Browsing by Subject "missing data"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Bayesian Causal Inferencee In Meta-Analysis
    (2019-05) Zhou, Jincheng
    While the randomized clinical trial (RCT) is the gold standard for investigating the effect of a medical intervention, noncompliance to assigned treatments can threaten a trial's validity. Noncompliance, if not appropriately controlled, can introduce substantial bias into the estimate of treatment effect. The complier average causal effect (CACE) approach provides a useful tool for addressing noncompliance, where CACE measures the effect of an intervention in the latent subgroup of the study population that complies with its assigned treatment (the compliers). Meta-analysis of RCTs has become a widely-used statistical technique to combine and contrast results from multiple independent studies. However, no existing methods can effectively deal with heterogeneous noncompliance in meta-analysis of RCTs. For example, the commonly used meta-analysis regression methods investigate the impact of study-level variables (e.g., mean age of the study population) on the study-specific treatment effect size by assuming the study-level covariates to be fixed. However, noncompliance rates generally differ between treatment groups within a study and are commonly considered as random rather than fixed post-randomization variables. In addition, noncompliance may dynamically interact with the primary outcome and thus affect the response to treatment. Thus, meta-regression methods are not suitable to controlling for noncompliance. This thesis focuses on developing Bayesian methods to estimate CACE in meta-analysis of RCTs with binary or ordinal outcomes. Bayesian hierarchical random effects models are developed to appropriately account for the inherent heterogeneity in treatment effect and noncompliance between studies and treatment groups. We first present a Bayesian hierarchical model to estimate the CACE where heterogeneous compliance rates are available for each study. Second, we extend our approach to deal with incomplete noncompliance when some RCTs do not report noncompliance data. The results are illustrated by a re-analysis of a meta-analysis comparing the effect of epidural analgesia in labor versus no or other analgesia in labor on the outcome cesarean section, where noncompliance varies substantially between studies. Simulations are performed to evaluate the performance of the proposed approach and to illustrate the importance of including appropriate random effects by showing the impact of over- and under-fitting. Furthermore, we develop an R package, BayesCACE, to provide user-friendly functions to implement CACE analysis for binary outcomes based on the proposed Bayesian hierarchical models. This package includes flexible functions for analyzing data from a single RCT and from a meta-analysis of multiple RCTs with either complete or incomplete noncompliance data. The package also provides various functions for generating forest, trace, posterior density, and auto-correlation plots, and to review noncompliance rates, visually assess the model, and obtain study-specific and overall CACEs.
  • Thumbnail Image
    Item
    Estimating themissing species bias in plant trait measurements
    (Wiley, 2015) Sandel, Brody; Gutiérrez, Alvaro G; Reich, Peter B; Schrodt, Franziska; Dickie, John; Kattge, Jens
    Aim Do plant trait databases represent a biased sample of species, and if so, can that bias be corrected? Ecologists are increasingly collecting and analysing data on plant functional traits, and contributing them to large plant trait databases. Many applications of such databases involve merging trait measurements with other data such as species distributions in vegetation plots; a process that invariably produces matrices with incomplete trait and species data. Typically, missing data are simply ignored and it is assumed that the missing species are missing at random. Methods Here, we argue that this assumption is unlikely to be valid and propose an approach for estimating the strength of the bias regarding which species are represented in trait databases. The method leverages the fact that, within a given database, some species have many measurements of a trait and others have few (high vs low measurement intensity). In the absence of bias, there should be no relationship between measurement intensity and trait values. We demonstrate the method using five traits that are part of the TRY database, a global archive of plant traits. Our method also leads naturally to a correction for this bias, which we validate and apply to two examples. Results Specific leaf area and seed mass were strongly positively biased (frequently measured species had higher trait values than rarely measured species), leaf nitrogen per unit mass and maximum height were moderately negatively biased, and maximum photosynthetic capacity per unit leaf area was weakly negatively biased. The bias-correction method yielded greatly improved estimates in the validation tests for the two most biased traits. Further, in our two applications, ecological interpretations were shown to be sensitive to uncorrected bias in the data. Conclusions Species inclusion in trait databases appears to be strongly biased in some cases, and failure to correct this can lead to incorrect conclusions.
  • Thumbnail Image
    Item
    Revision and phylogeny of the caddisfly subfamily Protoptilinae (Trichoptera: Glossosomatidae) inferred from adult morphology and mitochondrial DNA
    (Magnolia Press, 2013) Robertson, Desiree R.; Holzenthal, Ralph W.
    Protoptilinae Ross, 1956, is the most diverse subfamily belonging to the saddle- or tortoise-case-making caddisfly family Glossosomatidae Wallengren, 1891. The subfamily has a disjunct distribution: 5 genera are known from the East Palaearctic and Oriental regions; the remaining 13 are restricted to the Nearctic and Neotropical regions. Monophyly of Protoptilinae and each of 17 genera was tested using 80 taxa, 99 morphological characters, and mitochondrial DNA (COI). Additionally, homologies of morphological characters were assessed across genera and a standardized terminology for those structures was established. Mitochondrial DNA data were unavailable for 55 of the 80 taxa included in this study. To test the effects of the missing molecular data, 5 different datasets were analyzed using both parsimony and Bayesian methods. There was incongruence between the COI and morphological data, but results suggest the inclusion of COI data in a combined analysis, although incomplete, improved the overall phylogenetic signal. Bayesian and parsimony analyses of all 5 datasets strongly supported the monophyly of Protoptilinae. Monophyly of the following genera was also supported: Canoptila Mosely, 1939; Culoptila Mosely, 1954; Itauara Müller, 1888; Mastigoptila Flint, 1967; Mortoniella Ulmer, 1906; Protoptila Banks, 1904; and Tolhuaca Schmid, 1964. Several taxonomic changes were necessary for classification to reflect phylogeny accurately. Accordingly, Matrioptila Ross, 1938; Poeciloptila Schmid, 1991; Temburongpsyche Malicky, 1992; and Nepaloptila Kimmins, 1964, are designated new junior synonyms of Padunia Martynov, 1910. Additionally, the endemic Caribbean genera Campsiophora Flint, 1964, and Cubanoptila Sykora, 1973, are designated new junior synonyms of Cariboptila Flint, 1964. Diagnoses and a key to the subfamilies of Glossosomatidae and world genera of Protoptilinae incorporating these taxonomic changes are provided.