Samuel H. Preston

This book was certainly not what I expected. I expected a book detailing the plight faced by nineteenth-century families, stories about grieving mothers, about public health officials puzzling over how to solve the problem etc. The first chapter was a little like that. But the rest of the book was nothing but statistics. I warn you, reader: if you haven't taken a course in statistics you'll find this book very hard to understand. I certainly did.

There were sentences like: All three variables remain highly significant after accounting for degrees of freedom, but the multicategory size-of-place and region-of-residence variables are surpassed in F-ratios by more than parsimonious variables.

And: Weighting by number of children ever born also reduces the problem of heteroskedasticity, the tendency for larger variance to be associated with units containing further observations, a tendency that violates the assumptions of ordinary least squares regression.

Yeah. I have no idea what that means.

From what I could decipher, the authors puzzled over why the US had such a high child mortality rate when it was so prosperous. They analyzed things like whether the parents were immigrants and what county they came from, how long they breast-fed, whether or not the parents were literate, etc -- stuff I would never have considered. It was actually quite interesting, which is why I finished the book in spite of the fact that I could barely understand it. If you're willing to wade through all the scary numbers, this book is worth reading.… (plus d'informations)