Statistical Models and Explanation of Inferential Error
DOI:
https://doi.org/10.35588/cc.v3i2.5816Keywords:
Statistical models, Error, Inferential procedures, Causal explanation, Counterfactual dependency patternsAbstract
The main thesis of this work is that statistical models provide explanations about the errors that our data inference procedures from data to phenomena can make, in the context of classical or frequentist statistics. The explanation of the error is an operation which in turn, helps to avoid it where it is avoidable and intolerable; tolerate it where it is unavoidable and tolerable; and suspend judgment where it is unavoidable and intolerable. All these operations are elements of a reliable evidence-generating practice. This thesis is illustrated with the linear model in econometrics, the ordinary least squares estimator, the estimator and model of instrumental variables, and with the concepts related to hypothesis testing. Special emphasis is placed on the properties of unbiasedness, consistency, and precision as characterizations of the type of error that can be made using an estimator. It is argued that models explain by identifying patterns of counterfactual dependence between features of the stochastic mechanism represented by the model and features of the inferential procedures.
Downloads
Downloads
Published
Issue
Section
License
Regarding the copyright of the contributions (i.e. articles or others) published or sent to Culturas Científicas, it will be understood that the authors accept the following terms at the time of submitting:
-
The authors will retain the rights of the contributions they submit to Culturas Científicas.
-
The authors accept that their contribution is subject to the Creative Commons Attribution 4.0 International License (CC BY 4.0) once it is published.
This means that the published contribution may be shared and adapted—even commercially—as long as proper credit is given and no endorsement by the author (i.e. licensor) is implied.
-
Notwithstanding the foregoing, at the time of accepting their contribution, the authors grant its first publication exclusively to Culturas Científicas.
-
Once the contribution is accepted, the authors grant to Culturas Científicas the rights of reproduction and distribution of their article in all countries of the world for their exploitation in any medium, format or support.
-
The authors may adopt other non-exclusive license agreements for the contribution published by our journal (e.g. deposit in archives, publication in other journals, book compendiums or translation) committing to indicate that the original publication was made in Culturas Científicas or publishing the version that already has the identifiers of our journal.
We encourage the authors to spread their articles through the internet in sites designated for it in order to motivate their reading and discussion.
