How effective are warnings? A meta-analysis

We present the results of a meta-analysis on warnings which quantify their effects on users' behavioral compliance to, their recall of, and their attitudes toward, such warnings. We develop this comprehensive effort founded upon patterns observed in prior meta-analyses, but which are now updated and expanded in light of a further two additional decades of reported experimental work. We identified a total of thirty studies which qualified under the presently imposed inclusion criteria. This process resulted in the identification of a total of two-hundred and seventy-two effect sizes for analysis. Results from these analyses revealed that behavioral compliance to warnings is facilitated by the integration of such warnings into the work task, especially if accompanied by clear language explanations and placement in closest proximity to the person at risk. The noticeability of a warning proved to be mediated by its size, its immediate placement and, as with behavioral compliance, its integration into task processes. Recall of a warning's content was significantly facilitated by its level of sensory conspicuity, the presence of descriptive words and pictures, as well as the perceived seriousness of the communicated threat. Finally, people's attitudes to warnings were contingent upon its perceived seriousness, and the sematic terms used specifically to communicate that threat. These summated results can provide enhanced design guidelines to help toward the optimization of future warnings. Nevertheless, the absolute power of warnings' behavioral effects remains low and this last line of defense in the hierarchy of physical controls persists as a relatively weak form of protection from harm. However, warnings prove to be the first line of defense against cognitive damage and must therefore be accorded a greater prominence in a world where the character of threats and nature of work itself is changing rapidly.