Explore the vast range of titles available.

Mancl and DeRouen (2001, Biometrics57, 126-134) and Kauermann and Carroll (2001, JASA96, 1387-1398) proposed alternative bias-corrected covariance estimators for generalized estimating equations parameter estimates of regression models for marginal means. The finite sample properties of these estimators are compared to those of the uncorrected sandwich estimator that underestimates variances in small samples. Although the formula of Mancl and DeRouen generally overestimates variances, it often leads to coverage of 95% confidence intervals near the nominal level even in some situations with as few as 10 clusters. An explanation for these seemingly contradictory results is that the tendency to undercoverage resulting from the substantial variability of sandwich estimators counteracts the impact of overcorrecting the bias. However, these positive results do not generally hold; for small cluster sizes (e.g., <10) their estimator often results in overcoverage, and the bias-corrected covariance estimator of Kauermann and Carroll may be preferred. The methods are illustrated using data from a nested cross-sectional cluster intervention trial on reducing underage drinking.

Most general population web surveys are based on online panels maintained by commercial survey agencies. Many of these panels are based on non-probability samples. However, survey agencies differ in their panel selection and management strategies. Little is known if these different strategies cause differences in survey estimates. This paper presents the results of a systematic study designed to analyze the differences in web survey results between agencies. Six different survey agencies were commissioned with the same web survey using an identical standardized questionnaire covering factual health items. Five surveys were fielded at the same time. A calibration approach was used to control the effect of demographics on the outcome. Overall, the results show differences between probability and non-probability surveys in health estimates, which were reduced but not eliminated by weighting. Furthermore, the differences between non-probability surveys before and after weighting are larger than expected between random samples from the same population.

Background: A concern about kinematically aligned (KA) total knee arthroplasty (TKA) is that it relies on femoral components designed for mechanical alignment (MAd-FC) that could affect patellar tracking, in part, because of a trochlear groove orientation that is typically 6° from vertical. KA sets the femoral component coincident to the patient’s pre-arthritic distal and posterior femoral joint lines and restores the Q-angle, which varies widely. Relative to KA and the native knee, aligning the femoral component with MA changes most distal joint lines and Q-angles, and rotates the posterior joint line externally laterally covering the anterior femoral resection. Whether switching from a MAd- to a KAd-FC with a wider trochlear groove orientation of 20.5° from vertical results in radiographic measures known to promote patellar tracking is unknown. The primary aim was to determine whether a KAd-FC sets the trochlear groove lateral to the quadriceps line of force (QLF), better laterally covers the anterior femoral resection, and reduces lateral patella tilt relative to a MAd-FC. The secondary objective was to determine at six weeks whether the KAd-FC resulted in a higher complication rate, less knee extension and flexion, and lower clinical outcomes. Methods: Between April 2019 and July 2022, two surgeons performed sequential bilateral unrestricted caliper-verified KA TKA with manual instruments on thirty-six patients with a KAd- and MAd-FC in opposite knees. An observer measured the angle between a line best-fit to the deepest valley of the trochlea and a line representing the QLF that indicated the patient’s Q-angle. When the trochlear groove was lateral or medial relative to the QLF, the angle is denoted + or −, and the femoral component included or excluded the patient’s Q-angle, respectively. Software measured the lateral undercoverage of the anterior femoral resection on a Computed Tomography (CT) scan, and the patella tilt angle (PTA) on a skyline radiograph. Complications, knee extension and flexion measurements, Oxford Knee Score, KOOS Jr, and Forgotten Joint Score were recorded pre- and post-operatively (at 6 weeks). A paired Student’s T-test determined the difference between the KA TKAs with a KAd-FC and MAd-FC with a significance set at p < 0.05. Results: The final analysis included thirty-five patients. The 20.5° trochlear groove of the KAd-FC was lateral to the QLF in 100% (15 ± 3°) of TKAs, which was greater than the 69% (1 ± 3°) lateral to the QLF with the 6° trochlear groove of the MAd-FC (p < 0.001). The KAd-FC’s 2 ± 1.9 mm lateral undercoverage of the anterior femoral resection was less than the 4.4 ± 1.5 mm for the MAd-FC (p < 0.001). The PTA, complication rate, knee extension and flexion, and clinical outcome measures did not differ between component designs. Conclusions: The KA TKA with a KAd-FC resulted in a trochlear groove lateral to the QLF that included the Q-angle in all patients, and negligible lateral undercoverage of the anterior femoral resection. These newly described radiographic parameters could be helpful when investigating femoral components designed for KA with the intent of promoting patellofemoral kinematics.

To maintain acceptable response rates, the cost has grown for vaccination surveys that use traditional data collection modes, such as face-to-face and telephone interviews. Conducting a web or internet survey could be a low-cost alternative. However, because the internet is not used by everyone, we need to study how prevalence estimates in web surveys for vaccination surveillance could be affected by internet non-use. We analyzed data from the 2013–2017 Behavioral Risk Factor Surveillance System to assess undercoverage biases from internet non-use by partitioning into proportion of internet non-users and difference in prevalence of influenza and pneumococcal vaccinations between internet and internet non-users, respectively. The proportion of internet non-users decreased monotonically from 43.3% in 2013 to 35.4% in 2017; however, the undercoverage bias from internet use for pneumococcal vaccination increased from 0.8 to 1.5 percentage points at the same time. Overall, the undercoverage bias was −1.1 and 1.5 percentage points for influenza vaccination and pneumococcal vaccination in 2017, respectively. For both vaccinations, we found large absolute and relative biases among certain demographic subgroups. Although the proportion of internet non-users decreased in recent years, undercoverage bias of hybrid internet survey for influenza and pneumococcal vaccinations did not decrease. Despite a small overall undercoverage bias, the bias in subpopulation groups was not negligible.

We extend the empirical likelihood of Owen [Ann. Statist. 18 (1990) 90—120] by partitioning its domain into the collection of its contours and mapping the contours through a continuous sequence of similarity transformations onto the full parameter space. The resulting extended empirical likelihood is a natural generalization of the original empirical likelihood to the full parameter space; it has the same asymptotic properties and identically shaped contours as the original empirical likelihood. It can also attain the second order accuracy of the Bartlett corrected empirical likelihood of DiCiccio, Hall and Romano [Ann. Statist. 19 (1991) 1053—1061]. A simple first order extended empirical likelihood is found to be substantially more accurate than the original empirical likelihood. It is also more accurate than available second order empirical likelihood methods in most small sample situations and competitive in accuracy in large sample situations. Importantly, in many one-dimensional applications this first order extended empirical likelihood is accurate for sample sizes as small as ten, making it a practical and reliable choice for small sample empirical likelihood inference.

We develop a sequential sampling procedure for a class of stochastic programs. We assume that a sequence of feasible solutions with an optimal limit point is given as input to our procedure. Such a sequence can be generated by solving a series of sampling problems with increasing sample size, or it can be found by any other viable method. Our procedure estimates the optimality gap of a candidate solution from this sequence. If the point estimate of the optimality gap is sufficiently small according to our termination criterion, then we stop. Otherwise, we repeat with the next candidate solution from the sequence under an increased sample size. We provide conditions under which this procedure (i) terminates with probability one and (ii) terminates with a solution that has a small optimality gap with a prespecified probability.

Address-Based Sampling (ABS) is increasingly viewed as a potential remedy for the rising costs associated with in-person surveys of the U. S. general population, and for the dwindling coverage associated with telephone surveys. For small and moderate-sized surveys, ABS can help make in-person interviewing a viable mode of data collection. For large-scale studies, ABS can enable the transfer of resources from frame development to activities like training for refusal conversion, an examination of total survey error, or a nonresponse follow-up study. A synthesis of research studies estimating ABS coverage for in-person surveys shows nearly complete coverage of the household population in urban areas and varying degrees of undercoverage in rural areas. Less is known about ABS coverage of non-household populations, such as college students living in dormitories or persons residing in other group quarters. This research suggests the holistic question: Does the expediency of ABS outweigh the undercoverage that may accompany its use? The answer depends on the population being studied, the mode of data collection, and the effectiveness of frame supplementation methods. This article summarizes the current literature on the advantages and challenges of using ABS for in-person surveys as well as for mail and mixed-mode surveys.

The sandwich estimator, also known as robust covariance matrix estimator, heteroscedasticity-consistent covariance matrix estimate, or empirical covariance matrix estimator, has achieved increasing use in the econometric literature as well as with the growing popularity of generalized estimating equations. Its virtue is that it provides consistent estimates of the covariance matrix for parameter estimates even when the fitted parametric model fails to hold or is not even specified. Surprisingly though, there has been little discussion of properties of the sandwich method other than consistency. We investigate the sandwich estimator in quasi-likelihood models asymptotically, and in the linear case analytically. We show that under certain circumstances when the quasi-likelihood model is correct, the sandwich estimate is often far more variable than the usual parametric variance estimate. The increased variance is a fixed feature of the method and the price that one pays to obtain consistency even when the parametric model fails or when there is heteroscedasticity. We show that the additional variability directly affects the coverage probability of confidence intervals constructed from sandwich variance estimates. In fact, the use of sandwich variance estimates combined with t-distribution quantiles gives confidence intervals with coverage probability falling below the nominal value. We propose an adjustment to compensate for this fact.

A commonly known problem in population size estimation using registers, is that registers do not necessarily cover the whole population. This may be because they intend to cover part of the population (e.g., students), due to administrative delay or because part of the target population is not registered by default (e.g., illegal persons). One of the methods to estimate the population size in the presence of undercount is the capture-recapture method that combines the information of two or more samples. In the context of census estimation registers are used instead of samples. However, the method assumes that perfect linkage between the registers can be achieved. It is known that this assumption is often violated. In the setting of evaluating the population coverage of a census using a post-enumeration survey, a correction for linkage error was proposed. That correction was later generalized by relaxing some of the newly introduced conditions. However, the new correction method still implicitly assumed that the two registers are of equal size. We introduce a further generalization that includes both previously mentioned correction methods and at the same time deals with registers of different sizes. Specific parameter settings will correspond to the different correction methods. We show that the parameters of each method can be chosen such that the resulting estimates all equal the traditional Petersen estimate (1896) that would theoretically be obtained under truly perfect linkage.

We investigate the upper bounds on coverage probabilities of the subsampling-based confidence sets in the time series setting. Under the fixed-b asymptotic framework, where b is the ratio of block size to sample size, we derive the limiting coverage bound, and obtain the finite sample coverage bound by simulations. Our findings suggest that the coverage bound is strictly below 1 for positive b, it can be far away from 1, and the fixed-b subsampling method in Shao and Politis (2013) can exhibit serious undercoverage when the dimension of the parameter is large, the time series dependence is (positively) strong, or b is large. To alleviate the problem, we propose a generalized subsampling method that combines useful features of fixed-b subsampling and self-normalization, and demonstrate its effectiveness in terms of delivering more accurate coverage via numerical studies.