötscher (2005, 2006) obtained a number of negative results about estimating the distribution of a post-selection estimator, although they note their results do not necessarily preclude the possibility of postselection inference. The problem was most recently considered by Berk et al. (2013”
On the Large-Sample Minimal Coverage Probability of Confidence Intervals After Model Selection (2006, background): “Unfortunately, at closer inspection this conclusion turns out to not be warranted, and relying on it only creates an illusion of conducting valid inference (for details, see Leeb and Pötscher 2004, 2005).”
Penalized regression, standard errors, and Bayesian lassos (2010, result): “However, the results of Leeb and Pötscher (2005) also suggest that the distributional properties of post-model-selection estimators are quite intricate, and are not properly captured by the usual pointwise large-sample analysis using the maximal scaled mean squared error (MSE) and the maximal…”
Variable Selection in Predictive Regressions (2013, methodology): “Leeb and Potscher (2005, 2008) show that the distributions of estimators depend on the outcome of model-selection and cannot be uniformly estimated.”
Uniformly valid confidence intervals post-model-selection (2016, background): “…of valid statistical procedures in such situations is quite challenging, even if the candidate set of models does contain the true distribution (cf. Leeb and Pötscher (2005, 2006, 2008), Kabaila and Leeb (2006) and Pötscher (2009), and the references given in that literature), and has recently…”
Confidence intervals for high-dimensional inverse covariance estimation (2014, methodology): “We mention that several concerns have been raised considering simple post-model selection methods, which are elaborated on in the papers [18, 19] or [3].”
THE IMPACT OF A HAUSMAN PRETEST ON THE ASYMPTOTIC SIZE OF A HYPOTHESIS TEST (2009, background): “Andrews and Guggenberger (2009a) and Leeb and Pötscher (2005) consider tests concerning a parameter in a linear regression model after a “conservative” model selection procedure has been applied and find extreme size distortion for the two-stage test.”
A Simple Parametric Model Selection Test (2014, methodology): “In finite samples, some effect of the model selection step cannot be completely excluded (see, e.g., White 2000; Leeb and Pötscher 2005, 2008, and references therein, for a more detailed discussion).”
Valid post-selection inference (2013, background): “…specific problems have been the subject of incisive analyses and criticisms by the “Vienna School” of Pötscher, Leeb and Schneider; see, for example, Leeb and Pötscher (2003, 2005, 2006a, 2006b, 2008a, 2008b, 2008c), Pötscher (2006), Leeb (2006), Pötscher and Leeb (2009), Pötscher and…”
LONG MEMORY TESTING IN THE TIME DOMAIN (2007, background): “1) and (24) leads without any considerable modification to ∥∥∥∥∥ 1 T T ∑”
The Coverage Properties of Confidence Regions After Model Selection (2009, background): “Leeb & Pötscher (2005) also make the point that consistent model selection procedures, and their like, do not lead to naive confidence regions that, for sufficiently large sample size, have practically the same properties as if the true model were known.”