Date created: | Last Updated:

Category: Communication

Description: Background: To be effective, interventions must target important determinants of behavior, so determinant selection is a crucial step in intervention development. Therefore, many determinant studies are conducted where the relative importance of theoretically derived behavioral determinants and their sub-determinants (e.g. beliefs) is mapped for different behaviors and populations. However, to arrive at actual determinant selection on the basis of the rich data collected in such studies is no straightforward affair. This requires simultaneous evaluation of both univariate distributions and associations for many different variables, where judgments should be based on confidence intervals instead of point estimates. We introduce a novel approach to address this challenge. Methods: This approach (CIBER) is based on visualization of confidence intervals for the means and correlation coefficients for all determinants simultaneously. CIBER is provided in open source R package 'userfriendlyscience', and created to be usable by researchers with no experience in R. Results: We illustrate CIBER using data on the determinants of using a high dose of 3,4-methylenedioxymethamphetamine (or ecstasy). Previous experience with using a high dose and expected euphoria (positively) and regret and worry (negatively) were attitude's strongest predictors. Univariate analyses showed strong risk perceptions (awareness of high doses being unhealthy) and the expectation that higher doses cause more hallucinations. Discussion: Determinant selection requires simultaneous comparison of many confidence intervals, but these analyses can yield prohibitively many data points. CIBER efficiently presents those data points, enabling their evaluation and ultimately, well-informed selection of the determinants to target with behavior change methods (or techniques, BCTs).