With the proliferation of misinformation on the web, automatic misinformation detection methods are becoming an increasingly important subject of study. Large language models have produced the best results among content-based methods, which rely on the text of the article rather than the metadata or network features. However, finetuning such a model requires significant training data, which has led to the automatic creation of large-scale misinformation detection datasets. In these datasets, articles are not labelled directly. Rather, each news site is labelled for reliability by an established fact-checking organisation and every article is subsequently assigned the corresponding label based on the reliability score of the news source in question. A recent paper has explored the biases present in one such dataset, NELA-GT-2018, and shown that the models are at least partly learning the stylistic and other features of different news sources rather than the features of unreliable news. We confirm a part of their findings. Apart from studying the characteristics and potential biases of the datasets, we also find it important to examine in what way the model architecture influences the results. We therefore explore which text features or combinations of features are learned by models based on contextual word embeddings as opposed to basic bag-of-words models. To elucidate this, we perform extensive error analysis aided by the SHAP post-hoc explanation technique on a debiased portion of the dataset. We validate the explanation technique on our inherently interpretable baseline model.
Previous research shows that automatic tendency to approach alcohol plays a causal role in problematic alcohol use and can be retrained by Approach Bias Modification (ApBM). ApBM has been shown to be effective for patients diagnosed with alcohol use disorder (AUD) in inpatient treatment. This study aimed to investigate the effectiveness of adding an online ApBM to treatment as usual (TAU) in an outpatient setting compared to receiving TAU with an online placebo training. 139 AUD patients receiving face-to-face or online treatment as usual (TAU) participated in the study. The patients were randomized to an active or placebo version of 8 sessions of online ApBM over a 5-week period. The weekly consumed standard units of alcohol (primary outcome) was measured at pre-and post-training, 3 and 6 months follow-up. Approach tendency was measured pre-and-post ApBM training. No additional effect of ApBM was found on alcohol intake, nor other outcomes such as craving, depression, anxiety, or stress. A significant reduction of the alcohol approach bias was found. This research showed that approach bias retraining in AUD patients in an outpatient treatment setting reduces the tendency to approach alcohol, but this training effect does not translate into a significant difference in alcohol reduction between groups. Explanations for the lack of effects of ApBM on alcohol consumption are treatment goal and severity of AUD. Future ApBM research should target outpatients with an abstinence goal and offer alternative, more user-friendly modes of delivering ApBM training.
MULTIFILE
BackgroundConfounding bias is a common concern in epidemiological research. Its presence is often determined by comparing exposure effects between univariable- and multivariable regression models, using an arbitrary threshold of a 10% difference to indicate confounding bias. However, many clinical researchers are not aware that the use of this change-in-estimate criterion may lead to wrong conclusions when applied to logistic regression coefficients. This is due to a statistical phenomenon called noncollapsibility, which manifests itself in logistic regression models. This paper aims to clarify the role of noncollapsibility in logistic regression and to provide guidance in determining the presence of confounding bias.MethodsA Monte Carlo simulation study was designed to uncover patterns of confounding bias and noncollapsibility effects in logistic regression. An empirical data example was used to illustrate the inability of the change-in-estimate criterion to distinguish confounding bias from noncollapsibility effects.ResultsThe simulation study showed that, depending on the sign and magnitude of the confounding bias and the noncollapsibility effect, the difference between the effect estimates from univariable- and multivariable regression models may underestimate or overestimate the magnitude of the confounding bias. Because of the noncollapsibility effect, multivariable regression analysis and inverse probability weighting provided different but valid estimates of the confounder-adjusted exposure effect. In our data example, confounding bias was underestimated by the change in estimate due to the presence of a noncollapsibility effect.ConclusionIn logistic regression, the difference between the univariable- and multivariable effect estimate might not only reflect confounding bias but also a noncollapsibility effect. Ideally, the set of confounders is determined at the study design phase and based on subject matter knowledge. To quantify confounding bias, one could compare the unadjusted exposure effect estimate and the estimate from an inverse probability weighted model.
MULTIFILE