Su búsqueda

En autores o colaboradores
  • Bias attributable to missing data conditions is often ignored by applied researchers. In response, journal editors (e.g., von Elm et al., 2007) and governing bodies (e.g., Wilkinson & APA Task Force, 1999) are increasingly calling for missing data analysis as routine assessment. The term missing data is traditionally used in reference to the condition of a study participant missing one or more, but not all, survey responses, test items, or other data points. In addition, methods to control the bias resulting from nonequivalent comparison groups could be viewed as a “missing data” scenario (Dates & King, 2009), though in this chapter we emphasize the former. Consideration of the effects of missing data is critically needed in studies of giftedness. Frequent causes of missing data include study dropouts in longitudinal studies of children through adolescence and adulthood and incomplete data records due to oversight on the part of respondents, refusal to answer certain questions, or a host of other reasons. Missing values can sometimes be safely ignored if observed for only a very small percentage of items. Yet even a few omitted responses can significantly bias results if the data are missing systematically. For example, consider a hypothetical study of 10,000 children in which 5% are categorized as gifted. Observing a modest missing data rate of 2% may nonetheless be of concern should the absent values all fall within the gifted population (i.e., 40% of gifted children would have missing data). In such an instance, ignoring missing values or applying ineffective adjustments may drastically alter study outcomes and lead to invalid conclusions. It is therefore imperative that giftedness researchers carefully examine the consequences of even modest amounts of missing data and familiarize themselves with best practices for handling these conditions. This chapter proceeds by briefly describing a typology for categorizing missing data. We then provide a review of the strengths and weaknesses of a number of traditional and contemporary methodologies for handling missing data, followed by discussion and detailed application of a new, “hybrid” approach that is capable of adjusting for both item-level missing data and missing cases in a nonequivalent groups design, a scenario frequently faced by the applied researcher. (PsycInfo Database Record (c) 2024 APA, all rights reserved) (Source: chapter)

Última actualización de la base de datos: 2/10/26, 5:50 (CEST)

Explorar

Tipo de recurso

Año de publicación