Straightlining in a Survey Assessing Behavioral Health Treatment Services in Washington State

Primary author: Felix I. Rodriguez
Co-author(s): Rose Krebill-Prather; Kristen R. Petersen; Kent J. Miller

Primary college/unit: Edward R. Murrow College of Communication
Campus: Pullman

Straightlining, or non-differentiation in ratings across items, compromises the data quality of survey responses by introducing the possibility that satisficing or response bias has occurred. Recent studies on straightlining have examined the effects of demographic characteristics and mode of administration on this type of behavior. This study examines the extent of straightlining in survey responses of a sensitive population, using data from a statewide survey designed to evaluate publicly funded behavioral health services.

The Behavioral Health Enrollee Survey was administered in 2018 using a mixed-mode telephone/web design. Responses were collected from 2,135 randomly selected adults who received publicly funded behavioral health treatment services in Washington State from May through October of the preceding year.

First, the incidence of straightlining behavior is measured on four batteries of survey questions: quality of services, experience with services, perceived outcome of services, and feelings of being stigmatized. Then incidence of straightlining on each of these batteries is compared across groups by age, gender, minority status, behavioral health diagnosis, and mode of survey administration.

Preliminary results suggest straightlining occurs on each of the four batteries of questions. However, the extent to which straightlining is significantly related to other factors of interest varies from one battery to the next. Additional analyses examine more specifically what combinations of factors are related to straightlining. The results will shed light on whether these response patterns may be due to satisficing, or a more general lack of attention or care in survey responses for this population of behavioral health enrollees.