Bar Chart with Errors
Annotation category:
Chapter 3
Neonatologists (specialists in treating newborn babies) were asked whether
or not they would recommend each of four treatments (IV feeding, intestinal
surgery, cardiac surgery, and dialysis) for an ill newborn whose survival
depends on getting that treatment. These questions were posed under five
different assumptions about the newborn in question: normal, apart from the
illness needing treatment; infant's mother infected with HIV; infant itself
infected with HIV; Down syndrome; and Trisomy-13 (a serious neurological defect
that always leads to the infant's death after a few days or weeks). The bars
show the proportion of respondents who would recommend the needed treatment;
this varies from nearly 100% for the simpler treatments applied to the
otherwise-normal infant, down to 1% who would recommend dialysis for a baby with
Trisomy-13. The graph also shows 95%-confidence intervals, based on the
binomial distribution (chapter 3), for each proportion. Because of the large
sample size, the asymmetry of these confidence intervals in scarcely
discernable in the error bars.