Using Confidence Interval-Based Estimation of Relevance to Select Social-Cognitive Determinants for Behavior Change Interventions.

Rik Crutzen,Gjalt-Jorn Ygram Peters,Judith Noijen

doi:10.3389/fpubh.2017.00165

Rik Crutzen, Gjalt-Jorn Ygram Peters + Show 1 more

Open Access

https://doi.org/10.3389/fpubh.2017.00165

Copy DOI

Abstract

When developing an intervention aimed at behavior change, one of the crucial steps in the development process is to select the most relevant social-cognitive determinants. These determinants can be seen as the buttons one needs to push to establish behavior change. Insight into these determinants is needed to select behavior change methods (i.e., general behavior change techniques that are applied in an intervention) in the development process. Therefore, a study on determinants is often conducted as formative research in the intervention development process. Ideally, all relevant determinants identified in such a study are addressed by an intervention. However, when developing a behavior change intervention, there are limits in terms of, for example, resources available for intervention development and the amount of content that participants of an intervention can be exposed to. Hence, it is important to select those determinants that are most relevant to the target behavior as these determinants should be addressed in an intervention. The aim of the current paper is to introduce a novel approach to select the most relevant social-cognitive determinants and use them in intervention development. This approach is based on visualization of confidence intervals for the means and correlation coefficients for all determinants simultaneously. This visualization facilitates comparison, which is necessary when making selections. By means of a case study on the determinants of using a high dose of 3,4-methylenedioxymethamphetamine (commonly known as ecstasy), we illustrate this approach. We provide a freely available tool to facilitate the analyses needed in this approach.

Highlights

When developing an intervention aimed at behavior change, one of the crucial steps in the development process is to select the most relevant determinants [1]
The current paper demonstrates how to select the most relevant sub-determinants and how this can have an impact on choices made during intervention development
We have described an analytical approach, denoted as Confidence Interval-Based Estimation of Relevance (CIBER), to look at associations between sub-determinants and outcomes

Summary

INTRODUCTION

When developing an intervention aimed at behavior change, one of the crucial steps in the development process is to select the most relevant determinants [1]. If a sub-determinant is positively associated with behavior but left-skewed, most population members already have the desired value, so an intervention developer will want to reinforce it. The diamonds on the right hand panel show the association strengths (i.e., correlation coefficients with 95% confidence intervals) between individual items and determinants at different levels of psychological aggregation (attitude and intention in this example). Correlation coefficients, means, and confidence intervals of both need to be combined to select behavioral beliefs to be targeted in an intervention. The relevance is relatively low, because the scores on the middle panel indicate that participants are already convinced that using a high dose of ecstasy is much worse for their health If this belief is targeted in an intervention, this would mean that the belief needs to be confirmed, unless it is possible to tailor the intervention message to only target the small subgroup of participants who are not convinced of the dose/risk relationship.

A Practical Guide to Obtain Visualizations

DISCUSSION

ETHICS STATEMENT