Recrafting Youth Risk Assessment: Developing the Modified Positive Achievement Change Tool for Iowa

ABSTRACT

Risk assessments have become prevalent in the juvenile justice field. Many of these tools are adopted off-the-shelf and not adapted to fit the characteristics of a jurisdiction’s justice-involved youth. We examined the Positive Achievement Change Tool (PACT) in Iowa. Although used widely, the PACT is relatively unmodified. We updated the tool via item selection and weighting, gender-specific models and multiple outcomes, developing the Modified Positive Achievement Change Tool (M-PACT) for Iowa. We identified an average predictive accuracy increase of 7%. Evidence of reduced racial disparity was also observed, and research implications outline the need to customize assessments to improve predictive accuracy.     

Show me the meaning! The contextual relevance of images influences the recall and understanding of multimedia RSVP paragraphs

ABSTRACT

This study examined whether images facilitate contextual grouping as effectively as words. College students (N?=?102) read RSVP paragraphs with topics located at the beginning, middle, or end of the paragraph. Paragraph topic was presented as a topic-related image, topic-unrelated image, or written explicitly in text. Topic identification was equally accurate for topic-related images and topic sentences, and less accurate for unrelated images. Paragraphs from the text condition were recalled more accurately than the related and unrelated image conditions. Further, recognition of unrelated images was a positive predictor of the number of logical errors committed during recall. Therefore, related images did not improve memory and comprehension as much as text, and paragraphs with unrelated images negatively affected understanding of the meaning of the paragraph, but not recall. These findings have implications for multimedia information sources such as textbooks or articles that use images to clarify the meaning of the material.     

Relative possibilities of loops and redundant connections in neural nets

A model of the cortex (after Hebb) as a randomly connected set of neurons is examined. As traces made of larger numbers of cells are considered, the fraction of redundant connections likely to be contained in the trace increases. Within the trace, the fraction of connections likely to be reverberating loops decreases with larger networks.