Bias in Survey Completion By Alcohol and Drug Users When Using Interactive Voice Response

Schedule:
Friday, January 16, 2015: 3:00 PM
La Galeries 2, Second Floor (New Orleans Marriott)
* noted as presenting author
Nancy Jo Kepple, PhD, Assistant Professor, University of Kansas, Lawrence, KS
Jennifer Price Wolf, PhD, Associate Research Scientist, Prevention Research Center, Oakland, CA
Bridget Freisthler, PhD, Professor, University of California, Los Angeles, Los Angeles, CA
Background and Purpose: Interactive Voice Response (IVR), a computer-automated telephone interviewing technique, is an innovative method that may address concerns of socially desirable responding and child abuse reporting when asking caregivers to self-report maltreatment behaviors. However, IVR poses a methodological challenge around high dropout rates. A variety of factors may contribute to this dropout, such as respondent dislike of the automated system, fatigue, or increased experience of cognitive burden with the absence of a live interviewer. Heavy drinkers and current drug users are more likely to have impaired cognitive functioning, which may result in a lower likelihood of survey completion due to a higher experience of cognitive burden when using IVR. Differential dropout by these groups could potentially underestimate child maltreatment rates. To determine if biased child maltreatment estimates arise from IVR data collection, this study assessed if respondents who are heavy drinkers or current drug users were less likely to complete the IVR portion of the survey, controlling for differences in key demographic and parenting behaviors.

Methods: We conducted a listed telephone survey of 3,023 caregivers residing in 50 mid-sized cities in California. Respondents, who were primary caregivers of at least one child younger than 13 years old, answered questions with a live interviewer before being transferred to an IVR section. Respondents who dropped out at any point after this transfer were identified as having incomplete surveys (N = 211, 7% of full sample). Key independent variables were obtained during the live interviewer portion of the survey and included demographic characteristics, non-reportable but high risk parenting behaviors, and caregiver alcohol and drug use. Multilevel logistic regressions were used to measure if these factors were significantly associated with survey completion.

Results: Ex-drinkers, light drinkers, moderate drinkers, and heavy drinkers of alcohol were more likely to complete the survey than lifetime abstainers of alcohol. Analyses indicated no statistically significant differences between respondents who completed the IVR section and did not complete the IVR section by drug use. Only one demographic variable significantly differed between groups: Respondents born in the United States were more likely to complete the survey than those born elsewhere. No parenting variables significantly differed by survey completion.

Conclusions and Implications: The use of IVR technology resulted in only minimal bias. Contrary to expectation, heavy drinkers were more likely to complete the survey compared to lifetime abstainers. This observed difference may result in a slight overestimation of maltreatment rates when using IVR technology. The findings suggest IVR could prove to be useful for future surveys that aim to estimate abusive and/or neglectful parenting behaviors given the limited bias observed for demographic characteristics and high risk caregiver traits.