Differential Measurement Error Research Articles

Differential Measurement Error in House Price Indices

Differential Measurement Error in House Price Indices

A simulation study analyzing the impact of differential exposure measurement error of air pollution on preterm birth

A simulation study analyzing the impact of differential exposure measurement error of air pollution on preterm birth

A Finite-sample bias correction method for general linear model in the presence of differential measurement errors

A Finite-sample bias correction method for general linear model in the presence of differential measurement errors

Psychometric assessment of the Runyankole-translated Marlowe-Crowne Social Desirability Scale among persons with HIV in Uganda

BackgroundSocial desirability can negatively affect the validity of self-reported measures, including underreporting of stigmatized behaviors like alcohol consumption. The Marlowe-Crowne Social Desirability Scale (SDS) is widely implemented and comprised of Denial and Attribution Domains (i.e., tendencies to deny undesirable traits or attribute socially desirable traits to oneself, respectively). Yet, limited psychometric research has been conducted in sub-Saharan Africa, where the prevalence of unhealthy alcohol consumption is high as well as religiosity and hierarchical social norms. To address this gap, we (a) conducted an exploratory study assessing certain psychometric properties of the 28-item SDS (Runyankole-translated) among persons with HIV (PWH) in Uganda, and (b) examined the relationship between social desirability and self-reported alcohol use.MethodsWe pooled baseline data (N = 1153) from three studies of PWH engaged in alcohol use from 2017 to 2021. We assessed the translated scale’s construct validity (via confirmatory factor analysis), internal consistency, item performance, differential item functioning by gender, concurrent validity with the DUREL religiosity index domains, and the association between social desirability and self-reported alcohol use.ResultsParticipants had a mean age of 40.42 years, 63% were men, and 91% had an undetectable HIV viral load. The 28-item SDS had satisfactory construct validity (Model fit indices: RMSEA = 0.07, CFI = 0.84, TLI = 0.82) and internal consistency (Denial Domain ΩTotal = 0.82, Attribution Domain ΩTotal = 0.69). We excluded Item 14 (“I never hesitate to help someone in trouble”) from the Attribution Domain, which mitigated differential measurement error by gender and slightly improved the construct validity (Model fit indices: RMSEA = 0.06, CFI = 0.86, TLI = 0.85) and reliability (Attribution Domain ΩTotal = 0.72) of the 27-item modified SDS. Using the 27-item SDS, we found that social desirability was weakly correlated with religiosity and inversely associated with self-reported alcohol use after adjusting for biomarker-measured alcohol use and other confounders (β = -0.05, 95% confidence interval: -0.09 to -0.01, p-value = 0.03).ConclusionsWe detected and mitigated measurement error in the 28-item Runyankole-translated SDS, and found that the modified 27-item scale had satisfactory construct validity and internal consistency in our sample. Future studies should continue to evaluate the psychometric properties of the Runyankole-translated SDS, including retranslating Item 14 and reevaluating its performance.

Higher Order Moments for Differential Measurement Error, with Application to Tobin's q and Corporate Investment

Higher Order Moments for Differential Measurement Error, with Application to Tobin's q and Corporate Investment

Vaginal Microbiome Metagenome Inference Accuracy: Differential Measurement Error according to Community Composition.

Several studies have compared metagenome inference performance in different human body sites; however, none specifically reported on the vaginal microbiome. Findings from other body sites cannot easily be generalized to the vaginal microbiome due to unique features of vaginal microbial ecology, and investigators seeking to use metagenome inference in vaginal microbiome research are "flying blind" with respect to potential bias these methods may introduce into analyses. We compared the performance of PICRUSt2 and Tax4Fun2 using paired 16S rRNA gene amplicon sequencing and whole-metagenome sequencing data from vaginal samples from 72 pregnant individuals enrolled in the Pregnancy, Infection, and Nutrition (PIN) cohort. Participants were selected from those with known birth outcomes and adequate 16S rRNA gene amplicon sequencing data in a case-control design. Cases experienced early preterm birth (<32 weeks of gestation), and controls experienced term birth (37 to 41 weeks of gestation). PICRUSt2 and Tax4Fun2 performed modestly overall (median Spearman correlation coefficients between observed and predicted KEGG ortholog [KO] relative abundances of 0.20 and 0.22, respectively). Both methods performed best among Lactobacillus crispatus-dominated vaginal microbiotas (median Spearman correlation coefficients of 0.24 and 0.25, respectively) and worst among Lactobacillus iners-dominated microbiotas (median Spearman correlation coefficients of 0.06 and 0.11, respectively). The same pattern was observed when evaluating correlations between univariable hypothesis test P values generated with observed and predicted metagenome data. Differential metagenome inference performance across vaginal microbiota community types can be considered differential measurement error, which often causes differential misclassification. As such, metagenome inference will introduce hard-to-predict bias (toward or away from the null) in vaginal microbiome research. IMPORTANCE Compared to taxonomic composition, the functional potential within a bacterial community is more relevant to establishing mechanistic understandings and causal relationships between the microbiome and health outcomes. Metagenome inference attempts to bridge the gap between 16S rRNA gene amplicon sequencing and whole-metagenome sequencing by predicting a microbiome's gene content based on its taxonomic composition and annotated genome sequences of its members. Metagenome inference methods have been evaluated primarily among gut samples, where they appear to perform fairly well. Here, we show that metagenome inference performance is markedly worse for the vaginal microbiome and that performance varies across common vaginal microbiome community types. Because these community types are associated with sexual and reproductive outcomes, differential metagenome inference performance will bias vaginal microbiome studies, obscuring relationships of interest. Results from such studies should be interpreted with substantial caution and the understanding that they may over- or underestimate associations with metagenome content.

Value of central review of RECIST v1.1 outcomes in the AGITG INTEGRATE randomised phase 2 international trial for advanced oesophago-gastric cancer

PurposeActivity estimates should be accurately evaluated in phase 2 clinical trials to ensure appropriate decisions about proceeding to phase 3 trials. RECIST v1.1. progression-free survival (PFS) is a common endpoint in oncology; however, it can be influenced by assessment criteria and trial design. We assessed the value of central adjudication of investigator-assessed PFS times of participants in a double-blind, randomised phase 2 trial evaluating regorafenib versus placebo in advanced gastro-oesophageal cancer (AGITG INTEGRATE) to inform plans for central review in future trials.MethodsWe calculated the proportion of participants with a disagreement between the site investigator assessment and blinded independent central review and in whom central review resulted in a change, then evaluated the effect of central review on study conclusions by comparing hazard ratios (HRs) for PFS based on site review versus central review. Post-progression unblinding was assessed with similar methods. Simulation studies explored the effect of differential and non-differential measurement error on treatment effect estimation and study power.ResultsDisagreements between site assessments versus central review occurred in 8/147 (5.4%) participants, 5 resulting in amended date of progression (3.4%). PFS HRs (sites vs central review progression dates) were similar (0.39 vs 0.40). RECIST progression occurred in 82/86 (95%) of cases where post-progression unblinding was requested by the site investigator.ConclusionsBlinded independent central review was feasible and supported the reliability of site assessments, trial results, and conclusions. Modelling showed that when treatment effects were large and outcome assessments blinded, central review was unlikely to affect conclusions.

Differential pulse oximetry readings between ethnic groups and delayed transfer to intensive care units.

Pulse oximeters are widely used to monitor blood oxygen saturations, although concerns exist that they are less accurate in individuals with pigmented skin. This study aimed to determine if patients with pigmented skin were more severely unwell at the period of transfer to intensive care units (ICUs) than individuals with White skin. Using data from a large teaching hospital, measures of clinical severity at the time of transfer of patients with COVID-19 infection to ICUs were assessed, and how this varied by ethnic group. Data were available on 748 adults. Median pulse oximetry demonstrated similar oxygen saturations at the time of transfer to ICUs (Kruskal-Wallis test, P = 0.51), although median oxygen saturation measurements from arterial blood gases at this time demonstrated lower oxygen saturations in patients classified as Indian/Pakistani ethnicity (91.6%) and Black/Mixed ethnicity (93.0%), compared to those classified as a White ethnicity (94.4%, Kruskal-Wallis test, P = 0.005). There were significant differences in mean respiratory rates in these patients (P < 0.0001), ranging from 26 breaths/min in individuals with White ethnicity to 30 breaths/min for those classified as Indian/Pakistani ethnicity and 31 for those who were classified as Black/Mixed ethnicity. These data are consistent with the hypothesis that differential measurement error for pulse oximeter readings negatively impact on the escalation of clinical care in individuals from other than White ethnic groups. This has implications for healthcare in Africa and South-East Asia and may contribute to differences in health outcomes across ethnic groups globally.

Effects of differential measurement error in self-reported diet in longitudinal lifestyle intervention studies

BackgroundLifestyle intervention studies often use self-reported measures of diet as an outcome variable to measure changes in dietary intake. The presence of measurement error in self-reported diet due to participant failure to accurately report their diet is well known. Less familiar to researchers is differential measurement error, where the nature of measurement error differs by treatment group and/or time. Differential measurement error is often present in intervention studies and can result in biased estimates of the treatment effect and reduced power to detect treatment effects. Investigators need to be aware of the impact of differential measurement error when designing intervention studies that use self-reported measures.MethodsWe use simulation to assess the consequences of differential measurement error on the ability to estimate treatment effects in a two-arm randomized trial with two time points. We simulate data under a variety of scenarios, focusing on how different factors affect power to detect a treatment effect, bias of the treatment effect, and coverage of the 95% confidence interval of the treatment effect. Simulations use realistic scenarios based on data from the Trials of Hypertension Prevention Study. Simulated sample sizes ranged from 110-380 per group.ResultsRealistic differential measurement error seen in lifestyle intervention studies can require an increased sample size to achieve 80% power to detect a treatment effect and may result in a biased estimate of the treatment effect.ConclusionsInvestigators designing intervention studies that use self-reported measures should take differential measurement error into account by increasing their sample size, incorporating an internal validation study, and/or identifying statistical methods to correct for differential measurement error.

Predicting Within-City Spatial Variations in Outdoor Ultrafine Particle and Black Carbon Concentrations in Bucaramanga, Colombia: A Hybrid Approach Using Open-Source Geographic Data and Digital Images.

Outdoor ultrafine particles (UFP, <0.1 μm) and black carbon (BC) vary greatly within cities and may have adverse impacts on human health. In this study, we used a hybrid approach to develop new models to estimate within-city spatial variations in outdoor UFP and BC concentrations across Bucaramanga, Colombia. We conducted a mobile monitoring campaign over 20 days in 2019. Regression models were trained on land use data and combined with predictions from convolutional neural networks (CNN) trained to predict UFP and BC concentrations using satellite and street-level images. The combined UFP model (R2 = 0.54) outperformed the CNN (R2 = 0.47) and land use regression (LUR) models (R2 = 0.47) on their own. Similarly, the combined BC model also outperformed the CNN and LUR BC models (R2 = 0.51 vs 0.43 and 0.45, respectively). Spatial variations in model performance were more stable for the CNN and combined models compared to the LUR models, suggesting that the combined approach may be less likely to contribute to differential exposure measurement error in epidemiological studies. In general, our findings demonstrated that satellite and street-level images can be combined with a traditional LUR modeling approach to improve predictions of within-city spatial variations in outdoor UFP and BC concentrations.

Predicting Within-City Variations in Ultrafine Particle and Black Carbon Concentrations in Bucaramanga, Columbia Using Open Source Data and Images

BACKGROUND AND AIM: Outdoor ultrafine particles (UFP; 100 nm) and black carbon (BC) vary greatly within cities and may have adverse impacts on human health such as cardiovascular mortality and brain tumour incidence. Traditional regression methods for estimating spatial distributions of ambient air pollution depend on extensive, curated geospatial information system databases. This has resulted in a disparity wherein little is known about local air pollution in the data-sparse settings of low- and middle- income countries. We used a hybrid approach to develop new models to estimate within-city spatial variations in outdoor UFP and BC concentrations across Bucaramanga, Colombia. METHODS: We conducted a large-scale mobile monitoring campaign over twenty days in 2019. Land use regression (LUR) models were developed using land use parameters from curated and open-source databases. Convolutional Neural Network (CNN) models were developed using Google Maps satellite view and street view images. Combined models were developed by combining predictions from the LUR and CNN models. After training in the training and validation sets, predictions were generated in the test set and compared to measured values. RESULTS:The combined UFP model (R2=0.54) outperformed the CNN (R2=0.47) and land use regression (LUR) models (R2=0.47) on their own. Similarly, the combined BC model also outperformed the CNN and LUR BC models (R2 = 0.51 vs 0.43 and 0.45 respectively). Spatial variations in model performance were more stable for the CNN and combined models compared to the LUR models suggesting that the combined approach may be less likely to contribute to differential exposure measurement error in epidemiological studies. CONCLUSIONS:Estimates from these models can then be applied to population-based cohorts in order to evaluate population health risks. Additionally, our findings demonstrated that satellite and street-level images can be combined with a traditional LUR modelling approach to improve predictions of within-city spatial variations in outdoor UFP and BC concentrations. KEYWORDS: Ultrafine particles, Black Carbon, Deep learning, Images, Land use regression

A Multidimensional Item Response Theory Model for Continuous and Graded Responses With Error in Persons and Items.

Item response theory “dual” models (DMs) in which both items and individuals are viewed as sources of differential measurement error so far have been proposed only for unidimensional measures. This article proposes two multidimensional extensions of existing DMs: the M-DTCRM (dual Thurstonian continuous response model), intended for (approximately) continuous responses, and the M-DTGRM (dual Thurstonian graded response model), intended for ordered-categorical responses (including binary). A rationale for the extension to the multiple-content-dimensions case, which is based on the concept of the multidimensional location index, is first proposed and discussed. Then, the models are described using both the factor-analytic and the item response theory parameterizations. Procedures for (a) calibrating the items, (b) scoring individuals, (c) assessing model appropriateness, and (d) assessing measurement precision are finally discussed. The simulation results suggest that the proposal is quite feasible, and an illustrative example based on personality data is also provided. The proposals are submitted to be of particular interest for the case of multidimensional questionnaires in which the number of items per scale would not be enough for arriving at stable estimates if the existing unidimensional DMs were fitted on a separate-scale basis.

Measurement errors in the binary instrumental variable model

Summary Instrumental variable methods can identify causal effects even when the treatment and outcome are confounded. We study the problem of imperfect measurements of the binary instrumental variable, treatment and outcome. We first consider nondifferential measurement errors, that is, the mismeasured variable does not depend on other variables given its true value. We show that the measurement error of the instrumental variable does not bias the estimate, that the measurement error of the treatment biases the estimate away from zero, and that the measurement error of the outcome biases the estimate toward zero. Moreover, we derive sharp bounds on the causal effects without additional assumptions. These bounds are informative because they exclude zero. We then consider differential measurement errors, and focus on sensitivity analyses in those settings.

Galea and Hernán Respond to "Brings to the Table," "Differential Measurement Error," and "Causal Inference in Social Epidemiology".

Social epidemiology is concerned with the health effects of forces that are "above the skin." Although causal inference should be a key goal for social epidemiology, social epidemiology and quantitative causal inference have been seemingly at odds over the years. This does not have to be the case and, in fact, both fields stand to gain through a closer engagement of social epidemiology with formal causal inference approaches. We discuss the misconceptions that have led to an uneasy relationship between these 2 fields, propose a way forward that illustrates how the 2 areas can come together to inform causal questions, and discuss the implications of this approach. We argue that quantitative causal inference in social epidemiology is an opportunity to do better science that matters, a win-win for both fields.

Using fathers as a negative control exposure: Implications of measurement bias.

Negative control exposure analysis is a very effective tool in evaluating the effect of unmeasured confounding in observational epidemiological studies. Several biases, including recall bias, time-varying confounding factors, measurement bias and so on, can affect the credibility of negative control exposure analysis for causal interpretations. The article focuses on the implications of differential measurement error across exposed group and negative controls to causal interpretations on negative control exposure analysis.

Measurement Error Without the Proxy Exclusion Restriction

–This article studies the identification of the coefficients in a linear equation when data on the outcome, covariates, and an error-laden proxy for a latent variable are available. We maintain that the measurement error in the proxy is classical and relax the assumption that the proxy is excluded from the outcome equation. This enables the proxy to directly affect the outcome and allows for differential measurement error. Without the proxy exclusion restriction, we first show that the effects of the latent variable, the proxy, and the covariates are not identified. We then derive the sharp identification regions for these effects under any configuration of three auxiliary assumptions. The first weakens the assumption of no measurement error by imposing an upper bound on the noise-to-signal ratio. The second imposes an upper bound on the outcome equation coefficient of determination that would obtain had there been no measurement error. The third weakens the proxy exclusion restriction by specifying whether the latent variable and its proxy affect the outcome in the same or the opposite direction, if at all. Using the College Scorecard aggregate data, we illustrate our framework by studying the financial returns to college selectivity and characteristics and student characteristics when the average SAT score at an institution may directly affect earnings and serves as a proxy for the average ability of the student cohort.

The relationship between chronic stress, hair cortisol and hypertension

Inconsistencies in studies of chronic psychosocial stress and hypertension may be explained by the use of stress markers greatly influenced by circadian rhythm and transient stressors. We assessed whether hair cortisol, a marker that captures systemic cortisol over months, was independently associated with hypertension. We measured hair cortisol and blood pressure in 75 consecutive participants in the Survey of the Health of Wisconsin, using an ELISA test. Individuals with values ​≥ ​median (78.1 ​pg/mg) were considered exposed. We used approximate Bayesian logistic regression, with a prior odds ratio of 1.0–4.0, to quantify the multivariate-adjusted hair cortisol-hypertension association. Participants' average age was 46.9 years; 37.3% were male; and 25.3% were hypertensive. Hypertension prevalence was 2.23 times higher in exposed (95% CI: 1.69–3.03). This finding was unlikely explained by differential measurement errors, since we conducted blinded measurements of exposure and outcome. Sensitivity analyses showed the association was unlikely explained by an unmeasured confounder, survival bias, or reverse causality bias. Findings suggest elevated hair cortisol is a risk factor for hypertension. Although feasible, the clinical value of hair cortisol as a tool for hypertension risk stratification or for monitoring the effect of chronic psychosocial stress management interventions is still uncertain.

Simple Sensitivity Analysis for Differential Measurement Error.

Sensitivity analysis results are given for differential measurement error of either the exposure or outcome. In the case of differential measurement error of the outcome, it is shown that the true effect of the exposure on the outcome on the risk ratio scale must be at least as large as the observed association between the exposure and the mismeasured outcome divided by the maximum strength of differential measurement error. This maximum strength of differential measurement error is itself assessed as the risk ratio of the controlled direct effect of the exposure on the mismeasured outcome not through the true outcome. In the case of differential measurement error of the exposure, under certain assumptions concerning classification probabilities, the true effect on the odds ratio scale of the exposure on the outcome must be at least as large as the observed odds ratio between the mismeasured exposure and the outcome divided by the maximum odds ratio of the effect of the outcome on mismeasured exposure conditional on the true exposure. The results can be immediately used to indicate the minimum strength of differential measurement error that would be needed to explain away an observed association between an exposure measurement and an outcome measurement for this to be solely due to measurement error.

Multiplicative Interactions Under Differential Outcome Measurement Error with Perfect Specificity

Multiplicative Interactions Under Differential Outcome Measurement Error with Perfect Specificity

The Effect of Probing “Don’t Know” Responses on Measurement Quality and Nonresponse in Surveys

ABSTRACTIn survey interviews, “Don’t know” (DK) responses are commonly treated as missing data. One way to reduce the rate of such responses is to probe initial DK answers with a follow-up question designed to encourage respondents to give substantive, non-DK responses. However, such probing can also reduce data quality by introducing additional or differential measurement error. We propose a latent variable model for analyzing the effects of probing on responses to survey questions. The model makes it possible to separate measurement effects of probing from true differences between respondents who do and do not require probing. We analyze new data from an experiment, which compared responses to two multi-item batteries of questions with and without probing. In this study, probing reduced the rate of DK responses by around a half. However, it also had substantial measurement effects, in that probed answers were often weaker measures of constructs of interest than were unprobed answers. These effects were larger for questions on attitudes than for pseudo-knowledge questions on perceptions of external facts. The results provide evidence against the use of probing of “Don’t know” responses, at least for the kinds of items and respondents considered in this study. Supplementary materials for this article are available online.