J Korean Acad Nurs.
2012 Feb;42(1):48-55. 10.4040/jkan.2012.42.1.48.
Statistical Methods to Control Response Bias in Nursing Activity Surveys
- Affiliations
-
- 1Department of Nursing, Inha University, Incheon, Korea. lim20712@inha.ac.kr
- 2College of Nursing, University of Illinois at Chicago, Chicago, USA.
- KMID: 1245157
- DOI: http://doi.org/10.4040/jkan.2012.42.1.48
Abstract
- PURPOSE
The aim of this study was to compare statistical methods to control response bias in nursing activity surveys.
METHODS
Data were collected at a medical unit of a general hospital. The number of nursing activities and consumed activity time were measured using self-report questionnaires. Descriptive statistics were used to identify general characteristics of the units. Average, Z-standardization, gamma regression, finite mixture model, and stochastic frontier model were adopted to estimate true activity time controlling for response bias.
RESULTS
The nursing activity time data were highly skewed and had non-normal distributions. Among the 4 different methods, only gamma regression and stochastic frontier model controlled response bias effectively and the estimated total nursing activity time did not exceeded total work time. However, in gamma regression, estimated total nursing activity time was too small to use in real clinical settings. Thus stochastic frontier model was the most appropriate method to control response bias when compared with the other methods.
CONCLUSION
According to these results, we recommend the use of a stochastic frontier model to estimate true nursing activity time when using self-report surveys.
