Predictions of magnitudes (costs, durations, environmental events) are often given as uncertainty intervals (ranges). When are such forecasts judged to be correct? We report results of four experiments showing that forecasted ranges of expected natural events (floods and volcanic eruptions) are perceived as accurate when an observed magnitude falls inside or at the boundary of the range, with little regard to its position relative to the “most likely” (central) estimate. All outcomes that fell inside a wide interval were perceived as equally well captured by the forecast, whereas identical outcomes falling outside a narrow range were deemed to be incorrectly predicted, in proportion to the magnitude of deviation. In these studies, ranges function as categories, with boundaries distinguishing between right or wrong predictions, even for outcome distributions that are acknowledged as continuous, and for boundaries that are arbitrarily defined (for instance, when the narrow prediction interval is defined as capturing 50 percent and the wide 90 percent of all potential outcomes). However, the boundary effect is affected by label. When the upper limit of a range is described as a value that “can” occur (Experiment 5), outcomes both below and beyond this value were regarded as consistent with the forecast.

The Lee–Carter model has become a benchmark in stochastic mortality modeling. However, its forecasting performance can be significantly improved upon by modern machine learning techniques. We propose a convolutional neural network (NN) architecture for mortality rate forecasting, empirically compare this model as well as other NN models to the Lee–Carter model and find that lower forecast errors are achievable for many countries in the Human Mortality Database. We provide details on the errors and forecasts of our model to make it more understandable and, thus, more trustworthy. As NN by default only yield point estimates, previous works applying them to mortality modeling have not investigated prediction uncertainty. We address this gap in the literature by implementing a bootstrapping-based technique and demonstrate that it yields highly reliable prediction intervals for our NN model.

There exists an ever-increasing number of systematic reviews, with or without meta-analysis, in the field of nutrition. Concomitant with this increase is the increased use of such to guide future research as well as both practice and policy-based decisions. Given this increased production and consumption, a need exists to educate both producers and consumers of systematic reviews, with or without meta-analysis, on how to conduct and evaluate high-quality reviews of this nature in nutrition. The purpose of this paper is to try and address this gap. In the present manuscript, the different types of systematic reviews, with or without meta-analyses, are described as well as the description of the major elements, including methodology and interpretation, with a focus on nutrition. It is hoped that this non-technical information will be helpful to producers, reviewers and consumers of systematic reviews, with or without meta-analysis, in the field of nutrition.

A major use of neuropsychological assessment is to measure changes in functioning over time; that is, to determine whether a difference in test performance indicates a real change in the individual or just chance variation. Using 7 illustrative test measures and retest data from 384 neurologically stable adults, this paper compares different methods of predicting retest scores, and of determining whether observed changes in performance are unusual. The methods include the Reliable Change Index, with and without correction for practice effect, and models based upon simple and multiple regression. For all test variables, the most powerful predictor of follow-up performance was initial performance. Adding demographic variables and overall neuropsychological competence at baseline significantly but slightly improved prediction of all follow-up scores. The simple Reliable Change Index without correction for practice performed least well, with high error rates and large prediction intervals (confidence intervals). Overall prediction accuracy was similar for the other three methods; however, different models produce large differences in predicted scores for some individuals, especially those with extremes of initial test performance, overall competency, or demographics. All 5 measures from the Halstead–Reitan Battery had residual (observed − predicted score) variability that increased with poorer initial performance. Two variables showed significant nonnormality in the distribution of residuals. For accurate prediction with smallest prediction–confidence intervals, we recommend multiple regression models with attention to differential variability and nonnormality of residuals. (JINS, 1999, 5, 357–369.)