Note on the sampling error of the difference between correlated proportions or percentages

By Q. Mcnemar. 1947

How researchers see this paper

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

In Statistics and Probability

Citations in the corpus

Do Not Have Enough Data? Deep Learning to the Rescue! (2019, background): “LSTM Long Short Term Memory represents the type of classiﬁers that emerged after the advances in training recurrent neural networks, and the introduction of word embed-dings (Mikolov et al. 2013), LSTMs are commonly used for sequential and textual data classiﬁcation.”
Are We Modeling the Task or the Annotator? An Investigation of Annotator Bias in Natural Language Understanding Datasets (2019, methodology): “(2018), p-values were calculated using the McNemar’s test (McNemar, 1947) for MNLI and the Bootstrap test (Berg-Kirkpatrick et al.”
Leverage Financial News to Predict Stock Price Movements Using Word Embeddings and Deep Neural Networks (2015, methodology): “In order to test the significance of our results, we applied the McNemar’s Test [44], which is a statistical test used on paired nominal data.”
ECONET: Effective Continual Pretraining of Language Models for Event Temporal Reasoning (2020, background): “Table 8 in the appendix lists all improvements’ p-values per McNemar’s test (McNemar, 1947).”
Cyberbullying Detection With Fairness Constraints (2020, methodology): “Furthermore, we perform McNemar’s test (also known as withinsubjects chi-squared test) [64] to test whether a statistically significant difference between baseline (unconstrained) and constrained models exists in terms of cyberbullying classification accuracy.”
The Hitchhiker’s Guide to Testing Statistical Significance in Natural Language Processing (2018, methodology): “McNemar’s test (McNemar, 1947) This test is designed for paired nominal observations (binary labels).”
Some statistical issues in the comparison of speech recognition algorithms (1989, result): “McNemar ' s test is applicable when the errors made by an algorithm are independent , a condition which is cer - tainly met in an isolated word algorithm which does not make use of any context .”
Marginal models for categorical data (2002, methodology): “This test is best known as McNemar’s test (McNemar, 1947).”
Focused Entailment Graphs for Open IE Propositions (2014, methodology): “The F1 difference between these methods is statistically significant with McNemar’s test (1947) with p 0.”
High-Performance Work Systems and Job Control (2013, background): “The sampling error is within recommended limits (McNemar, 1947), suggesting that response bias was not a concern.”
Detecting Asymmetric Semantic Relations in Context: A Case-Study on Hypernymy Detection (2017, methodology): “01 under the McNemar’s test (McNemar, 1947) Supervised Model Config.”
Retrospective Reader for Machine Reading Comprehension (2020, methodology): “According to our task setting, we used McNemar’s test [McNemar, 1947] to test the statistical significance of our results.”