In the course of the evaluation of experiments on a certain technology, different chemometric methods has been used. The aim of the experiments was the determination of those technical circumstances where methane could be most efficiently converted to formaldehyde by catalytic oxidation. The process
A cautionary case study of approaches to the treatment of missing data
β Scribed by Christopher Paul; William M. Mason; Daniel McCaffrey; Sarah A. Fox
- Publisher
- Springer
- Year
- 2008
- Tongue
- English
- Weight
- 278 KB
- Volume
- 17
- Category
- Article
- ISSN
- 1613-981X
No coin nor oath required. For personal study only.
β¦ Synopsis
This article presents findings from a case study of different approachesto the treatment of missing data. Simulations based on data from the Los AngelesMammography Promotion in Churches Program (LAMP) led the authors to the followingcautionary conclusions about the treatment of missing data: (1) Automatedselection of the imputation model in the use of full Bayesian multiple imputation canlead to unexpected bias in coefficients of substantive models. (2) Under conditionsthat occur in actual data, casewise deletion can perform less well than we were led toexpect by the existing literature. (3) Relatively unsophisticated imputations, such as mean imputation and conditional mean imputation, performed better than the technicalliterature led us to expect. (4) To underscore points (1), (2), and (3), the article concludes that imputation models are substantive models, and require the same cautionwith respect to specificity and calculability.
π SIMILAR VOLUMES
## Abstract Effective waste management must address waste reduction, reuse, recovery/recycle and, as the least progressive option, waste treatment. Catalysis can serve as an integral green processing tool, ensuring lower operating pressures/temperatures with a reduction in energy requirements while
## Abstract A review of the world literature indicates that there is a multiplicity of treatment approaches for hyperhidrosis. Most of these concentrate on the symptom itself and range from topical applications to sympathectomy. This paper demonstrates the value of an integrative approach in which
One of the main problems of data mining is imperfection of input data. Such data may be uncertain, vague, and incomplete. In our data set, describing preterm birth, many attribute values were missing, that is, the input data set was incomplete. The main approach to solving the missing attribute valu
The purpose of this paper is to describe the degree of compliance with quality of life measures in two clinical trials conducted by the Australian New Zealand Breast Cancer Trials Group comparing different chemotherapy policies for metastatic breast cancer. Quality of life was assessed by the patien