Increasing response rates to an online behavioral survey offered to students attending a large minority-serving university: The Student Behavioral Health Survey (SBHS-Web)

Background:  Electronic surveys have numerous advantages over traditional modes of data collection, but response rates are notoriously low.  For example, the National College Health Assessment II (NCHA-II) offered to students attending our university in fall 2012 produced 5% response over a 39-day period.  We designed an experiment in effort to increase the response rate to a similarly conducted student behavioral health survey (SBHS-Web).

Methods:  The 40-item SBHS-Web aimed to measure sensitive behaviors among students attending a large, minority-serving university.  A two-factor experimental design measured the effect of letter/questionnaire design (increased personalization, reduced length) and pre-notification on responses.  Email addresses of 8,000 students enrolled in spring 2013 were randomly selected and allocated into four conditions of equal size:  standard (S), standard pre-notification (SP), innovative (I), and innovative pre-notification (IP).  Differences in response were tested using ANOVA procedures.

Results:  The median age of 632 respondents (7.9%) was 23 years (range 18-60).  Most were undergraduate students (77%), unmarried (77%), female (60%), Hispanic (59%), employed (59%), and living off campus with family (54%).  All conditions were comparable on sociodemographic factors.  The response rate in condition IP (14.0%) was significantly greater than in conditions S (6.5%*), SP (6.3%*), and I (4.9%*)(*p<0.001).  Response rates in conditions S, SP, and I were no different than in NCHA-II.

Conclusion:  The combined effects of personalization, reduced length, pre-notification, and consent resulted in a significantly higher response rate among students enrolled at our university.  Additional research is needed to further identify factors that will increase response rates and contribute to survey validity.