Truncation Bias Research Articles

Bias-Corrected Estimation of Price Impact in Securities Litigation

The single-firm event studies that securities litigants use to detect the impact of a corrective disclosure on a firm’s stock price have low statistical power. As a result, observed price impacts are biased against defendants and systematically overestimate the effect on firm value. We use the empirical distribution of daily stock returns to analyze the bias and develop bias-corrected estimators of price impact in securities litigation. Because of low statistical power, the ex ante incentives against committing securities fraud are also too low. We analyze the adjustment for optimal deterrence and find that it is material, but is nowhere equal to the opposing truncation bias.

Extending technological trajectories to latest technological changes by overcoming time lags

Bibliometric methods of mapping technological trajectories encounter difficulty when identifying recent technological changes, mainly due to two forms of time lag: publication lag between invention and patent publication, and citation lag between publication and citation. To address this problem, we propose a novel way of identifying recent important publications, and of automatically connecting these to the patents-based main path. Our method reduces truncation bias due to publication and citation lags, and extends the technological trajectory to reach the latest technology at the research forefront. By generating links between patent and product, our approach can anticipate future product features that are driven by technological changes. In practice, our method enables managers to identify incremental innovation opportunities along mainstream technology dynamics, and further to detect signals of emerging technologies, wild cards, and new applications in other disciplines. An illustrative case regarding image technology is described and discussed.

The truncation problem

The truncation procedure is an approximation commonly used in plane-parallel radiative transfer codes, which consists in removing the forward scattering peak observed in the phase function of large particles (few microns) or cloud droplets, due to diffraction. This approximation allows faster calculations but introduces biases in the radiances modelled at the ground-based level and also at the top of the atmosphere in narrow angular intervals (cloud bows and glory). In this study, we recalled the principle of the truncation and present a new method to correct its flaws. In comparison with previous studies, we present here a full comprehensive correction and analysis of the truncation biases for the downward and upward radiances. For ground-level measurements, we add to truncated calculations an approximate expression of the successive scatterings in the truncated forward peak to restore the solar aureole. In case of satellite measurements, we reduce the biases found for narrow angular signatures simply by changing the expression of the primary scattering. After correction, maximal errors do not exceed 0.001 for the degree of linear polarization for optical thickness smaller than 2.0, which is a sufficient accuracy for most applications based on polarimetric measurements. This correction is available for both total and polarized radiances and is now implemented in successive order of scattering code used in the Generalized Retrieval of Aerosol and Surface Properties (GRASP) algorithm. The analysis of the problem is based on the method of successive orders, but the suggested corrections are applicable for any other resolution method (e.g. adding-doubling).

Bias-Corrected Estimation of Price Impact in Securities Litigation

The single-firm event studies that securities litigants use to detect the impact of a corrective disclosure on a firm’s stock price have low statistical power. As a result, observed price impacts are biased against defendants and systematically overestimate the effect on firm value. We use the empirical distribution of daily stock returns to analyze the bias and develop bias-corrected estimators of price impact in securities litigation. Because of low statistical power, the ex ante incentives against committing securities fraud are also too low. We analyze the adjustment for optimal deterrence and find that it is material, but is nowhere equal to the opposing truncation bias.

Semiparametric Correction for Endogenous Truncation Bias With Vox Populi-Based Participation Decision

We synthesize the knowledge present in various scientific disciplines for the development of semiparametric endogenous truncation-proof algorithm, correcting for truncation bias due to endogenous self-selection. This synthesis enriches the algorithm's accuracy, efficiency and applicability. Improving upon the covariate shift assumption, data are intrinsically affected and largely generated by their own behavior (cognition). Refining the concept of Vox Populi (Wisdom of Crowd) allows data points to sort themselves out depending on their estimated latent reference group opinion space. Monte Carlo simulations, based on 2,000,000 different distribution functions, practically generating 100 million realizations, attest to a very high accuracy of our model.

Intergenerational Educational Mobility in Rural Economy: Evidence from China and India

We extend the Becker-Tomes (1986) model of intergenerational educational mobility to a rural economy characterized by occupational dualism (farm vs. nonfarm) and provide a comparative analysis of rural India and rural China. Using two exceptional data-sets, we estimate father-sons intergenerational educational persistence in farm and nonfarm households free of truncation bias due to coresidency. The sons in rural India faced lower educational mobility compared to the sons in rural China in the 1990s and earlier. Father's nonfarm occupation and education were complementary in determining a son's schooling in India, but separable in China. However, the separability observed for the older cohorts in rural China broke down for the younger cohort. Evidence from supplementary data on economic mechanisms shows that the extended Becker-Tomes model provides plausible explanations for both the cross-country heterogeneity (India vs. China) and the evolution of mobility across cohorts in China.

Effectiveness of Population-Based Cancer Screening Only Based on Survival Data Considering Latent Lead-Time and Truncation Biases

Background: When evaluating the effectiveness of population-based screening program (comparison of survival between screen-detected and clinically detected cases), latent lead time and truncation are both inherent biases from screen-detected cases. Following the disease natural history, the time spent in preclinical screen-detectable cases (PCDP) called sojourn time is an important factor to determine these 2 biases, lead-time bias and length bias. The first is the time from diagnosis at screen to hypothetical diagnosis (entering clinical phase) depending on the sojourn time should be subtracted from the overall survival time to achieve a fair comparison. The second is pertaining to oversampling cases with long sojourn time at prevalent screen. The survival benefit is often overestimated without considering these 2 biases. Aim: To unbiasedly estimate the effectiveness of population-based screening in terms of hazard ratio between screen-detected and clinically detected cases. Methods: Walter and Stitt (1985) has already proposed a parametric approach to adjust lead-time bias. However, lead-time bias and truncation biases are correlated as both have shared information on sojourn time in common, and therefore we aim to consider these 2 biases at the same time. Both parametric and semiparametric approaches are proposed to address the hazard ratio of risk of cancer death adjusting for lead-time and truncation biases based on data from the Swedish W-county trial. The estimation of parameters computer algorithms was also adopted with non-Bayesian and Bayesian approaches using SAS and R. Results: By using parametric approach, the hazard ratio was inflated from 0.26 (0.18-0.36) without adjustment to 0.38 (0.25-0.55) with lead-time adjustment and 0.43 (0.27-0.60) with both lead-time and truncation adjustments. The use of semiparametric model led to 0.59 (0.49-0.70) of adjusted hazard ratio. After considering tumor attributes, the hazard ratio was inflated to 0.76 (0.45-1.11) and 0.87 (0.71-1.04) separately under parametric and semiparametric approaches. Conclusion: Unbiased evaluation of effectiveness of population-based screen with adjustment for lead-time and truncation biases based on survival data are efficient and feasible.

Access to antiretroviral therapy in HIV-infected children aged 0-19 years in the International Epidemiology Databases to Evaluate AIDS (IeDEA) Global Cohort Consortium, 2004-2015: A prospective cohort study.

IntroductionAccess to antiretroviral therapy (ART) is a global priority. However, the attrition across the continuum of care for HIV-infected children between their HIV diagnosis and ART initiation is not well known. We analyzed the time from enrollment into HIV care to ART initiation in HIV-infected children within the International Epidemiology Databases to Evaluate AIDS (IeDEA) Global Cohort Consortium.Methods and findingsWe included 135,479 HIV-1-infected children, aged 0–19 years and ART-naïve at enrollment, between 1 January 2004 and 31 December 2015, in IeDEA cohorts from Central Africa (3 countries; n = 4,948), East Africa (3 countries; n = 22,827), West Africa (7 countries; n = 7,372), Southern Africa (6 countries; n = 93,799), Asia-Pacific (6 countries; n = 4,045), and Latin America (7 countries; n = 2,488). Follow-up in these cohorts is typically every 3–6 months. We described time to ART initiation and missed opportunities (death or loss to follow-up [LTFU]: last clinical visit >6 months) since baseline (the date of HIV diagnosis or, if unavailable, date of enrollment). Cumulative incidence functions (CIFs) for and determinants of ART initiation were computed, with death and LTFU as competing risks. Among the 135,479 children included, 99,404 (73.4%) initiated ART, 1.9% died, 1.4% were transferred out, and 20.4% were lost to follow-up before ART initiation. The 24-month CIF for ART initiation was 68.2% (95% CI: 67.9%–68.4%); it was lower in sub-Saharan Africa—ranging from 49.8% (95% CI: 48.4%–51.2%) in Central Africa to 72.5% (95% CI: 71.5%–73.5%) in West Africa—compared to Latin America (71.0%, 95% CI: 69.1%–72.7%) and the Asia-Pacific (78.3%, 95% CI: 76.9%–79.6%). Adolescents aged 15–19 years and infants <1 year had the lowest cumulative incidence of ART initiation compared to other ages: 62.2% (95% CI: 61.6%–62.8%) and 66.4% (95% CI: 65.7%–67.0%), respectively. Overall, 49.1% were ART-eligible per local guidelines at baseline, of whom 80.6% initiated ART. The following children had lower cumulative incidence of ART initiation: female children (p < 0.01); those aged <1 year, 2–4 years, 5–9 years, and 15–19 years (versus those aged 10–14 years, p < 0.01); those who became eligible during follow-up (versus eligible at enrollment, p < 0.01); and those receiving care in low-income or lower-middle-income countries (p < 0.01). The main limitations of our study include left truncation and survivor bias, caused by deaths of children prior to enrollment, and use of enrollment date as a proxy for missing data on date of HIV diagnosis, which could have led to underestimation of the time between HIV diagnosis and ART initiation.ConclusionsIn this study, 68% of HIV-infected children initiated ART by 24 months. However, there was a substantial risk of LTFU before ART initiation, which may also represent undocumented mortality. In 2015, many obstacles to ART initiation remained, with substantial inequities. More effective and targeted interventions to improve access are needed to reach the target of treating 90% of HIV-infected children with ART.

Semi-Parametric Correction for Endogenous Truncation Bias with Vox Populi Based Participation Decision

Not having access to the entire data distribution entails severe biases in the resulting parameter estimates of any empirical investigation and which hampers the understanding and analysis of the data generation process. We integrate and synthesize the knowledge present in various scientific disciplines for the development of semiparametric endogenous truncation-proof algorithms correcting for truncation bias due to endogenous self-selection. Thus, computer science (pattern recognition), machine (unsupervised) learning, electrical engineering (signal extraction), economics, psychology, and management science all contribute to the ideas and techniques embedded in the offered algorithms developed. This synthesis enriches the algorithms' accuracy, efficiency and applicability. Under truncation the unobserved portion of the distribution is similar to the unobserved pixels problems in pattern recognition dealt with in computer science, machine learning and artificial intelligence algorithms. However, data in the social sciences are intrinsically affected and largely generated by their own behavior (cognition). Thus, incorporating behavioral aspects from economics, psychology and management science to play a role in the model allows for endogeneity to take place, be modeled and controlled. The algorithms offered improve upon the covariate-shift assumption in machine learning in that, each data point's decision to truncate itself out from the original distribution is an important building block generating the estimation algorithms. Refining the concept of Vox Populi (Wisdom of Crowd) allows data points to sort themselves out depending on their estimated latent reference group opinion space. The opinion space is composed of experts' observed and unobserved characteristics. The later are captured by latent classes. The emerging algorithms are semiparametric and are of the orthonormal polynomials sequence family (such as Fourier) known to have flexibility and are useful in non-parametric analysis. The most attractive feature of this estimator for our purpose is that it intrinsically prevents potential multicollinearity problems, a feature the kernel estimator does not possess. Each datum is generated by a different distribution function characterized as a finite mixture of various continuous distribution functions which are not restricted to be unimodal or symmetric. Thus, the proposed algorithm is distribution free. The number of reference groups is not arbitrarily imposed but rather estimated using SCAD (smoothly clipped absolute deviation) penalization mechanism. Monte-Carlo simulations based on 2,000,000 different distribution functions, practically generating 100 millions realizations which are not i.i.d and attest to a very high accuracy of our model.

Public-to-Private Buyouts and Innovation

We study the effect of public-to-private buyout transactions on investments in innovation using an international sample over the 1997-2017 period. We use patent counts and citations to proxy for the quantity, quality, and economic importance of innovation. Our results are based on time analysis and matched sample regressions. The data indicate that buyouts are associated with a significant reduction in patents and patent citations, including a reduction in radical (i.e., more scientific) patents. When we split the sample into institutional and management buyouts, the negative effect of buyouts is confirmed only for institutional buyouts. This suggests that only institutional buyouts prevent target firms from adopting long-term investments. This finding is confirmed by reductions in innovator employment and innovation efficiency subsequent to going private. Moreover, the data indicate that the negative effect is most prevalent for transactions where the cost of the deal’s debt financing is higher than that of the debt post-buyout. We rule out some alternative explanations for these findings, including but not limited to outliers, truncation bias, and endogeneity.

Comparison of different digital elevation models and satellite imagery for lineament analysis: Implications for identification and spatial arrangement of fault zones in crystalline basement rocks of the southern Black Forest (Germany)

GIS-based remote sensing techniques and lineament mapping provide additional information on the spatial arrangement of faults and fractures in large areas with variable outcrop conditions. Due to inherent censoring and truncation bias mapping of lineaments is still a challenging task. In this study we show how statistical evaluations help to improve the reliability of lineament mappings by comparing two digital elevation models (ASTER, LIDAR) and satellite imagery data sets in the seismically active southern Black Forest. A statistical assessment of the orientation, average length, and the total length of mapped lineaments reveals an impact of the different resolutions of the data sets that allow to define maximum (censoring bias) and minimum (truncation bias) observable lineament length for each data set. The increase of the spatial resolution of the digital elevation model from 30 m × 30 m to 5 m × 5 m results in a decrease of total lineament length by about 40% whereby the average lineament lengths decrease by about 60%. Lineament length distributions of both data sets follow a power law distribution as documented elsewhere for fault and fracture systems. Predominant NE-, N-, NNW-, and NW-directions of the lineaments are observed in all data sets and correlate with well-known, mappable large-scale structures in the southern Black Forest. Therefore, mapped lineaments can be correlated with faults and hence display geological significance. Lineament density in the granite-dominated areas is apparently higher than in the gneiss-dominated areas. Application of a slip- and dilation tendency analysis on the fault pattern reveals largest reactivation potentials for WNW-ESE and N-S striking faults as strike-slip faults whereas normal faulting may occur along NW-striking faults within the ambient stress field. Remote sensing techniques in combination with highly resolved digital elevation models and a slip- and dilation tendency analysis thus can be used to quickly get first order results of the spatial arrangement of critically stressed faults in crystalline basement rocks.

Survivor bias in Mendelian randomization analysis.

Mendelian randomization studies employ genotypes as experimental handles to infer the effect of genetically modified exposures (e.g. vitamin D exposure) on disease outcomes (e.g. mortality). The statistical analysis of these studies makes use of the standard instrumental variables framework. Many of these studies focus on elderly populations, thereby ignoring the problem of left truncation, which arises due to the selection of study participants being conditional upon surviving up to the time of study onset. Such selection, in general, invalidates the assumptions on which the instrumental variables analysis rests. We show that Mendelian randomization studies of adult or elderly populations will therefore, in general, return biased estimates of the exposure effect when the considered genotype affects mortality; in contrast, standard tests of the causal null hypothesis that the exposure does not affect the mortality rate remain unbiased, even when they ignore this problem of left truncation. To eliminate "survivor bias" or "truncation bias" from the effect of exposure on mortality, we next propose various simple strategies under a semi-parametric additive hazard model. We examine the performance of the proposed methods in simulation studies and use them to infer the effect of vitamin D on all-cause mortality based on the Monica10 study with the genetic variant filaggrin as instrumental variable.

Left Truncation Bias to Explain the Protective Effect of Smoking on Preeclampsia: Potential, But How Plausible?

An inverse association between maternal smoking and preeclampsia has been frequently observed in epidemiologic studies for several decades. In the May 2015 issue of this journal, Lisonkova and Joseph described a simulation study suggesting that bias from left truncation might explain the inverse association. The simulations were based on strong assumptions regarding the underlying mechanisms through which bias might occur. To examine the sensitivity of the previous authors' conclusions to these assumptions, we constructed a new Monte Carlo simulation using published estimates to frame our data-generating parameters. We estimated the association between smoking and preeclampsia across a range of scenarios that incorporated abnormal placentation and early pregnancy loss. Our results confirmed that the previous authors' findings are highly dependent on assumptions regarding the strength of association between abnormal placentation and preeclampsia. Thus, the bias they described may be less pronounced than was suggested. Under empirically derived constraints of these critical assumptions, left truncation does not appear to fully explain the inverse association between smoking and preeclampsia. Furthermore, when considering processes in which left truncation may result from the exposure, it is important to precisely describe the target population and parameter of interest before assessing potential bias. We comment on the specification of a meaningful target population when assessing maternal smoking and preeclampsia as a public health issue. We describe considerations for defining a target population in studies of perinatal exposures when those exposures cause competing events (e.g., early pregnancy loss) for primary outcomes of interest.

Further exploration of dissemination bias in qualitative research required to facilitate assessment within qualitative evidence syntheses

ObjectivesTo conceptualise and discuss dissemination bias in qualitative research. ResultsIt is likely that the mechanisms leading to dissemination bias in quantitative research, including time lag, language, gray literature, and truncation bias also contribute to dissemination bias in qualitative research. These conceptual considerations have informed the development of a research agenda. ConclusionFurther exploration of dissemination bias in qualitative research is needed, including the extent of non-dissemination and related dissemination bias, and how to assess dissemination bias within qualitative evidence syntheses. We also need to consider the mechanisms through which dissemination bias in qualitative research could occur to explore approaches for reducing it.

Truncation bias corrections in patent data: Implications for recent research on innovation

We review the effectiveness of various adjustment methods in correcting the truncation-bias in patents data and the implications for existing studies. The NBER patents-database was recently updated, extending the sample from 2006 to 2010. The updated sample is largely free of truncation-bias over the period covered by the NBER-2006 sample, allowing us to evaluate the bias-adjustment methods. We find that existing adjustments perform poorly towards the end of NBER-2006 sample. We re-examine multiple studies from the recent literature on innovation and show that findings based on the last few years of NBER-2006 data are not supported in the updated patents-database.

Estimating Intergenerational Mobility with Incomplete Data: Coresidency and Truncation Bias in Rank-Based Relative and Absolute Mobility Measures

The rank-based measures of intergenerational mobility have become increasingly popular in economics literature. Recent evidence shows that rank-based measures are less affected by measurement error and life-cycle bias compared with other standard measures such as intergenerational regression coefficient and intergenerational correlation. However, most of the available household surveys suffer from sample truncation, because coresidency is used to define household membership. There is no evidence on how sample truncation affects the rank-based mobility estimates relative to intergenerational regression coefficient and intergenerational correlation. This paper provides evidence on this in the context of intergenerational schooling persistence, using two exceptional surveys from India and Bangladesh that include all children irrespective of residency status. The analysis shows that the measures of relative mobility (slopes) are biased downward in coresident samples, but the average bias in rank correlation is less than half of that in intergenerational regression coefficient, and comparable to that in intergenerational correlation in magnitude. The intercept estimates are biased upward, with the largest bias found in the intercept of the regression used for intergenerational correlation. Truncation bias in rank-based absolute mobility estimates is the lowest in most cases. The results strengthen the case for rank-based measures of intergenerational mobility when working with the standard household surveys.

Semi-Parametric Correction for Endogenous Truncation Bias with Vox Populi Based Participation Decision

We synthesize the knowledge present in various scientific disciplines for the development of semiparametric endogenous truncation-proof algorithm, correcting for truncation bias due to endogenous self-selection. This synthesis enriches the algorithm's accuracy, efficiency and applicability. Improving upon the covariate shift assumption, data are intrinsically affected and largely generated by their own behavior (cognition). Refining the concept of Vox Populi (Wisdom of Crowd) allows data points to sort themselves out depending on their estimated latent reference group opinion space. Monte Carlo simulations, based on 2,000,000 different distribution functions, practically generating 100 million realizations, attest to a very high accuracy of our model.

Measuring nonfundamentalness for structural VARs

As nonfundamental vector moving averages do not have causal VAR representations, standard structural VAR methods are deemed inappropriate for recovering the economic shocks of general equilibrium models with nonfundamental reduced forms. In the previous literature it has been pointed out that, despite nonfundamentalness, structural VARs may still be good approximating models. I characterize nonfundamentalness as bias depending on the zeros of moving average filters. However, measuring the nonfundamental bias is not trivial because of the simultaneous occurrence of lag truncation bias. I propose a method to disentangle the bias based on population spectral density and derive a measure for the nonfundamental bias in population. In the application, I find that the SVAR exercises of Sims (2012) are accurate because the nonfundamental bias is mild.

Incident Ischemic Heart Disease After Long-Term Occupational Exposure to Fine Particulate Matter: Accounting for 2 Forms of Survivor Bias.

Little is known about the heart disease risks associated with occupational, rather than traffic-related, exposure to particulate matter with aerodynamic diameter of 2.5 µm or less (PM2.5). We examined long-term exposure to PM2.5 in cohorts of aluminum smelters and fabrication workers in the United States who were followed for incident ischemic heart disease from 1998 to 2012, and we addressed 2 forms of survivor bias. Left truncation bias was addressed by restricting analyses to the subcohort hired after the start of follow up. Healthy worker survivor bias, which is characterized by time-varying confounding that is affected by prior exposure, was documented only in the smelters and required the use of marginal structural Cox models. When comparing always-exposed participants above the 10th percentile of annual exposure with those below, the hazard ratios were 1.67 (95% confidence interval (CI): 1.11, 2.52) and 3.95 (95% CI: 0.87, 18.00) in the full and restricted subcohorts of smelter workers, respectively. In the fabrication stratum, hazard ratios based on conditional Cox models were 0.98 (95% CI: 0.94, 1.02) and 1.17 (95% CI: 1.00, 1.37) per 1 mg/m3-year in the full and restricted subcohorts, respectively. Long-term exposure to occupational PM2.5 was associated with a higher risk of ischemic heart disease among aluminum manufacturing workers, particularly in smelters, after adjustment for survivor bias.

Abstract 17352: Predictors of Ventricular Arrhythmias and Sudden Death in a Quebec Cohort With Brugada Syndrome

Introduction: Patients with Brugada syndrome (BrS) are at risk for ventricular arrhythmias (VAs) and sudden death. Identification of high-risk individuals beyond those with syncope or resuscitated SCD remains a major challenge. We sought to assess the value of clinical, electrophysiological, and ECG features of depolarization and repolarization in predicting arrhythmic events and SCD in a Quebec population. Methods: Consecutive patients with BrS diagnosed between 2002 and 2013 were recruited from three major university hospitals in the province of Quebec, Canada. Relevant clinical features were recorded at baseline. Qualifying ECGs with the highest type 1 ST-segment elevations were reviewed and analyzed by two electrophysiologists blinded to clinical history, with the following parameters assessed: S duration in V1, aVR sign, QRS duration in V6, R-J interval in V2, maximal Tp-e, duration, Tp-e dispersion, QRS-f, ST elevation and QTc in V2, and presence of an inferolateral ER pattern. Survival analyses considered a left truncation model to account for potential sampling bias, considering that those who died suddenly before recognition of BrS were necessarily excluded. Results: A total of 105 consecutive patients, 79.8% male, were diagnosed with BrS at a median age of 48 years and were followed for 5.0±1.2 years. Ten (9.5%) had a history of SCD, 37 (35.2%) had syncope, and 7 (6.7%) experienced 20 arrhythmic events during follow-up (7 anti-tachycardia pacing; 13 shocks). In multivariate Cox regression analyses adjusted for left truncation and Firth bias correction, a spontaneous type I ECG pattern (HR 10.80; 95% CI 1.03 to 113.87; p=0.0476)], maximal Tp-e duration ≥100ms (HR 29.73; 95% CI 1.33 to 666.37; p= 0.0325) and QRS duration in V6&gt;110 ms (HR 15.27; 95% CI 1.07 to 217.42; p=0.0443) were independently associated with VAs or aborted SCD. Family history of SCD, AF, and inducible VAs during programmed ventricular stimulation were not associated with the primary endpoint. Conclusion: In a cohort of patients with BrS from the province of Quebec, VAs and SCD were independently associated with standard 12-lead ECG features including a spontaneous type 1 pattern, depolarization (QRS in V6 &gt;110 ms), and repolarization (maximal Tp-e duration) criteria.