Accuracy of data transfer: double data entry and estimating levels of error

Aims and objectives Most nursing research using quantitative empirical data involves entering information collected on data collection forms into a computer. This paper brings to attention issues related to the introduction of errors during this transfer of data and makes some recommendations as to how this might be dealt with. Background Beyond concerns with the reliability and validity of data collection instruments, the issue of data accuracy and the introduction of errors in research data sets receive little mention in the nursing research literature. Given that studies have reported up to a 26·9 per cent error rate in data transfer (Goldberg et al. Proceedings American Medical Informatics Association, Annual Symposium November 2008), it is important that this potential problem is addressed in the context of nursing research. Design Discursive paper. Conclusions An approach to checking the accuracy of data entered onto a computer via the keyboard is suggested, and a method for estimating the level of confidence that could be claimed regarding the number of errors introduced is outlined. Relevance to clinical practice If nursing is to be investigated by quantitative methods, it would seem that the accuracy of study findings is paramount. This paper identifies one area of research where errors are prone to arise and suggests a method for containing these.