• Presentation

"This call may be monitored or recorded for quality assurance purposes"


Stockdale, J. D., Thornburg, V. E., & Aldridge, A. P. (2010, May). #8220;This call may be monitored or recorded for quality assurance purposes#8221;. Presented at IFD&TC 2010, .


Research shows that reporting of alcohol and illicit drug use varies from one mode of data collection to another, with telephone respondents less likely to report these sensitive behaviors than respondents interviewed face-to-face, largely due to issues of trust and confidentiality between respondents and interviewers. One of the factors that may help to explain lower rates of reported sensitive behaviors is that respondents are told at the beginning of their telephone interview that their interview may be monitored or recorded for quality assurance purposes. And, literature shows that this notification is being regularly utilized in the research and marketing fields and is required for research studies by some institutional review boards (IRBs). An RTI study sponsored by the Substance Abuse and Mental Health Services Administration presented us with a unique opportunity to explore whether this notification process influences rates of alcohol and illicit substance reporting, since this study’s data included several hundred telephone interviews completed before and then after a notification message was included at the front end of the interview.

We hypothesized that respondents who are informed their responses are being recorded are less likely to report their past and recent alcohol and illicit substance behaviors. In examining the data, we varied our analysis design specifications to determine how sensitive our results are to the influences of respondent heterogeneity and state fixed effects (effects which can be caused by demographic characteristics within a particular state). Within a multivariate regression discontinuity framework we estimated the effect of being recorded and found large decreases in the prevalence of reporting any illicit substance (11.5 to 16.4 percentage points) but only moderate statistical significance (p = .083 to .168).