Saturday, January 06, 2007
Drawing lines through data, predetermined patterns, and good imaginations
One of the topics my PhD advisor was the most concerned about was user bias in presenting data. I spent a lot of time learning how to propagate errors through calculations, and how to decide what kind of curves are appropriate to fit to various 2D data sets. I came across an article in a recent publication (pub and author name withheld for now) that got me thinking. Above I have posted a modified figure from this publication, the x-axis is time, the y-axis is a value measured in sediment cores. Take a few minutes and decide how you would draw a line through this data if you needed to, I will come back with their version in a few days. I have erased everything but the data from this figure.
Correlating data with lines or curves can of course be informed by expected patterns, where one value has been shown to vary regularly with the second. I am not an expert in the branch of earth science discussed in this article, so perhaps I am missing something. That being said, enjoy.
One more note, my first beef is the lack of uncertainties associated with the data points. They should either be on the figure or there should have been a disclaimer in the figure caption mentioning that the uncertainties are smaller than the size of the symbol. You will have to take my word that it doesn't.