Overcoming defensive responding in the estimation of prevalence of non-medical use of prescription stimulants.

Prevalence of Non-Medical Use of Prescription Stimulants (NMUPS) is estimated to be high among young adults enrolled in college. However, precise estimation of the prevalence of NMUPS is challenging owing to biases affecting self-report of sensitive and potentially illegal behaviors. This study aimed to estimate the prevalence and risk factors of NMUPS using the crosswise randomized response technique (CRRT) and compare findings to the traditionally-used direct self-report (DSR) method. This study utilized a cross-sectional, randomized experimental design to survey adult undergraduate students at a major southeastern university in the United States. Eligible respondents were randomly assigned to a DSR group or a CRRT group. Those in the DSR group were presented a direct question about NMUPS, but those in the CRRT group were asked to indicate whether their response to the NMUPS question was the 'same' or 'different' compared to a random non-sensitive question. Prevalence of NMUPS was found to be 18.6% (95% CI:18.5%–18.7%) in the DSR group and 32.5% (95% CI:32.1%–32.9%; p = 0.003) in the CRRT group. Logistic regression analysis predicting NMUPS in the DSR group showed that it was significantly associated with positive expectancies (OR:3.50; 95% CI:2.44–5.02), negative expectancies (OR:0.49; 95% CI:0.35–0.68), perceived norms (OR:1.71; 95% CI:1.27–2.29), and religious beliefs (OR:0.69; 95% CI:0.52–0.92). The setting and mechanism of the survey is likely closely related to the validity of prevalence estimation of sensitive behaviors. This study found that prevalence of sensitive behaviors such as NMUPS is significantly higher when respondents are provided increased anonymity. • Prevalence of sensitive behaviors is underestimated due to evasive survey responses. • Prevalence of NMUPS was found to be 18.6% using direct self-report in online survey. • Prevalence increased to 32.5% using the crosswise randomized response technique. • Methods of survey administration play a significant role in validity of results. [ABSTRACT FROM AUTHOR]