Data aggregation: Criteria for psychometric evaluation

Abstract

Criteria for the examination of data collected at the individual level and aggregated to the group are proposed. These four empirical criteria relate to content validity, representativeness, reliability, and validity of the aggregated scores as indicators of group phenomena. An example of the use of the criteria is provided using data from a large research project on the implementation of a professional practice model on nine nursing units. The use of the criteria proposed provide a systematic approach to the evaluation of individual data for use at the group level.

Intensive Care Units, Communication Between Nurses and Physicians, and Patients’ Outcomes

“…Data aggregation to the unit level requires that at least 50% of participants in each unit complete the survey for that unit to be considered adequate for group representation. 30 Analyses comparing the 4 ICUs that had low response rates with the other 21 ICUs were conducted by using t tests and χ 2 tests, and the groups did not differ significantly on most variables. All 25 ICUs were included to conserve data.…”

The Relationship Between Incident Reporting by Nurses and Safety Management in Hospitals

“…To ensure that individual-level data were representative of group-level phenomena, a composition model was applied. 33,34 On the basis of these analyses, the relationship between perceptions of incident reporting, the frequency of incident reporting on wards, and safety management was examined using the Pearson correlation coefficient. All of the analyses were conducted using SPSS 13.0 for Windows student version.…”

Linking Unit Collaboration and Nursing Leadership to Nurse Outcomes and Quality of Care

“…A 50% response rate is a generally accepted criterion for supporting the accuracy of inferences made from aggregated data. 17 We included 5 adult unit types: critical care, step-down, medical, surgical, and medical-surgical combined units. Based on these inclusion criteria, our analytic sample for this study included 29 742 RNs in 1 228 units from 200 acute care hospitals in 41 states.…”