Investigation and treatment of missing item scores in test and questionnaire data

Multivariate Behavioral Research 06/2010; 38(4). DOI: 10.1207/s15327906mbr3804_4

ABSTRACT This article first discusses a statistical test for investigating whether or not the pattern of missing scores in a respondent-by-item data matrix is random. Since this is an asymptotic test, we investigate whether it is useful in small but realistic sample sizes. Then, we discuss two known simple imputation methods, person mean (PM) and two-way (TW) imputation, and we propose two new imputation methods, response-function (RF) and mean response-function (MRF) imputation. These methods are based on few assumptions about the data structure. An empirical data example with simulated missing item scores shows that the new method RF was superior to the methods PM, TW, and MRF in recovering from incomplete data several statistical properties of the original complete data. Methods TW and RF are useful both when item score missingness is ignorable and nonignorable.

  • Journal of Pediatric Psychology 01/2013; · 2.91 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Pediatric chronic pain, which can result in deleterious effects for the child, bears the risk of aggravation into adulthood. Intensive interdisciplinary pain treatment (IIPT) might be an effective treatment, given the advantage of consulting with multiple professionals on a daily basis. Evidence for the effectiveness of IIPT is scarce. Here, we investigated the efficacy of an IIPT within a randomized controlled trial by comparing an intervention group (IG) (n=52) to a waiting-list control group (WCG) (n=52). We made assessments before (PRE), immediately after treatment (POST), as well as at short-term (POST6MONTHS) and long-term (POST12MONTHS) follow-up. We determined a combined endpoint, "improvement" (pain intensity, disability, school absence) [16,21], and investigated three additional outcome domains (anxiety, depression, catastrophizing). We also investigated changes in economic parameters (healthcare use, parental work absenteeism, subjective financial burden) and their relationship to the child's improvement. Results at POST showed that significantly more children in the IG than in the WCG were assigned to improvement (55% compared to 14%; Fisher p<.001; 95% CI for incidence difference: 0.21%-0.60%). While immediate effects were achieved for disability, school absence, depression and catastrophizing, pain intensity and anxiety did not change until short-term follow-up. More than 60% of the children in both groups were improved long-term. The parents reported significant reductions in all economic parameters. The results from the present study support the efficacy of the IIPT. Future research is warranted to investigate differences in treatment response and to understand the changes in economic parameters in nonimproved children.
    Pain 09/2013; · 5.64 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Se investigó la influencia que tienen los valores perdidos sobre la estimación de las propiedades psicométricas de los tests de personalidad. Se ha utilizado un enfoque aplicado, simulando situaciones que puedan te-ner relevancia cara a la práctica profesional. Se diseñaron dos estudios de simulación a partir de datos reales obtenidos de la aplicación de la prueba ESQUIZO-Q que evalúa esquizotipia. En el primero de ellos se utilizó una muestra de 3056 personas y en el segundo una de 200; en ambos casos se emplearon cuatro niveles de pérdida de respuestas y ocho procedimientos de imputación de los valores perdidos. Se estudió su influencia sobre las es-timaciones del coeficiente α de Cronbach, la estructura factorial de la prue-ba y la ordenación de las puntuaciones en el cuestionario. Los resultados apuntan a que en presencia de niveles bajos de valores perdidos, incluso los métodos más simples, ofrecen soluciones muy razonables desde el punto de vista práctico. Desde la perspectiva más estadística el procedimiento de Expectación-Maximización (EM) es el que presenta un mejor comporta-miento global en los diferentes criterios manejados. Destaca también el pobre comportamiento de los métodos de sustitución por el valor anterior o posterior de cara a mantener la estructura factorial de los datos.
    Anales de Psicología 01/2013; · 0.55 Impact Factor


Available from