A SAS Macro for Jackknifing the Results of Discriminant Analyses.

In discriminant analysis, the correlations between the variables and the discriminant functions, structure coefficients, are used in interpretation. Rules of thumb are used to set a cut-off value. It would be useful for structure coefficients to be at least statistically significantly different from zero before being used in an interpretation. Unfortunately, there are no formulas for the standard error of structure coefficients. However, the jackknife procedure can be applied to produce an estimate of the standard error. A SAS macro procedure for performing a jackknife analysis on structure coefficients is described together with issues and caveats on its use in multivariate methods.

What value is given to quality of life assessment by health professionals considering response to palliative chemotherapy for advanced cancer?

A study was designed to obtain information on the importance of quality of life assessment (QL) during palliative chemotherapy. A questionnaire was answered by 542 health professionals (392 general practitioners, 20 specialist oncologists, and 130 oncology nurses). In both simulated patient situations and multiple-choice questions, all groups rated QL higher than other standard methods of assessment. General practitioners and oncologists appeared to weight the assessment criteria more equally than nurses who gave strong emphasis to QL. In the simulated patient situation, there was a small degree of interaction between QL and other assessment criteria. However, the analysis showed that QL was regarded as an independent variable and was considered to be the most important objective of palliative chemotherapy for advanced cancer.

The relationship of child abuse indicators to the assessment of perceived risk and to the court's decision to separate.

The most serious decision taken in child abuse cases is to formally separate the child and the family. There has been concern that the criteria for making these decisions should be made explicit. The consequences of the separation for the child, family, and society is one set of reasons for this concern. Another is that with the increasing number of cases being reported, there is a need for reliable, consistent, and efficient decisions across cases. This paper operationally defines a set of indicators for the separation decision. One hundred fifty-two cases of suspected child abuse from the Royal Children's Hospital were coded in terms of the indicators and the outcome of the case. The outcome was the court's decision to separate the child. The study investigated whether and in what way these indicators were related to the separation outcome. In addition the perceived risk to the child was operationally defined, and the relationship of the indicators and the court's decision to perceived risk was studied. Statistical analysis of the data showed that the indicators most associated with the separation outcome were severity of abuse, parenting, and the family social system. Two important implications for practice are that information on explicit indicators should be recorded for cases, and that they can provide a base for training of new child protection workers.