UC Santa Barbara

Uncertainty is native to science and thus also to wholly accurate science communication, However, uncertainties that are inevitable in individual findings of science and in larger processes of science are often not clearly communicated to the public. Instead, many public-facing science communicators purposefully avoid discussing the uncertainties that are attached to the science they communicate – often out of fear of adverse effects of those uncertainty frames. To date, it has remained unclear whether these fears are well-founded because much of the extant literature has produced competing theory and findings.

Much of the uncertainty about uncertainty in the extant literature has been caused by inconsistent and uncoordinated conceptual and operational definitions of uncertainty and uncertainty frames. Therefore, this dissertation develops a new and clarifying conceptual explication of distinct uncertainty types, which enables a more nuanced interpretation of the extant empirical literature. Specifically, that frames of consensus uncertainty have been associated with none of the reported findings of positive effects of uncertainty frames – only negative effects and nonsignificant effects. Conversely, frames of technical uncertainty and scientific uncertainty frames have been associated with positive and nonsignificant effects. Frames of deficient uncertainty have not been a focus of the extant empirical literature.

However, the summary observations from this informal meta-analytic review are the product of disparate methods, issue contexts, concepts, and measures – all of which are confounding factors that render confident meta-analytic conclusions impossible. The literature lacked a controlled experiment that compared the effects of each uncertainty frame type within one consistent methodology.

All of these factors together have created a situation where progress toward the answers to a question with universal importance and tangible applications has been obscured despite many uncoordinated efforts occurring within disciplinary silos. Therefore, this dissertation was an effort to move this field of research forward – providing a rigorous and robust set of findings that inform the relative effects of different types of uncertainty frames in science communication, and the role of motivated reasoning in responses to portrayals of uncertainty in science – all tested in one, large, controlled experimental design.

This dissertation employs an online survey experiment in a between-subjects five (frame type) by three (issue context) factorial design to specify and compare the effects of four distinct uncertainty frame types (consensus, deficient, scientific, and technical uncertainties, respectively) and one control condition on attitudinal outcome variables of claim belief, credibility perceptions, and behavioral intentions. These tests are replicated in each of three distinct issue contexts: the effects of climate change, the effects of GMO labeling laws, and the occupational hazards posed by vibrating machinery.

Using an opt-in online sample and quotas that approximated U.S. census levels of education, age, and gender, this dissertation asked participants to read a simulated news article that contained a report of new scientific evidence. These claims of new science findings were the experimental manipulations, and as such used a frame of one (or none) of the uncertainty types. After reading their assigned news article, participants then indicated their belief in the claim, credibility perceptions of the scientists, and intentions to engage in relevant behaviors that indicated support for the claim.

By applying a rigorous sequence of explication and validation in the development of the measures and the measurement model, this dissertation demonstrated clear evidence that the measures have strong convergent and discriminant validity. By first establishing the manipulation check with rigorous testing and revision, and the basic structure of the conceptual model with SEM, this dissertation builds a foundation of confidence in conceptual, theoretical, and methodological validity upon which to base interpretations of the later tests of interaction effects.

These tests of conditional effects found that – in the issue contexts of climate change and vibrating machinery – frames of consensus uncertainty are associated with significantly lower belief in the claim and perceptions of credibility of the scientists, compared to other types of uncertainty frames. Importantly, and interestingly, the other three types of uncertainty frames did not significantly differ from each other in their associations with levels of the attitudinal outcome variables. Also, while the tests of the conceptual model demonstrated strong motivated reasoning effects, these effects do not differ across uncertainty frame types. These findings have important implications for theory, research, and practice, and multiple interpretations of them are discussed at length in the latter portions of this dissertation.