The problem of bias when nursing facility staff administer customer satisfaction surveys

Customer satisfaction instruments are being used with increasing frequency to assess and monitor residents' assessments of quality of care in nursing facilities. There is no standard protocol, however, for how or by whom the instruments should be administered when anonymous, written responses are not feasible. Researchers often use outside interviewers to assess satisfaction, but cost considerations may limit the extent to which facilities are able to hire outside interviewers on a regular basis. This study was designed to investigate the existence and extent of any bias caused by staff administering customer satisfaction surveys.

Customer satisfaction data were collected in 1998 from 265 residents in 21 nursing facilities in North Dakota. Half the residents in each facility were interviewed by staff members and the other half by outside consultants; scores were compared by interviewer type. In addition to a tabulation of raw scores, ordinary least-squares analysis with facility fixed effects was used to control for resident characteristics and unmeasured facility-level factors that could influence scores.

Significant positive bias was found when staff members interviewed residents. The bias was not limited to questions directly affecting staff responsibilities but applied across all types of issues. The bias was robust under varying constructions of satisfaction and dissatisfaction.

A uniform method of survey administration appears to be important if satisfaction data are to be used to compare facilities. Bias is an important factor that should be considered and weighed against the costs of obtaining outside interviewers when assessing customer satisfaction among long term care residents.