Search

In the health science literature, a common approach of validating a regression equation is data-splitting, where a portion of the data fits the model (fitting sample) and the remainder (validation sample) estimates future performance. The R2 and SEE obtained by predicting the validation sample with the fitting sample equation is a proper estimate of future performance, tending to correct for the natural upward bias of the R2 and SEE obtained from fitting sample alone. Data-splitting has several disadvantages, however. These include: 1) difficulty, arbitrariness, and inconvenience of matching samples; 2) the need to report two sets of statistics to determine homogeneity; and 3) the lack of equation stability due to diluted sample size. The PRESS statistic and associated residuals do not require the data to be split, yield alternative unbiased estimates of R2 and SEE, and provide useful case diagnostics. This procedure is easy to use, is widely available in modern statistical packages, but is rarely utilized. The two methods are contrasted here using a simulation from original data for predicting body density from anthropometric measurements of a group of 117 women. The PRESS approach is particularly appropriate for smaller datasets; methods of reporting these statistics are recommended.

Although most studies have concentrated on fatal residential fire injuries, which are a leading cause of fatal injuries in the United States, few investigators have examined in detail nonfatal injuries as a consequence of residential fires. This population-based study used the Washington State Fire Incident Reporting System to assess the incidence and descriptive epidemiology of fatal and nonfatal burns or respiratory tract damage resulting from unintentional residential fires. For the 2-year period 1984-85 in King County, WA, the mortality rate due to injury in a residential fire was 0.7 per 100,000 per year, and the incidence of nonfatal injuries was 5.6 per 100,000 per year. Of 17 fatalities, 59 percent of the deaths occurred at the scene of the fire. Of 128 persons with nonfatal injuries, 19 percent were hospitalized; although the 55 percent seen as outpatients and the 26 percent treated by the fire department or paramedics at the fire scene usually had minor injuries, they would not have been captured if only traditional data sources had been employed. Those injured averaged 2.8 days of restricted activity, but the range was from less than 1 day to 1 year. Injuries were more common in the households with a low socioeconomic status and among nonwhites, especially American Indians. Variation in incidence by age, sex, and source of ignition for deaths and nonfatal injuries suggests appropriate targets for future fire injury prevention programs.