We are grateful to George Kellas and Paula Schwanenflugel for their comments on earlier drafts.

In 2014, following a Central Region pilot assessment, the National Weather Service implemented large-scale use of an experimental product of enhanced severe weather warnings known as impact-based warnings (IBWs). The overarching goal of these IBWs is to improve the threat warning process and motivate appropriate responses by using event tags and additional text that provides more specificity about the magnitude of the storm and its potential consequences. These IBWs are designed to be used by individuals in the field to make more effective decisions. Although qualitative research has shown overall satisfaction with IBWs (Harrison et al. 2014; Losego et al. 2013), little published experimental research has been conducted on these new enhanced warnings. The research reported here therefore empirically investigates the effectiveness of the new IBW experimental product. In three experiments, participants adopted the role of a plant manager and read both IBWs and non-IBWs. At three different decision points, participants made decisions about shutting down the plant and having employees shelter in place. The results of all three experiments show that the IBWs produced higher likelihoods of closing the plant and sheltering in place, but only after the additional IBW text (providing information about the hazard, source, and impact) was presented. Interestingly, participant background knowledge of tornadoes and severe weather had little impact on their shelter in place decisions. The results support the conclusion that the additional enhanced text in IBWs promotes a higher stated likelihood of sheltering in place, at least as measured by behavioral intentions.

This study examined inferential processing during reading. Adults and 3rd, 5th, and 8th graders read stories implying a consequence and answered questions. Some of the inferences were more necessary than others for comprehension according to T. Trabasso and P. van den Broek's (1985) causal criteria, and the readers engaged in either superficial or integrative reading. Results showed that elaborative inferences were not as likely to be generated as those more necessary for comprehension. The necessity manipulation also produced a similar pattern of responding among the 4 groups, suggesting that even the young children were sensitive to the causal criteria. In addition, the integrative reading condition prompted a general slowdown for the 2 younger groups, whereas the 2 older groups were not as affected by reading condition. This pattern was interpreted in terms of differences in attentional demands and working memory capacity.

This study addressed age-related changes in the ability to draw inferences, and examined the generation process of the inferences themselves. Second-, fifth-, and eighth-grade children, and college undergraduates read eight stories from which two types of inferences could be drawn: those that are critical to the comprehension of a story (backward inferences), and those that are not (forward inferences). Subjects then answered questions testing whether the appropriate inferences had been drawn. Backward inferences were drawn more reliably and more rapidly than were forward inferences, at all grade levels. The results suggest that children as young as second grade are able to draw inferences from text, particularly when the coherence of the text requires such inferences, but that the ability to draw both kinds of inferences still increases with age. It is also suggested that, for all ages, backward inferences are most likely to be drawn at the time the text is encoded, while forward inferences are most likely to be drawn only when needed, at retrieval. RESUME Cohe'rence textuelle et de'veloppement des capacitb de production d'in fkrencesBeaucoup soutiennent que la capacite a faire des inferences est un important dkterminant de la comprkhension en lecture. Quoiqu'il y ait abondance de recherches montrant que l'enfant et l'adulte sont des producteurs d'infkrence prolixes, quand on s'intkresse au lieu de production des infkrences, l'image est beaucoup moins Claire. La question n'est plus de savoir si les enfants et les adultes peuvent faire des infkrences, mais de savoir quand on fait ces inferences. Les inferences sont-elles encodees lors de la comprkhension et, par conskquent, sont-elles stockkes en tant que

No abstract