Tuesday, March 06, 2012

An introduction to the structure of biomarker equations

An Introduction to the Structure of Biomarker Equations
David G. Blanchflower
Dartmouth College, Stirling University, IZA,
and National Bureau of Economic Research, USA
blanchflower@dartmouth.edu
Nicholas A. Christakis
Harvard Medical School
christak@hcp.med.harvard.edu
Andrew J. Oswald
University of Warwick and IZA
andrew.oswald@warwick.ac.uk
December 2011


Abstract
Most economists are not familiar with so-called biomarker data. We attempt here to provide an introduction to such data and to describe the econometric structure of simple biomarker equations. We draw upon information on the heart rate, systolic and diastolic blood pressure, fibrinogen, and C-reactive protein levels of 100,000 adults. We show that it is extremely important to control for fruit and vegetable consumption (more so than is conventionally recognized in health economics). Once that is done, there are income gradients only in heart-rate and C-reactive protein equations, and those gradients are small. Education enters remarkably weakly in these biomarker equations.

2 comments:

Mark McGovern said...

Haven't read this yet but just looking at the abstract it seems strange to control for things that education or income is likely to affect (like diet) and then say that there's no social gradient.

Liam Delaney said...

fair point, but if the gradients could be explained by current consumption that would be a very strong result. think the authors are clear this is a preliminary pass through the data and an attempt to open a literature to economics.