54 - An Experimental Analysis of Research Assent in Children

Background:  Guidelines define assent as an affirmative response by which an individual who cannot provide legal consent may agree to participate in a research activity. However, little is known about potential independent variables involved in the assenting response. This study evaluated factors influencing assent using a dual choice- and engagement-based model.

Methods:  A potential approach to assessing assent validity is correlating an initial assent response with subsequent assent probes while the activity assented to and any alternative activity remain available. This arrangement allows for the assessment of assent validity metrics, including assent-choice correspondence (i.e., extent to which an initial assent response predicts subsequent assent responses) and assent-engagement correspondence (i.e., extent to which assent predicts engagement with the assented activity). This experimental model of assent was evaluated using a single-subject research design, specifically a reversal design. This design, commonly used in behavioral research, was selected to understand the impact of independent variables on a within-subject basis. Nineteen children aged 4 to 8 participated and experienced one of four types of assent-seeking protocols. The dependent variables included initial assent responses, subsequent assent responses, and activity engagement. The dependent variables were measured across conditions varying in activity availability and relative preference. Typical single-subject designs involve no more than five participants. As this study was evaluating an experimental model, more participants were included so aggregated analyses could be conducted.

Results:  Across participants, children showed high levels of assent for moderately preferred research-like activities when no alternatives were available (as might be expected in a typical research context with children). When leisure activities were offered concurrently, levels of assent to the research-like activity declined. Even low-preference leisure activities competed effectively with higher-preference research-like activities when offered simultaneously. In other words, most children assented to participating in the research-like activity only when no other alternative activities were available. Across all conditions, when assent occurred, it typically was followed by engagement in the research-like activity, but this correspondence declined over time. Aggregated analyses indicated no apparent impact of assent-seeking protocol (concurrent-chains. pictorial, vocal, written). Similarly, even though participants showed variability in their performance of putative assent prerequisite skills, their performance of these skills did not have a significant correlation with assent probability or assent-engagement correspondence.

Limitations: As a laboratory analysis of assenting behavior, our definition of assent in terms of selecting a research-like activity in the context of a multiple-choice scenario is by no means a standard in the literature. Also, our choice of experimental design balanced economy and within-subject replication. Only one of the three test conditions was reversed for each participant. Ideally, we would have been able to expose each participant to each of the test conditions.

Discussion:  Our study suggests that a dual choice- and engagement-based experimental model to study assent can be utilized successfully with children. The results showed that a single-activity presentation format produced the highest and most consistent level of assent. Concurrent-activity presentations produced relatively low levels of assent. The degree of research activity displacement was a function of the relative preference of the research and leisure activities. Future research should consider the nature of the alternatives to assent necessary for true choice. An impoverished alternative to assent albeit effective in securing assent, may not be sufficient as a meaningful assent-seeking procedure.