Sources Of Bias Research Articles

Survival Benefit Associated With Participation in Clinical Trials of Anticancer Drugs: A Systematic Review and Meta-analysis.

Many cancer clinical investigators view clinical trials as offering better care for patients than routine clinical care. However, definitive evidence of clinical benefit from trial participation (hereafter referred to as the participation effect) has yet to emerge. To conduct a systematic review and meta-analysis of the evidence examining whether patient participation in cancer trials was associated with greater survival benefit compared with routine care. Studies were found through PubMed and Embase (January 1, 2000, until August 31, 2022), as well as backward and forward citation searching. Studies were included that compared overall survival of trial participants and routine care patients. Data extraction and methodological quality assessment were completed by 2 independent coders using Covidence software. Data were pooled using a random-effects model and analyzed based on the quality of the comparison between trial participants and routine care patients (ie, extent to which studies controlled for bias and confounders). The hazard ratio (HR) for overall survival of trial participants vs routine care patients. Thirty-nine publications were included, comprising 85 comparisons of trial participants and routine care patients. The meta-analysis revealed a statistically significant overall survival benefit for trial participants (HR, 0.76 [95% CI, 0.69-0.82]) when all studies were pooled, regardless of design or quality. However, survival benefits diminished in study subsets that matched trial participants and routine care patients for eligibility criteria (HR, 0.85 [95% CI, 0.75-0.97]) and disappeared when only high-quality studies were pooled (HR, 0.91 [95% CI, 0.80-1.05]). They also disappeared when estimates were adjusted for potential publication bias (HR, 0.94 [95% CI, 0.86-1.03]). Many studies suggest a survival benefit for cancer trial participants. However, these benefits were not detected in studies using designs addressing important sources of bias and confounding. Pooled results of high-quality studies are not consistent with a beneficial effect of trial participation on its own.

It's not all about residual confounding: a plea for QBA for epidemiologic researchers and educators.

We spend a great deal of time on confounding in our teaching, in our methods development and in our assessment of study results. This may give the impression that uncontrolled confounding is the biggest problem that observational epidemiology faces, when in fact, other sources of bias such as selection bias, measurement error, missing data, and misalignment of zero time may often (especially if they are all present in a single study) lead to a stronger deviation from the truth. Compared to the amount of time we spend teaching how to address confounding in a data analysis, we spend relatively little time teaching methods for simulating confounding (and other sources of bias) to learn their impact and develop plans to mitigate or quantify the bias. We review a paper by Desai et al that uses simulation methods to quantify the impact of an unmeasured confounder when it is completely missing or when a proxy of the confounder is measured. We use this article to discuss how we can use simulations of sources of bias to ensure we generate better and more valid study estimates, and we discuss the importance of simulating realistic datasets with plausible bias structures to guide data collection. If an advanced life form exists outside of our current universe and they came to earth with the goal of scouring the published epidemiologic literature to understand what the biggest problem epidemiologists have, they would quickly discover that the limitations section of publications would provide them with all the information they needed. And most likely what they would conclude is that the biggest problem that we face is uncontrolled confounding. It seems to be an obsession of ours.

Acupoint Sensitivity in Health and Disease: A Systematic Review.

Introduction: The concept of acupoints is a key defining feature of acupuncture, yet the scientific basis of acupoints remains unclear. In recent years, there has been an emerging body of animal studies demonstrating an association between cutaneous sensitivity and visceral pathophysiology, through which acupoints over the skin are sensitized in pathologic conditions. Several studies with humans have also been conducted to assess whether the sensitivity of acupoints is distinct in healthy versus clinical populations. However, no systematic review has been conducted to collate and synthesize the status and quality of human studies on this topic. Methods: A systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA). Literature search was performed by combining variations of search terms related to acupoints and pain sensitivity in PubMed, EMBASE, and Alt HealthWatch (EBSCOHost). Screening of titles and abstracts and review of full-text articles for eligibility were performed by two independent investigators. Using a predefined template, information on subject characteristics, pathologic conditions, names of assessed acupoints, and relevant main findings were extracted from the included studies. The methodological quality of included studies was assessed using a modified Newcastle-Ottawa Scale (NOS) for case-control studies. A quality assessment checklist was also developed by the present authors to examine the quality of reporting of experimental variables that were considered important for evaluating acupoint sensitivity. Results: A total of 3453 studies were identified from the database search, of which 11 met the eligibility criteria to be included in this review. Six studies examined the mechanical sensitivity of body acupoints, and the remaining five studies examined the mechanical sensitivity of auricular points. Overall, findings suggest that the sensitivity of acupoints may be distinct in healthy versus clinical populations. However, there were various potential sources of bias and substantial heterogeneity across included studies in clinical conditions and acupoints. Conclusion: There is at present insufficient evidence to support or refute that acupoints in humans are sensitized in pathologic conditions. There were various methodological issues, including small sample size and poor reporting of experimental design and variables, which limit the ability to draw a definitive conclusion on this topic. It is also largely unclear whether it is the general body regions rather than specific acupoints that may be sensitized, as most studies did not include nonacupoint location(s) for comparison. Thus, further rigorous research is warranted.

Integrated and DC-powered superconducting microcomb

Frequency combs, specialized laser sources emitting multiple equidistant frequency lines, have revolutionized science and technology with unprecedented precision and versatility. Recently, integrated frequency combs are emerging as scalable solutions for on-chip photonics. Here, we demonstrate a fully integrated superconducting microcomb that is easy to manufacture, simple to operate, and consumes ultra-low power. Our turnkey apparatus comprises a basic nonlinear superconducting device, a Josephson junction, directly coupled to a superconducting microstrip resonator. We showcase coherent comb generation through self-started mode-locking. Therefore, comb emission is initiated solely by activating a DC bias source, with power consumption as low as tens of picowatts. The resulting comb spectrum resides in the microwave domain and spans multiple octaves. The linewidths of all comb lines can be narrowed down to 1 Hz through a unique coherent injection-locking technique. Our work represents a critical step towards fully integrated microwave photonics and offers the potential for integrated quantum processors.

Increasing the etanercept dose in a treat-to-target approach in juvenile idiopathic arthritis: does it help to reach the target? A post-hoc analysis of the BeSt for Kids randomised clinical trial

BackgroundEtanercept has been studied in doses up to 0.8 mg/kg/week (max 50 mg/week) in juvenile idiopathic arthritis (JIA) patients. In clinical practice higher doses are used off-label, but evidence regarding the relation with outcomes is lacking. We describe the clinical course of JIA-patients receiving high-dose etanercept (1.6 mg/kg/week; max 50 mg/week) in the BeSt for Kids trial.Methods92 patients with oligoarticular JIA, RF-negative polyarticular JIA or juvenile psoriatic arthritis were randomised across three treat-to-target arms: (1) sequential DMARD-monotherapy (sulfasalazine or methotrexate (MTX)), (2) combination-therapy MTX + 6 weeks prednisolone and (3) combination therapy MTX + etanercept. In any treatment-arm, patients could eventually escalate to high-dose etanercept alongside MTX 10mg/m2/week.Results32 patients received high-dose etanercept (69% female, median age 6 years (IQR 4–10), median 10 months (7–16) from baseline). Median follow-up was 24.6 months. Most clinical parameters improved within 3 months after dose-increase: median JADAS10 from 7.2 to 2.8 (p = 0.008), VAS-physician from 12 to 4 (p = 0.022), VAS-patient/parent from 38.5 to 13 (p = 0.003), number of active joints from 2 to 0.5 (p = 0.12) and VAS-pain from 35.5 to 15 (p = 0.030). Functional impairments (CHAQ-score) improved more gradually and ESR remained stable. A comparable pattern was observed in 11 patients (73% girls, median age 8 (IQR 6–9)) who did not receive high-dose etanercept despite eligibility (comparison group). In both groups, 56% reached inactive disease at 6 months. No severe adverse events (SAEs) occurred after etanercept dose-increase. In the comparison group, 2 SAEs consisting of hospital admission occurred. Rates of non-severe AEs per subsequent patient year follow-up were 2.27 in the high-dose and 1.43 in the comparison group.ConclusionsEscalation to high-dose etanercept in JIA-patients who were treated to target was generally followed by meaningful clinical improvement. However, similar improvements were observed in a smaller comparison group who did not escalate to high-dose etanercept. No SAEs were seen after escalation to high-dose etanercept. The division into the high-dose and comparison groups was not randomised, which is a potential source of bias. We advocate larger, randomised studies of high versus regular dose etanercept to provide high level evidence on efficacy and safety.Trial registrationDutch Trial Register; NTR1574; 3 December 2008; https://onderzoekmetmensen.nl/en/trial/26585.

Reward Sensitivity and Noise Contribute to Negative Affective Bias: A Learning Signal Detection Theory Approach in Decision-Making

In patients with mood disorders, negative affective biases – systematically prioritising and interpreting information negatively – are common. A translational cognitive task testing this bias has shown that depressed patients have a reduced preference for a high reward under ambiguous decision-making conditions. The precise mechanisms underscoring this bias are, however, not yet understood. We therefore developed a set of measures to probe the underlying source of the behavioural bias by testing its relationship to a participant’s reward sensitivity, value sensitivity and reward learning rate. One-hundred-forty-eight participants completed three online behavioural tasks: the original ambiguous-cue decision-making task probing negative affective bias, a probabilistic reward learning task probing reward sensitivity and reward learning rate, and a gambling task probing value sensitivity. We modelled the learning task through a dynamic signal detection theory model and the gambling task through an expectation-maximisation prospect theory model. Reward sensitivity from the probabilistic reward task (β = 0.131, p = 0.024) and setting noise from the probabilistic reward task (β = –0.187, p = 0.028) both predicted the affective bias score in a logistic regression. Increased negative affective bias, at least on this specific task, may therefore be driven in part by a combination of reduced sensitivity to rewards and more variable responses.

Deep learning models for atypical serotonergic cells recognition

Background:The serotonergic system modulates brain processes via functionally distinct subpopulations of neurons with heterogeneous properties, including their electrophysiological activity. In extracellular recordings, serotonergic neurons to be investigated for their functional properties are commonly identified on the basis of “typical” features of their activity, i.e. slow regular firing and relatively long duration of action potentials. Thus, due to the lack of equally robust criteria for discriminating serotonergic neurons with “atypical” features from non-serotonergic cells, the physiological relevance of the diversity of serotonergic neuron activities results largely understudied. New Methods: We propose deep learning models capable of discriminating typical and atypical serotonergic neurons from non-serotonergic cells with high accuracy. The research utilized electrophysiological in vitro recordings from serotonergic neurons identified by the expression of fluorescent proteins specific to the serotonergic system and non-serotonergic cells. These recordings formed the basis of the training, validation, and testing data for the deep learning models. The study employed convolutional neural networks (CNNs), known for their efficiency in pattern recognition, to classify neurons based on the specific characteristics of their action potentials. Results:The models were trained on a dataset comprising 27,108 original action potential samples, alongside an extensive set of 12 million synthetic action potential samples, designed to mitigate the risk of overfitting the background noise in the recordings, a potential source of bias. Results show that the models achieved high accuracy and were further validated on ”non-homogeneous” data, i.e., data unknown to the model and collected on different days from those used for the training of the model, to confirm their robustness and reliability in real-world experimental conditions. Comparison with existing methods: Conventional methods for identifying serotonergic neurons allow recognition of serotonergic neurons defined as typical. Our model based on the analysis of the sole action potential reliably recognizes over 94% of serotonergic neurons including those with atypical features of spike and activity. Conclusion:The model is ready for use in experiments conducted with the here described recording parameters. We release the codes and procedures allowing to adapt the model to different acquisition parameters or for identification of other classes of spontaneously active neurons.

A framework for understanding selection bias in real-world healthcare data

Abstract Using administrative patient-care data such as Electronic Health Records (EHR) and medical/pharmaceutical claims for population-based scientific research has become increasingly common. With vast sample sizes leading to very small standard errors, researchers need to pay more attention to potential biases in the estimates of association parameters of interest, specifically to biases that do not diminish with increasing sample size. Of these multiple sources of biases, in this paper, we focus on understanding selection bias. We present an analytic framework using directed acyclic graphs for guiding applied researchers to dissect how different sources of selection bias may affect estimates of the association between a binary outcome and an exposure (continuous or categorical) of interest. We consider four easy-to-implement weighting approaches to reduce selection bias with accompanying variance formulae. We demonstrate through a simulation study when they can rescue us in practice with analysis of real-world data. We compare these methods using a data example where our goal is to estimate the well-known association of cancer and biological sex, using EHR from a longitudinal biorepository at the University of Michigan Healthcare system. We provide annotated R codes to implement these weighted methods with associated inference.

Understanding and Mitigating Bias in Imaging Artificial Intelligence.

Artificial intelligence (AI) algorithms are prone to bias at multiple stages of model development, with potential for exacerbating health disparities. However, bias in imaging AI is a complex topic that encompasses multiple coexisting definitions. Bias may refer to unequal preference to a person or group owing to preexisting attitudes or beliefs, either intentional or unintentional. However, cognitive bias refers to systematic deviation from objective judgment due to reliance on heuristics, and statistical bias refers to differences between true and expected values, commonly manifesting as systematic error in model prediction (ie, a model with output unrepresentative of real-world conditions). Clinical decisions informed by biased models may lead to patient harm due to action on inaccurate AI results or exacerbate health inequities due to differing performance among patient populations. However, while inequitable bias can harm patients in this context, a mindful approach leveraging equitable bias can address underrepresentation of minority groups or rare diseases. Radiologists should also be aware of bias after AI deployment such as automation bias, or a tendency to agree with automated decisions despite contrary evidence. Understanding common sources of imaging AI bias and the consequences of using biased models can guide preventive measures to mitigate its impact. Accordingly, the authors focus on sources of bias at stages along the imaging machine learning life cycle, attempting to simplify potentially intimidating technical terminology for general radiologists using AI tools in practice or collaborating with data scientists and engineers for AI tool development. The authors review definitions of bias in AI, describe common sources of bias, and present recommendations to guide quality control measures to mitigate the impact of bias in imaging AI. Understanding the terms featured in this article will enable a proactive approach to identifying and mitigating bias in imaging AI. Published under a CC BY 4.0 license. Test Your Knowledge questions for this article are available in the supplemental material. See the invited commentary by Rouzrokh and Erickson in this issue.

Machine Hands on Flaws to Machine: The Surprising Sources of Biases in Machine Learning Models

Machine Hands on Flaws to Machine: The Surprising Sources of Biases in Machine Learning Models

Foreground removal with ILC methods for AliCPT-1

One of the main goals of most future CMB experiments is the precise measurement of CMB B-mode polarization, whose major obstacle is the Galactic foregrounds. In this paper, we evaluate the foreground cleaning performance of the variants of the ILC method on partial sky B-modes and analyze the main sources of biases on the BB power spectrum. Specially, we compare the NILC, the cILC (in three domains) and the cMILC methods for AliCPT-1 simulations. We find that the cILC methods implemented in harmonic space and needlet space are both competent to clean different models of foregrounds, which bias the tensor-to-scalar ratio about 0.008 at maximum, and constrain the tensor-to-scalar ratio to r < 0.043(95%CL) for the AliCPT-1 configuration. We also note that the deviation of the estimated noise bias from the actual one for ILC, dubbed the noise bias error (NBE) in this paper, might make significant effects on the power spectrum for a small footprint and low signal-to-noise ratio CMB experiment. We finally obtain its relation with respect to the noise residual which fits well with the simulated results.

Ambient Seismic Noise Tomography Within the Floridan Aquifer System, Santa Fe River‐Sink Rise, Florida, US

Ambient Seismic Noise Tomography Within the Floridan Aquifer System, Santa Fe River‐Sink Rise, Florida, US

A simple method to construct circular repeated measurement design classes and efficiently control carry over effects

Repeated measurement designs (RMDs) are economical and commonly utilized in pharmacology, psychology, and animal science experiments. However, RMDs might have carryover effects, which become the main source of bias in estimation of treatment effects. Carry over effects can economically be controlled with the use of minimal balanced RMDs, strongly balanced RMDs or their generalized classes. No general procedure is available in literature to obtain these important classes of RMDs. An R-Package has been developed in this article to check the given values v and p for the possible existence of minimal circular balanced and strongly balanced RMDs (along with their generalized classes), generate the sets of shifts for possible designs, and to calculate the efficiency of separability and carry over effects for the generated design.

Effects of avoidance versus use of neuromuscular blocking agents for facilitation of tracheal intubation in children and infants.

The European Society of Anesthesiology and Intensive Care recommends the use of neuromuscular blocking agents (NMBA) in adults, to facilitate tracheal intubation and reduce its associated complications. Children who undergo tracheal intubation may suffer some of the same complications, however, no consensus exists regarding the use of NMBA for tracheal intubation in the pediatric population. We will explore the existing evidence assessing the effects of avoidance versus the use of NMBA for the facilitation of tracheal intubation in children and infants. This protocol follows the preferred reporting items for systematic reviews and meta-analyses protocols recommendations. We will include all randomized controlled clinical trials assessing the effects of avoidance versus the use of NMBA for facilitation of tracheal intubation (oral or nasal) using direct laryngoscopy or video laryngoscopy in pediatric participants (<18 years). Our primary outcome is incidence of difficult tracheal intubation. Secondary outcomes include incidence of serious adverse events, failed intubation, events of upper airway discomfort or injury, and difficult laryngoscopy. We will conduct a thorough database search to identify relevant trials, including CENTRAL, MEDLINE, EMBASE, BIOSIS, Web of Science, CINAHL, and trial registries. Two review authors will independently handle the screening of literature and data extraction. Each trial will be evaluated for major sources of bias with the "classic risk of bias tool" used in the Cochrane Collaboration tool from 2011. We will use Review manager (RevMan) or R with the meta package to perform the meta-analysis. We will perform a trial sequential analysis on the meta-analysis of our primary outcome, providing an estimate of statistical reliability. Two review authors will independently assess the quality of the body of evidence using the grading of recommendations assessment, development, and evaluation (GRADE) approach. We will use GRADEpro software to conduct the GRADE assessments and to create "Summary of the findings" tables.

Use of quantitative bias analysis to evaluate single-arm trials with real-world data external controls.

Use of real-world data (RWD) for external controls added to single-arm trials (SAT) is increasingly prevalent in regulatory submissions. Due to inherent differences in the data-generating mechanisms, biases can arise. This paper aims to illustrate how to use quantitative bias analysis (QBA). Advanced non-small cell lung cancer (NSCLC) serves as an example, where many small subsets of patients with molecular tumor subtypes exist. First, some sources of bias that may occur in oncology when comparing RWD to SAT are described. Second, using a hypothetical immunotherapy agent, a dataset is simulated based on expert input for survival analysis of advanced NSCLC. Finally, we illustrate the impact of three biases: missing confounder, misclassification of exposure, and outcome evaluation. For each simulated scenario, bias was induced by removing or adding data; hazard ratios (HRs) were estimated applying conventional analyses. Estimating the bias-adjusted treatment effect and uncertainty required carefully selecting the bias model and bias factors. Although the magnitude of each biased and bias-adjusted HR appeared moderate in all three hypothetical scenarios, the direction of bias was variable. These findings suggest that QBA can provide an intuitive framework for bias analysis, providing a key means of challenging assumptions about the evidence. However, the accuracy of bias analysis is itself dependent on correct specification of the bias model and bias factors. Ultimately, study design should reduce bias, but QBA allows us to evaluate the impact of unavoidable bias to assess the quality of the evidence.

Assessing measurement uncertainties on high-speed DCB testing of composites

This study investigates dynamic behavior in Double Cantilever Beam (DCB) tests across loading rates from 6 mm/min to 3 m/s. Comprehensive instrumentation has been used to investigate the sources of measurement errors (load, opening and crack length) or biases (asymmetry) and to quantify their effects on the fracture toughness determination using three variations of the Corrected Beam Theory. In particular, challenges in establishing zero load and displacement references were revealed through monitoring initial test phases. The resulting uncertainties in the fracture toughness have been estimated to be between 7 and 15%. Arms asymmetry during higher loading rates (1 m/s and 3 m/s) has been recorded using Digital Images Correlation. Using this data and a simple beam model, mode mixity (GII/GI from Williams approach) computed is in the order of 10%–20%.

Combined multiplex panel test results are a poor estimate of disease prevalence without adjustment for test error.

Multiplex panel tests identify many individual pathogens at once, using a set of component tests. In some panels the number of components can be large. If the panel is detecting causative pathogens for a single syndrome or disease then we might estimate the burden of that disease by combining the results of the panel, for example determining the prevalence of pneumococcal pneumonia as caused by many individual pneumococcal serotypes. When we are dealing with multiplex test panels with many components, test error in the individual components of a panel, even when present at very low levels, can cause significant overall error. Uncertainty in the sensitivity and specificity of the individual tests, and statistical fluctuations in the numbers of false positives and false negatives, will cause large uncertainty in the combined estimates of disease prevalence. In many cases this can be a source of significant bias. In this paper we develop a mathematical framework to characterise this issue, we determine expressions for the sensitivity and specificity of panel tests. In this we identify a counter-intuitive relationship between panel test sensitivity and disease prevalence that means panel tests become more sensitive as prevalence increases. We present novel statistical methods that adjust for bias and quantify uncertainty in prevalence estimates from panel tests, and use simulations to test these methods. As multiplex testing becomes more commonly used for screening in routine clinical practice, accumulation of test error due to the combination of large numbers of test results needs to be identified and corrected for.

Comprehensive High-resolution Chemical Spectroscopy of Barnard’s Star with SPIRou

Determination of fundamental parameters of stars impacts all fields of astrophysics, from galaxy evolution to constraining the internal structure of exoplanets. This paper presents a detailed spectroscopic analysis of Barnard’s star (otherwise known as Gl 699) that compares an exceptionally high-quality (an average signal-to-noise ratio of ∼1000 in the entire domain), high-resolution near-infrared (NIR) spectrum taken with Canada-France-Hawaii Telescope/SPIRou to PHOENIX-ACES stellar atmosphere models. The observed spectrum shows thousands of lines not identified in the models with a similarly large number of lines present in the model but not in the observed data. We also identify several other caveats, such as continuum mismatch, unresolved contamination, and spectral lines significantly shifted from their expected wavelengths; all of these can be a source of bias for the determination of abundance. Out of >104 observed lines in the NIR that could be used for chemical spectroscopy, we identify a short list of a few hundred lines that are reliable. We present a novel method for determining the effective temperature (T eff) and overall metallicity of slowly rotating M dwarfs that uses several groups of lines as opposed to bulk spectral fitting methods. With this method, we infer T eff = 3231 ± 21 K for Barnard's star, consistent with the value of 3238 ± 11 K inferred from the interferometric method. We also provide measurements of the abundance of 15 different elements for Barnard's star, including the abundances of four elements (K, O, Y, Th) never reported before for this star. This work emphasizes the need to improve current atmosphere models to fully exploit the NIR domain for chemical spectroscopy analysis.

Gentrification and Racial Distrust in Communities: Evidence from 911 Calls

The prevalence of racial bias in policing has long concerned social scientists and policymakers. This article studies a predecessor mechanism that constitutes an important source of policing bias in American society: calls by individuals to the police to investigate “suspicious” behaviors, often involving neighbors. We construct a novel data set of more than 39 million 911 calls across 14 U.S. cities from 2011 to 2020. These data, obtained through the digitization initiatives of local governments, provide us with a unique opportunity to study neighborhood-level trust and social cohesion and demonstrate how changes to a neighborhood’s composition lead to systematic increases in the prevalence of “unfounded” suspicion calls to the police. Across a range of specifications, the proportion of unfounded suspicion calls increases as more non-Black residents move into neighborhoods with historically high levels of Black residents. This relationship is exacerbated in gentrifying neighborhoods and those with public spaces that enable more contact between community members. However, we also find some evidence that Black leadership and public support of Black citizens in communities mitigate the association between non-Black residents and the proportion of unfounded 911 calls. We discuss our results and implications for future research and policy. This paper was accepted by Anindya Ghose, information systems. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.00362 .

Investigating item complexity as a source of cross-national DIF in TIMSS math and science

BackgroundLarge scale international assessments depend on invariance of measurement across countries. An important consideration when observing cross-national differential item functioning (DIF) is whether the DIF actually reflects a source of bias, or might instead be a methodological artifact reflecting item response theory (IRT) model misspecification. Determining the validity of the source of DIF has implications for how it is handled in practice.MethodWe demonstrate a form of sensitivity analysis that can point to model misspecification induced by item complexity as a possible cause of DIF, and show how such a cause of DIF might be accommodated through attempts to generalize the IRT model for the studied item(s) in psychometrically and psychologically plausible ways.ResultsIn both simulated illustrations and empirical data from TIMSS 2011 and TIMSS 2019 4th and 8th Grade Math and Science, we have found that using a form of proposed IRT model generalization can substantially reduce DIF when IRT model misspecification is at least a partial cause of the observed DIF.ConclusionsBy demonstrating item complexity as a possible valid source of DIF and showing the effectiveness of the proposed approach, we recommend additional attention toward model generalizations as a means of addressing and/or understanding DIF.